Feasibility Assessment

Appendix C.

Feasibility Assessment for Epidemiological
Studies at Pease International Tradeport,
Portsmouth, New Hampshire
November 2017

0

Contents

Summary ..................................................................................................................................................... 2
Introduction ................................................................................................................................................. 8
Site history .................................................................................................................................................. 9
Community concerns ................................................................................................................................ 11
Exposure assessment ................................................................................................................................. 12
Summary of literature review ................................................................................................................... 14
Adult cancers and other adult diseases ............................................................................................. 14
Health effects in children.................................................................................................................... 14
Sources of adverse outcome data for the Pease population ...................................................................... 15
Sources of exposure data .......................................................................................................................... 17
Feasibility of an epidemiological study of children at the Pease Tradeport ............................................. 18
Feasibility of an epidemiological study of adults at the Pease Tradeport................................................. 32
Feasibility of an epidemiological study of former military service and civilian workers at the former
Pease Air Force base ................................................................................................................................. 41
Other study designs and health-related endpoints .................................................................................... 42
References ................................................................................................................................................. 45
Tables ........................................................................................................................................................ 59
Appendix ................................................................................................................................................... 76
Literature review ................................................................................................................................ 77
Description of sample size calculations ............................................................................................. 93
Appendix tables ................................................................................................................................. 115
Comments from the Pease CAP and ATSDR responses……………………………………………….164

1

Summary
This report describes the activities and the conclusions of ATSDR’s feasibility assessment of possible
future drinking water epidemiological studies at the Pease International Tradeport, Portsmouth, New
Hampshire (“Pease”). The drinking water at Pease was contaminated with perfluoroalkyl substances
(PFAS), in particular perfluorooctane sulfonate (PFOS) and perfluorohexane sulfonate (PFHxS), from
the use of aqueous film-forming foam (AFFF) at the former Pease Air Force Base. The base used AFFF
for firefighting training and to extinguish flammable liquid fires. In 2015, the New Hampshire
Department of Health and Human Services (NH DHHS) established a PFAS blood testing program at
Pease. A total of 1,578 persons submitted a blood sample for analysis. The results from the blood testing
program indicated that the exposed population had higher serum levels of PFOS and PFHxS than did the
U.S. population.
In March 2016, ATSDR established a community assistance panel (CAP) as a mechanism for the
community to voice its concerns and provide input on decisions concerning potential health activities at
Pease. A key concern expressed by the community was the lack of information on the possible shortterm and long-term health effects to children and adults exposed to the PFAS contaminants in the
drinking water at Pease. Specifically, the community was concerned about cancers, elevated lipids,
effects on thyroid and immune function, and developmental delays in children.
ATSDR then assessed whether epidemiological studies focusing on populations at Pease were feasible
and whether such studies could answer the concerns of the community. When evaluating whether an
epidemiological study would be scientifically feasible, ATSDR used three main criteria:
1. Meaningful and credible results — a study should have sufficient validity and precision, be
capable of detecting health-related effects, and be as responsive as possible to the community’s
questions and concerns. Ideally, a study should also be capable of detecting health-related
effects, for example a 20% to 100% increase in risk with sufficient statistical power (i.e.,
statistical power ≥80%).
2. Scientific importance — a study should evaluate biologically plausible diseases and other healthrelated endpoints (also called “effect biomarkers”) and improve our understanding of possible
health effects of PFAS exposures.
3. Public health significance — a study should provide a basis for determining if PFAS exposures
increase the risks for specific adverse health effects, and if so, what public health actions are
necessary to reduce the risks. The study should also be relevant to other populations with similar
exposures.
The feasibility assessment is guided by these three criteria and does not address considerations of
financial or operational feasibility. Feasibility was also assessed in terms of whether sufficient
participation (sample size) could be obtained from within the Pease community to achieve sufficient
statistical power for the health-related endpoints being considered, or whether the study would need to
be expanded to other communities beyond the Pease population.
ATSDR reviewed the epidemiological literature on PFAS exposures to identify the health-related
endpoints that have been studied and current data gaps, in particular, for the effects of PFHxS. The
2

literature review also was used to identify adverse effect sizes observed in the PFAS studies for PFAS
serum levels similar to those found in the Pease population.
The literature review found that most information on potential health effects concerned exposures to
perfluorooctanoic acid (PFOA). In particular, numerous studies have been conducted of West Virginia
and Ohio residents and workers exposed to PFOA from a chemical plant (the “C8” studies) [Frisbee
2009]. Studies of other workforces also were primarily focused on PFOA exposures. The literature
review found that less information was available about the potential health effects of PFOS exposures,
and very little information was available on the potential health effects of exposures to PFHxS. Because
the primary contaminants in the drinking water at the Pease Tradeport were PFOS and PFHxS,
epidemiological studies of the Pease populations have the potential to fill key knowledge gaps and
address the community’s concerns.
The literature review identified many health-related endpoints evaluated in previous epidemiological
studies of PFAS exposures. These included cancers, lipids, effects on thyroid and immune function, and
developmental delays. They also included effects on kidney and liver function and sex hormones, and
diseases such as endometriosis, ulcerative colitis and osteoporosis. Many of these health-related
endpoints were also previously raised by the community and the Pease CAP.
In considering possible study designs, ATSDR focused on the methods used in previous epidemiological
research of PFAS exposures. Adopting study design methods consistent with previous research would
facilitate the interpretation and synthesis of findings across studies. The literature review found that
most of the epidemiological studies of PFAS exposures were cross-sectional and evaluated serum PFAS
measurements. Some studies also evaluated cumulative PFAS serum levels that were estimated from
modeling methods. ATSDR concluded that any study of populations exposed to the PFAS-contaminated
drinking water at the Pease Tradeport should be cross-sectional and evaluate measured serum PFAS
measurements as well as estimated cumulative PFAS serum levels. ATSDR also concluded that methods
used to evaluate health-related endpoints in the Pease Tradeport populations should be consistent with
methods used in previous epidemiological research of PFAS exposures.

Potential Study Designs
A. Cross-sectional study of children
The first design is a cross-sectional study of children who were exposed to the PFAS-contaminated
drinking water while attending the two day-care centers at Pease. Inclusion would be limited to children
who attended the day-care centers any time before June 2014, and who would be in the age range of 4–
17 years at the time the study begins. During the 2015 blood testing program at Pease, 379 children aged
1–14 years contributed blood samples. If a study were to begin in 2018, these children would be ages 4–
17 years. The study would involve re-contacting these participants and obtaining new blood samples. To
increase the sample size, the study would also recruit and obtain blood samples from children who
attended the day-care centers at Pease, but who did not participate in the New Hampshire blood testing
program. Because PFAS-contaminated drinking water exposures could occur to children in utero and
during breastfeeding if the mother worked at the Pease Tradeport, the study would include these
additional children if the exposures began prior to June 2014 and their ages are 4 – 17 years at the time
the study begins.
3

A comparison group of children, who did not attend day care at the Pease Tradeport and whose parents
did not work at the Pease Tradeport or have occupational exposures to PFAS, would be recruited and
blood samples collected. The comparison group would be sampled from the Portsmouth public schools
and selected to have similar demographics as the Pease children.
Based on the health-related endpoints included in the final study, blood samples could be used to
evaluate PFAS serum levels and several biomarkers of effect, including lipids, thyroid function, kidney
function, immune function, and sex hormones. The children could also be assessed for neurological
endpoints such as intelligence quotient (IQ), learning problems, and attention-deficit/hyperactivity
disorder (ADHD) behaviors.
Calculations were conducted assuming a sample size of 350 exposed children who attended day care at
the Pease Tradeport and 175 unexposed children from the Portsmouth area who did not attend day care
at the Pease Tradeport. Additional sample size calculations assumed a sample size of 500 exposed
children and 250 unexposed children. The sample size calculations also assumed a simple comparison of
exposed versus unexposed children. A second approach was to determine the sample sizes needed to
detect effects found in other PFAS studies of children with serum PFAS levels similar to those observed
in the Pease children population. For some health-related endpoints, there was insufficient information
to conduct any sample size calculations.
Based on sample size considerations, health-related endpoints were grouped into three categories: 1)
feasible to study, 2) possible to study (but would require a larger sample size than 350 exposed children
and 175 unexposed children), and 3) not feasible to study using the Pease children population unless
additional populations exposed to PFAS-contaminated drinking water from other affected communities
are included in the study.
Health-related endpoints feasible to study in children at Pease
•
•
•
•

Mean difference in lipids (total cholesterol, LDL, HDL, triglycerides)
Mean difference in estimated glomerular filtration rate (eGFR), a measure of kidney function
Insulin-like growth factor – 1 (a measure of growth hormone deficiency)
Overweight/Obesity

Health-related endpoints that may be possible to study in children at Pease (although a larger
sample size from the Pease community will likely be needed)
•
•
•
•
•
•
•
•
•

Mean difference in uric acid, a measure of kidney function
Elevated total cholesterol (hypercholesterolemia)
Elevated uric acid (hyperuricemia)
IQ/neurobehavioral
Thyroid function
Sex hormones
Asthma and atopic dermatitis (immune function)
Rhinitis (stuffy, runny nose)
Antibody responses to rubella, mumps and diphtheria vaccines
4

Health-related endpoints not feasible to study using the Pease children population (in order to
address these health endpoints, populations from other sites beyond the Pease community with PFAScontaminated drinking water would need to be included along with the Pease children population)
•
•
•
•
•

Attention deficit/hyperactivity disorder (ADHD)
Autism spectrum disorder
Delayed puberty
Thyroid disease
Childhood cancers

To evaluate exposure-response trends, the study participants would need to be split into tertiles or
quartiles based on their serum PFAS levels. This might require a larger sample size for some of the
health-related endpoints listed as feasible to study.
B. Cross-sectional study of adults
The second cross-sectional study design would involve obtaining blood samples from adults aged ≥18
years who worked anytime at the Pease Tradeport during January 2008–May 2014. This study would
evaluate PFAS serum levels, lipids, thyroid function, liver function, kidney function, and immune
function. The study would also evaluate diseases such as kidney disease, liver disease, cardiovascular
disease, thyroid disease, ulcerative colitis, rheumatoid arthritis, osteoporosis, osteoarthritis, and
endometriosis. In the 2015 blood testing program at Pease, 1,182 adults aged ≥18 years participated, and
1,083 (91.6%) adults reported that they last worked at Pease during 2008–2014.
Calculations were conducted assuming a sample size of 1,500 adults exposed while employed at the
Pease Tradeport and 1,500 unexposed adults from the Portsmouth area who never worked at the Pease
Tradeport. The sample size calculations also assumed a simple comparison of exposed versus unexposed
adults. A second approach was to determine the sample sizes needed to detect effects found in other
PFAS studies of adults with serum PFAS levels similar to those observed in the Pease adult population.
Based on sample size considerations, health-related endpoints were grouped into three categories: 1)
feasible to study, 2) possible to study (but would require a larger sample size than 1,500 exposed and
1,500 unexposed adults), and 3) not feasible to study using the Pease adult population unless additional
populations exposed to PFAS-contaminated drinking water are included in the study.
Health-related endpoints feasible to study in adults at Pease
•
•
•
•
•
•
•
•

Mean difference in lipids (total cholesterol, LDL, HDL, triglycerides)
Elevated total cholesterol (hypercholesterolemia)
Mean difference in uric acid, a measure of kidney function
Elevated uric acid (hyperuricemia)
Thyroid disease (unconfirmed)
Cardiovascular disease
Hypertension
Osteoarthritis and osteoporosis
5

•

Mean differences in serum immunoglobin (IgA, IgE, IgG, IgM), and C-reactive protein (an indicator
of inflammation); increase in antinuclear antibodies (an indicator of autoimmune reaction);
alterations in specific cytokines

Health-related endpoints that may be possible to study in adults at Pease (although a larger sample
size from the Pease community may be needed)
•
•
•
•
•

Liver function
Thyroid disease (confirmed)
Thyroid function
Endometriosis
Pregnancy-induced hypertension

Health endpoints not feasible to study using the Pease adult population (i.e., populations from other
sites beyond the Pease community with PFAS-contaminated drinking water would need to be included
to evaluate these health-related endpoints)
•
•
•
•
•
•
•

Liver disease
Kidney disease
Ulcerative colitis
Rheumatoid arthritis
Lupus
Multiple sclerosis
Kidney cancer (and other adult cancers)

To evaluate exposure-response trends, the study participants would need to be split into tertiles or
quartiles based on their serum PFAS levels. This might require a larger sample size for some of the
health endpoints listed as feasible to study.

C. Mortality study of former military service and civilian worker personnel
A third study design that was considered would evaluate mortality and cancer incidence among former
military service and civilian worker personnel at the former Pease Air Force Base and other military
bases where drinking water was contaminated with PFOS and PFHxS from the use of AFFF.
Comparison military bases would also need to be identified that had no PFAS-contaminated drinking
water or drinking water contamination from other chemicals above the U.S. Environmental Protection
Agency’s maximum contaminant levels (MCLs). Personal identifier information (e.g., Social Security
number, name, date of birth, sex) necessary for data linkage with the national death index and state and
federal cancer registries could be obtained from the Defense Manpower Data Center.
However, based on sample size considerations, ATSDR concluded that it is not feasible to conduct a
mortality or cancer incidence study that is limited to the military service and civilian workers who were
stationed or worked at the Pease Air Force Base. Such a study would require, in addition to the Pease
6

Air Force Base populations, several thousands of exposed populations from military bases where PFAScontaminated drinking water occurred, as well as several thousands of comparison populations from
military bases that did not have drinking water contamination.

Conclusions
The feasibility assessment concluded that it is possible to evaluate some health-related endpoints if a
sufficient number of children and adults from the Pease population participate. Other health-related
endpoints would require larger numbers of exposed individuals and would require the inclusion of
populations from other sites who were exposed to PFAS-contaminated drinking water. The feasibility
assessment concluded that a third study design, a mortality and cancer incidence study of former
military service and civilian worker personnel, would not be feasible solely with the population at Pease.
No single study of the Pease population will provide definitive answers to the community about whether
their exposures to the PFAS-contaminated drinking water caused their health problems. All
epidemiological studies of environmental exposures and health outcomes have limitations and
uncertainties. Whether a study will find an association between an environmental exposure and health
effects cannot be known prior to conducting the study. The ability of a study of the Pease population to
provide useful information will depend to a great extent on the success of recruiting sufficient number of
study participants.
The feasibility of successfully evaluating particular health-related endpoints (or effect biomarkers) could
change depending on final study design and goals.

7

Introduction
This report describes the approach and the conclusions of the Agency for Toxic Substance and Disease
Registry’s (ATSDR’s) feasibility assessment of possible drinking water epidemiological studies at the
Pease International Tradeport (“Pease”), Portsmouth, New Hampshire. The purpose of the feasibility
assessment was to determine whether epidemiological studies are reasonable to conduct at Pease and
whether data exist to conduct scientifically credible epidemiological studies. This feasibility assessment
report for possible future studies at Pease International Tradeport was distributed to the Pease
Community Assistance Panel (CAP) for members’ review and input. Input from the CAP was intended
to help ATSDR ensure the proposed research is relevant to community concerns. The report is not
intended to be a protocol or systematic literature review. The final study design, including sample size,
the health endpoints that can be considered and the development of the study protocol itself, including
the statistical analysis approach have yet to be determined. The Pease CAP will have an opportunity to
review and provide input on a draft of the study design before it is finalized. The feasibility assessment
does not represent a commitment by ATSDR to conduct research at Pease International Tradeport, given
that funding and staffing to conduct the described research are not available at this time.
Three criteria were used to determine whether epidemiological studies are warranted at Pease:
1. Meaningful and credible results —a study should have sufficient validity and precision, be
capable of detecting health-related effects, and be as responsive as possible to the community’s
questions and concerns. Ideally, a study should also be capable of detecting health-related
effects, for example a 20% to 100% increase in risk with sufficient statistical power (i.e.,
statistical power ≥80%). To achieve sufficient validity, a study should minimize biases such as
selection bias and confounding bias. Sufficient precision can be achieved by a sample size that
has at least 80% statistical power to detect health-related effect sizes observed in other studies
for PFAS serum levels similar to those in the Pease population.
2. Scientific importance — a study should evaluate biologically plausible diseases and other
health-related endpoints (also called “effect biomarkers”) and improve our understanding of
possible health effects of PFAS exposures and fill important data gaps. Evidence for the
biological plausibility of a health-related endpoint can come from animal studies of PFAS
exposures, information on how PFAS exposures cause adverse effects (i.e., mechanistic
information), and epidemiological studies. Since PFHxS and PFOS serum levels were elevated in
the Pease population compared to national data, a Pease study should focus on data gaps
concerning the health effects of exposures to these chemicals. The feasibility assessment
included a literature search of epidemiological studies of PFAS exposures to identify the healthrelated endpoints evaluated in these studies and the data gaps that exist on the health effects of
PFHxS and PFOS.
3. Public health significance — a study should provide a basis for determining if PFAS exposures
increase the risks for specific adverse health effects, and if so, what public health actions are
necessary to reduce the risks. In particular, the study should provide a basis for early medical
intervention for health outcomes that are not routinely evaluated in physical exams. The study
should also be relevant to other populations with similar exposures.

8

In addition to the above criteria, a feasibility assessment must address specific questions:
1. Can the study population be enumerated and selected to minimize selection bias? (Selection bias
occurs when the probability of selection is related both to exposure status and to disease status.)
2. Is there an appropriate comparison population?
3. Is there a complete exposure pathway, well-defined exposed population, and ability to assign
levels of exposure with adequate accuracy?
4. Is there justification for studying the specific health outcome(s) being considered? (e.g., is there
suggestive biological evidence? A finding in a previous study?)
5. Can the health effect(s) be validly ascertained or measured?
6. Is the exposed population sufficiently large so that risks can be estimated with precision?
7. Can information be obtained on other risk factors that need to be taken into account?
8. Can a study answer the questions of concern to the Pease community?

Site history
The Pease International Tradeport is located in Portsmouth, New Hampshire. It contains over 250
companies employing more than 9,525 people. In 1993, companies began to operate at the Pease
Tradeport. Two day-care centers are located at the Tradeport. One of the day-care centers estimated that
about 695 children attended the center during 1996–2016. The other day-care center could not easily
compile total enrollment statistics, but its capacity is 220 children, they usually enroll about 180–195
children at a time, and they have been operating for almost 7 years. As of July 2015, the estimated
population of Portsmouth was 21,530 (http://www.census.gov/quickfacts/table/PST045215/3362900).
According to the 2010 census, 4.7% were children younger than 5 years, 11.9% were children ages 6–17
years, 67.5% were adults ages 18–64 years, and 15.9% were adults ages 65 years and older.
Additionally, 51.5% of the population were female, 91.5% were white, and 95.6% of persons ages 25
years and older were high school graduates.
The area on which the Tradeport is located was originally built in 1951 as part of the Pease Air Force
Base. In October 1989, 3,465 military personnel were assigned to the base, accompanied by 4,746
dependents. The Air Force estimated that 537 civilian employees worked on-base at that time (ATSDR
1999). During 1970–1990, an average of 3,000 personnel and their families were assigned to the base at
any one time. Before 1970, the base supported a maximum of 5,000 personnel (ATSDR 1999).
Three major supply wells provided drinking water to the base: the Haven, Smith, and Harrison wells.
Before 1981, the wells fed directly into the distribution system so that a particular area of base would
primarily receive water from the nearest well. After 1981, the water from the three wells were mixed
together and treated before entering the distribution system. These same three supply wells provided
drinking water to the Pease Tradeport after it opened.
In 1977, water from the base wells was found to contain trichloroethylene (TCE). Two of the three wells
serving the base were contaminated. The maximum concentrations of TCE measured in the Haven and
Harrison supply wells were 391 micrograms per liter (µg/L) and 28.5 µg/L, respectively. After the
discovery of the contamination, those wells were shut down and the city of Portsmouth supplied
drinking water to the base during 1977–1978. In the fall of 1978, the wells were back in operation. TCE
levels in the Haven well fluctuated between 50 µg/L and 115 µg/L from the fall of 1978 through January
1980, then fell below 50 µg/L, with an occasional spike above 50 µg/L through October 1980. From
9

November 1980 through July 1981, TCE levels averaged about 30 µg/L, then fell to around 10 µg/L
from August 1981 through May 1983. Levels continued to decline, but did not remain consistently
below the current U.S. Environmental Protection Agency (EPA) maximum contaminant level (MCL) in
drinking water of 5 µg/L until January 1986 (ATSDR 1999).
The base officially closed in October 1991, and most of the property was transferred to the Pease
Development Authority (PDA). During 1993, the business and aviation industrial park began operation.
The City of Portsmouth entered into a long-term lease and operation agreement with the PDA to operate
and maintain the public water system serving the Tradeport.
From approximately 1970 until the base closed, aqueous film-forming foam (AFFF) was used to
extinguish and prevent flammable liquid fires. AFFF was also used during firefighting training at the
base. Several perfluoroalkyl substances (PFAS) were used in the manufacturing of AFFF, including
perfluorooctanoic acid (PFOA), perfluorooctane sulfonate (PFOS), and perfluorohexane sulfonate
(PFHxS). AFFF containing PFAS likely leached into the soil and groundwater and migrated to the three
supply wells serving the Pease Tradeport. It is not known when these wells were contaminated with
PFAS, but it is possible that the contamination began before the opening of the Tradeport, when the Air
Force base was still in operation.
The Haven, Smith and Harrison wells have also served the Tradeport. In addition, the City of
Portsmouth has the capability to supply water to the Tradeport via its main distribution system. Monthly
pumping records for the three wells were provided by the City of Portsmouth, Department of Public
Works. Up through 1999, the Haven well on average provided about 56% of the total water supply at the
Tradeport, with the Smith well providing 44% and the Harrison well out of service. In 2000-2001, the
Haven well supplied 88% of the supply and the Smith well supplied 12%. From 2003 until it was taken
out of service in May 2014, the Haven well on average supplied about half the water supply. By 2006,
the Harrison well was back in service and the Smith and Harrison wells together supplied on average
about half of the water supply at the Tradeport. After May 2014, the Smith and Harrison wells supplied
56% of the Tradeport water supply and the City of Portsmouth provided the other 44%.
In 2009, EPA established provisional health advisory levels for PFOS and PFOA of 0.2 µg/L and 0.4
µg/L, respectively [US EPA 2009]. In 2013, sampling of monitoring wells at the former Pease Air Force
Base fire training areas detected PFOS and PFOA as high as 95 μg/L and 56 μg/L. In May 2016, EPA
established a new lifetime health advisory for PFOS and PFOA that said the combined concentrations of
PFOS and PFOA in drinking water should not exceed 0.07 µg/L [US EPA 2016a]. No drinking water
health advisory level has been established for PFHxS or other PFAS chemicals. While the EPA has a
lifetime health advisory for PFOS and PFOA, no federal regulatory standards for these contaminants
have been issued.
In April and May 2014, the three supply wells serving the Tradeport were sampled for PFAS. In the
April sampling, the Haven well had PFOS, PFOA, and PFHxS levels of 2.5 µg/L, 0.35 µg/L, and 0.83
µg/L, respectively. In the May sampling, the Haven well had PFOS, PFOA, and PFHxS levels of 2.4
µg/L, 0.32 µg/L, and 0.96 µg/L. Other PFASs were also detected in the Haven well. The Harrison well
had much lower levels of these contaminants with maximum PFOS, PFOA, and PFHxS levels of 0.048
µg/L, 0.009 µg/L, and 0.036 µg/L, respectively. The Smith well had maximum levels of PFOS and
PFHxS of 0.018 µg/L and 0.013 µg/L, respectively, with an estimated level of PFOA of about 0.004
µg/L.
10

No samples of the Pease Tradeport distribution system for PFAS are available from the period when the
Haven well was in operation. We can use a simple mixing model to estimate the PFAS levels in the
distribution system, assuming that contamination concentrations are approximately uniform throughout
the system. The model takes into account the pumping rates for each of the three wells, the total water
demand, and the concentrations of PFAS in the wells during the April and May 2014 sampling. Using
this simple approach, the estimated levels of PFOS, PFOA, and PFHxS in the Pease Tradeport
distribution system in April 2014 would be approximately 1.4 µg/L, 0.2 µg/L, and 0.5 µg/L,
respectively.
In April 2015, the City of Portsmouth created a community advisory board (CAB) to address the PFAS
contamination in the Tradeport drinking water. The CAB was established to act as a liaison between the
affected community and the New Hampshire Department of Health and Human Services (NH DHHS),
to represent the diverse views of the affected community, to review the blood testing conducted by NH
DHHS, and to provide input into future direction of the blood testing program (CAB 2015). The CAB
held 14 public meetings during May through December 1, 2015, and disbanded after issuing its final
report of its activities on December 21, 2015. Among the recommendations of the CAB in its final
report were the following:
1. Establish a community body to coordinate ongoing issues with ATSDR, NH DHHS, and the U.S.
Air Force’s Restoration Advisory Board at Pease and to provide an effective mechanism for
communication with all persons working or cared for at the Pease Tradeport.
2. A new community body should, along with its partner agencies, provide health education to the
public regarding environmental chemical exposures and how exposures and risks can be reduced.
In February 2016, ATSDR began recruiting community volunteers to serve as members of a Pease
community assistance panel (CAP). Technical advisors who could help CAP members in reviewing the
scientific information on PFAS and proposed health activities were also recruited. The purpose of the
CAP was to provide a mechanism for the community to participate directly in ATSDR’s health activities
related to the exposures to the contaminated drinking water at the Tradeport. The CAP would provide
input concerning possible health activities proposed by ATSDR. CAP members would also work with
ATSDR to gather and review community health concerns, provide information on how people might
have been exposed to hazardous substances, and inform ATSDR about ways to involve the community.
The first public meeting of the CAP was held in May 2016 in Portsmouth. The second public meeting
was held in September 2016. ATSDR has also convened monthly calls with the CAP.

Community concerns
The final report of the CAB, issued on December 21, 2015, noted that “…the lack of any definitive
information regarding the possible health effects of PFC [perfluorinated compound] exposure remains a
source of frustration and concern.” [CAB 2015] The report concluded, “There is a great need to better
understand what if any health effects might result for PFC exposure, and at what levels of exposure
these risks might be manifested.”
In an email sent to ATSDR in November 2015, the CAB asked that ATSDR consider the following
question: “What, if any, long-term health effects, such as specific cancers, elevated blood lipids, thyroid
function, immune function and developmental delays, are associated with the PFC exposure at Pease?
This question should be broken down with regard to specific populations including children,
11

nursing/pregnant women, firefighters, and adult exposed workers.” This question was reiterated at the
first in-person CAP meeting in May 2016. Some CAP members, as parents, were very concerned about
the health of their children who were exposed at a critical, early age of development while attending the
two day-care centers at the Pease Tradeport. They noted the lack of pediatric studies associated with
PFAS exposure and wanted ATSDR to consider testing the exposed children for health endpoints such
as lipids. CAP members also voiced concern about the exposed adult population, especially former
military service personnel and civilian workers at the former Pease Air Force Base. Concern was also
expressed for firefighters who were exposed to contaminated drinking water at Pease and also directly to
AFFF as part of their firefighting duties. CAP members expressed their desire for a longitudinal
approach (compared to a cross-sectional approach) to evaluate short-term and long-term health
conditions, including cancers.

Exposure assessment
Using the information currently available on PFAS concentrations in the supply wells during April and
May 2014, supply well pumping data, the total demand in the system, and assuming that PFAS
concentrations in the supply wells during the April–May 2014 sampling reflect historical concentrations
(given the persistence of these chemicals in the environment), a simple but crude assessment of PFAS
drinking water exposures could be conducted. However, to accurately estimate historical PFAS
concentrations in the Haven, Harrison, and Smith supply wells and the distribution system they served,
both during the operation of the Air Force base and the Tradeport, would require the following steps:
1. Obtain information on the locations and use of AFFF at the Air Force base, including accidental
releases.
2. Model the migration of contaminants from the soil where AFFF was used or released to the
groundwater and then to the supply wells.
3. Model the PFAS concentrations throughout the distribution system.

Historical reconstruction of PFAS concentrations in the drinking water distribution system would be
needed to assess exposures to service personnel and civilian employees who were at the Air Force base
during its operations, and to workers and day-care attendees at the Tradeport.
Another important source of information on exposures at the Pease Tradeport was the NH DHHS PFAS
blood testing program conducted during April–October 2015. A person was eligible for this program if
he or she had worked at, lived on, or attended childcare at the Pease Tradeport or Pease Air Force Base,
or lived in a home near the Pease Tradeport that was served by a PFAS-contaminated private well. A
total of 1,578 persons volunteered to submit a blood sample for PFASs testing [NH DHHS 2016]. This
was a convenience (or volunteer) sample, not a statistically based sample. Nevertheless, the testing
program provided important information on the extent and magnitude of exposures to the PFAScontaminated drinking water at the Pease Tradeport.
Table 1 shows the serum concentrations of PFOS, PFOA, PFHxS, and perfluorononanoic acid (PFNA)
for the 366 children younger than 12 years at the time of testing and comparison values from studies
conducted in Texas [Schecter 2012] and California (Wu 2015). Data from the National Health and
Nutrition Examination Survey (NHANES) are not available for children younger than 12 years.
12

NHANES testing for serum PFAS was restricted to those ages 12 years and older. The California study
[Wu 2015] conducted a random sample of households in northern California and obtained blood samples
from 68 children ages 2–8 years for PFAS analyses during December 2007–November 2009. The
parents of the children had higher education levels than the general population. The Texas study
[Schecter 2012] analyzed serum samples collected from 300 children ages ≤12 years at a children’s
hospital during 2009. Whether the children in the Texas study were healthy or receiving treatment for
illness was not reported. None of the California and Texas children were known to be exposed to PFAScontaminated drinking water. The children in both studies were considered to be representative of
general population exposures to PFAS via diet and consumer products.
Table 1 shows that the median and geometric mean serum PFHxS and PFOS levels in the Pease children
(ages <12 years) are considerably higher than background median and geometric mean levels seen in the
Texas and California studies. For PFOA, the Pease children have slightly higher levels than the
reference group in the Texas study, but lower than in the California study. However, the comparisons
with Texas and California results might not be appropriate given the difference in sampling years.
Nationally, serum levels of PFOS and PFOA have been declining sharply over time. For example, in the
1999–2000 NHANES cycle, the geometric mean serum PFOA level for persons aged ≥12 years was 5.2
µg/L. By the 2013–2014 cycle, it had declined to 1.9 µg/L. Serum PFOS declined even more sharply,
from 30.4 µg/L during the 1999–2000 cycle to 5.0 µg/L in the 2013–2014 cycle. PFHxS also declined,
but more gradually, from 2.1 µg/L during the 1999–2000 cycle to 1.3 µg/L in the 2013–2014 cycle. In
the NHANES 2013–2014 cycle, children ages 12–19 years had geometric mean PFOA, PFOS, and
PFHxS serum levels of 1.66 µg/L, 3.54 µg/L, and 1.27 µg/L, respectively. Therefore, the most
appropriate PFAS comparison values for the Pease blood testing program would be serum levels
obtained near in time to the Pease sampling (i.e., 2015). Such comparison values are not currently
available.
Table 2 shows the serum concentrations of PFOS, PFOA, PFHxS, and PFNA for the 1,212 participants
ages 12 years and older at the time of testing and comparison values from NHANES for 2013–2014 (the
most recent years data are currently available). Table 2 indicates that, similar to the children at Pease,
the median and geometric mean serum levels of PFHxS and PFOS among those ages ≥12 years are
considerably higher than those in the NHANES 2013–2014 cycle. The median and geometric mean
serum PFOA among those at Pease were also slightly elevated compared with NHANES results.
In analyses conducted by NH DHHS, geometric mean PFHxS serum levels were higher for persons who
drank ≥4 cups of water per day compared to those who drank <4 cups per day. Of all the PFAS serum
levels measured, water consumption had the strongest effect on PFHxS serum levels. In particular, water
consumption had the highest effect on PFHxS serum levels among persons aged ≤19 years (β = 0.31, SE
= 0.15, marginal effect = 36.4%). Geometric mean PFOS and PFOA serum levels were also higher
among persons who drank ≥4 cups of water per day compared with those who drank <4 cups per day
[NH DHHS 2016]. Linear trends were observed for geometric mean serum levels of PFOS, PFOA, and
PFHxS and increasing time spent at the Pease Tradeport. The trend was strongest for PFOS and PFHxS
[NH DHHS 2016].

13

Summary of literature review
ATSDR reviewed published health studies to identify health-related endpoints that have been studied
and the data gaps that exist, in particular, for the effects of PFHxS and PFOS. The literature review also
was used to identify adverse effect sizes observed in the PFAS studies for PFAS serum levels similar to
those found in the Pease population.
The Appendix has a listing of the epidemiological literature on PFAS exposures and adult cancers, other
adult diseases, and adverse outcomes in children. Tables 3 and 4 provide a summary. In these tables, a
“+” indicates that at least one study had a finding for a specific PFAS chemical that suggests an
increased risk of an adverse outcome (e.g., an odds ratio [OR] or risk ratio [RR] of ≥1.20), and a “*”
indicates that no study has been conducted for that PFAS chemical. In these tables, an “I” indicates that
the findings from studies have not suggested an increased risk for an adverse outcome (e.g., all odds
ratios or risk ratios are <1.20) but the information is too limited to conclude that there is no association
between the PFAS exposure and the adverse outcome.
These tables are for illustrative purposes, to indicate where data gaps exist and therefore additional
research may be needed. Tables 3 and 4, and the tables and descriptions of the studies in the appendix,
should not be interpreted as implying causation or as an assessment of the weight of evidence for an
association. Currently, epidemiological research on the health effects of PFAS exposures is at an early
stage. This is particularly true for PFHxS in addition to PFAS chemicals other than PFOA and PFOS.
However, even for PFOA and PFOS, additional research on all the health-related endpoints mentioned
in these tables will be needed to provide sufficient evidence for causal assessments and to address
community health concerns.

Adult cancers and other adult diseases
Based on its assessment of the epidemiological literature, ATSDR concluded that there was limited or
no information concerning associations with PFAS exposures and most cancers and other adult diseases
(Table 3). In particular, very few studies have evaluated PFHxS exposures and cancers and other adult
diseases. Although more information is available for PFOS exposures and cancers and other adult
diseases than for PFHxS exposures, the information is still very limited and therefore inadequate to
determine whether PFOS exposures increase the risk for most of the adult diseases evaluated. Although
more information is available on PFOA exposure, the information is still too limited to determine
whether a causal association exists between PFOA and specific cancers and other adult disease.
Therefore, additional research on the effects of PFHxS, PFOS, and PFOA would be needed to determine
whether exposures increase the risk for many adult cancers and non-cancer diseases.

Health effects in children
There is some evidence that PFAS exposures are associated with decreased birth weight, small fetus size
for gestational age, measures of intrauterine growth retardation, and preterm birth. In particular, two
meta-analyses have found an overall decrease in birthweight associated with PFOA and PFOS [Verner
2015; Bach 2015]. However, the findings across studies are inconsistent for these outcomes and for
14

other adverse birth outcomes, and few studies have evaluated PFHxS. Several studies of infants have
found that prenatal PFAS exposures affect thyroid function, but only two studies have evaluated thyroid
function in older children. A few studies have found elevated uric acid with PFAS exposures, but the
possibility of reverse causation cannot be ruled out. Four studies of PFAS exposures and testosterone
and other sex hormones have been conducted. However, the findings have not been consistent across
studies and further research is needed. Three of the studies did find that PFAS exposures decreased
testosterone in boys or girls. There is some evidence from four studies that PFAS exposures might be
associated with ADHD, but findings have not been consistent across studies. Evaluating the evidence for
PFAS exposures and neurobehavioral outcomes is difficult for several reasons: 1) the studies used
different methods to measure the outcomes, 2) studies are inconsistent in the outcomes evaluated, and 3)
too few studies have been conducted. A few studies have found associations between PFAS exposures
and a decline in antibody response to specific vaccines, but only two studies evaluated the same vaccine
(i.e., rubella). In summary, there are considerable data gaps concerning the health effects in children of
PFAS exposures. This is because of the small number of studies conducted, inconsistencies in methods
and findings across studies, and limited sample sizes in some studies. As for other adverse outcomes,
few studies have evaluated the effects on children of PFHxS exposures.

Sources of adverse outcome data for the Pease population
The adverse outcomes of interest for PFAS exposure that can be ascertained from the birth certificate are
pregnancy-induced hypertension, diabetes, small for gestational age (SGA), low birth weight, birth
weight, preterm birth, and gestational age. Although the birth certificate has a checklist for congenital
anomalies, the most reliable data on birth defects are provided by population-based birth defect
registries. Birth defects registries exist in 41 states, including New Hampshire. The New Hampshire
Birth Conditions Program (NHBCP), based at the Geisel School of Medicine at Dartmouth College,
began collecting data on births occurring in-state to New Hampshire residents in 2003
(http://www.cdc.gov/ncbddd/birthdefects/states/newhampshire.html). Data reported on 46 different birth
defects are ascertained for infants aged ≤1 year are collected through active surveillance methods.
Congenital hypothyroidism data can be obtained from the newborn screening program. Newborn
screening for congenital hypothyroidism is conducted in every state, including New Hampshire.
The birth certificate has information on sex of the child, plurality, gestational and pre-pregnancy
diabetes, previous preterm birth, parity and gravidity, cigarette smoking before and during pregnancy,
principal source of payment for the delivery (a measure of socio-economic status), date of last
pregnancy, date of last normal menses, date of first and last prenatal care visit and total number of
prenatal care visits, race/ethnicity of the mother and father, education of the mother and father, parents’
names and address, mother’s marital status, labor and delivery complications, and whether the infant is
being breastfed at discharge. The New Hampshire Division of Vital Records Administration collects
information on births in New Hampshire from hospitals and midwives, birth certificates, and interstate
exchange agreements for births occurring out-of-state to New Hampshire residents
(http://www.dhhs.nh.gov/dphs/hsdm/birth/ ).
Mortality information is available from the National Death Index (NDI) operated by the National Center
for Health Statistics (NCHS), Centers for Disease Control and Prevention. Currently, 2014 data are
complete and available for searches. “Early release data” for 2015 are ≥90% complete (98% complete
for New Hampshire) and also available for searches. NDI “plus” provides information on cause of death
(underlying, contributing and all other causes of death listed on the death certificate) and date and state
of death based on death certificate data provided by the states. The NDI has data starting from 1979.
15

New Hampshire death certificate data are available from the New Hampshire Division of Vital Records
Administration, which collects information on deaths of New Hampshire residents and deaths occurring
in New Hampshire (http://www.dhhs.nh.gov/dphs/hsdm/death/index.htm). Information on deaths of
New Hampshire residents that occur out-of-state is captured through interstate exchange agreements.
Information on underlying cause of death and up to 14 contributing causes of death is collected.
Complete data are available approximately 24–48 months after the close of a calendar year.
Population-based cancer registries exist in all 50 states and Washington, DC. The New Hampshire State
Cancer Registry (NHSCR) is a statewide, population-based cancer surveillance program that has
collected incidence data on all cancer cases diagnosed or treated in the state since 1985
(http://geiselmed.dartmouth.edu/nhscr/). NHSCR, which is contracted to the Geisel School of Medicine
at Dartmouth College, currently collects data from the larger hospitals in the state. NHSCR also receives
case reports from physician practices, free standing radiation oncology centers, pathology laboratories
and other sources. NHSCR staff assist hospitals with fewer than 100 cases per year with reporting.
Through interstate data exchange agreements, NHSCR also receives case reports for New Hampshire
residents who are diagnosed outside the state.
The New Hampshire Uniform Hospital Discharge Data Set (UHDDS) collects discharge data from all
health care facilities in the state (acute care hospitals, specialty hospitals, freestanding hospital
emergency facilities, and walk-in urgent care centers), as required by law
(http://www.dhhs.nh.gov/dphs/hsdm/hospital/index.htm). Discharge data from Maine, Massachusetts,
and Vermont hospitals for New Hampshire residents are included in the UHDDS via interstate data
exchange agreements. The dataset includes transfers of NH residents. Chronic diseases such as asthma,
chronic obstructive pulmonary disease, angina, hypertension, congestive heart failure, hypoglycemia,
and diabetes are included in the UHDDS. Limitations of this dataset are that discharges are not deduplicated and one person with multiple admissions might falsely increase the number of persons
hospitalized. Additionally, state law requires health care professionals to report information on chronic
health conditions relating to children, infectious diseases, immunizations, and autism to NH DHHS
(http://www.healthinfolaw.org/state-topics/30,67/f_topics).
To ascertain autism or ADHD reliably, a review of school special education records and medical records
from providers that conduct developmental evaluations of children or provide treatment is necessary. In
Portsmouth, records are available from three elementary schools (serving grades K–5), one middle
school (serving grades 6–8), and one high school (serving grades 9–12). Projected enrollment for the
2016–17 school year was 988 students in the elementary schools, 516 students in the middle school, and
1,183 students in the high school (http://cityofportsmouth.com/school/FY16BudgetBooklet.pdf). In
school year 2015–2016, the Portsmouth Public Schools provided special education services to 416
students. Among those students, 121 (29.1%) had an orthopedic impairment, 36 (8.7%) had a
speech/language impairment, 32 (7.7%) had a developmental delay, 25 (6.0%) had autism, 17 (4.1%)
had an emotional disturbance, 11 (2.6%) had some other disability, and 174 (41.8%) were classified as
having a “specific learning disability.”
Various studies have focused on West Virginia and Ohio residents and workers exposed to PFOA from a
chemical plant (the “C8” studies) [Frisbee 2009]. In a C8 study that evaluated ADHD, affected persons
were identified via questionnaire, which included a question requesting information on medications used
[Stein 2011]. For chronic diseases, the C8 studies relied primarily on self-reported information from
questionnaires with attempted confirmation of self-reports by obtaining medical records.
16

Sources of exposure data
An important source of exposure information is PFAS biomonitoring. Measuring serum levels of PFAS
chemicals provides information on the amount of these chemicals that has entered the body from all
sources. At Pease, 1,578 persons volunteered to submit blood samples for PFAS analyses during the NH
DHHS biomonitoring program in 2015. In the C8 study, blood samples for PFAS analyses were
obtained from 66,899 persons during the 13-month baseline period, 2005–2006 [Frisbee 2009].
Biomonitoring for PFAS is useful in estimating past exposures, given the long half-lives of PFOS
(approximately 5.4 years) and PFHxS (approximately 8.5 years). Although biomonitoring integrates
PFAS exposures from all sources, including diet and consumer products, PFAS levels in serum from
populations exposed to PFAS-contaminated drinking water will mostly reflect the drinking water
exposures, unless the person is or was also exposed occupationally (e.g., firefighters, PFAS
manufacturing workers).
The use of PFAS biomonitoring in epidemiological studies has some limitations. A key limitation is the
issue of “reverse causation,” in which the disease under investigation (e.g., kidney disease or kidney
function) affects the elimination of PFAS in the body, causing higher serum levels of PFAS. Other
problems include potential confounding by a factor that is both a risk factor for the disease of interest
and a factor influencing serum PFAS levels (e.g., parity in the evaluation of adverse birth outcomes).
Another limitation is that biomonitoring results, by themselves, might not provide sufficient information
to estimate historical exposures. Estimating historical exposures is necessary to assess cumulative
exposure and to characterize periods of special vulnerability to PFAS exposures, such as prenatal or
early childhood exposures.
Modeling methods are used to reconstruct historical PFAS serum levels. The results of PFAS
biomonitoring can be used to validate estimates of PFAS serum levels obtained from modeling.
C8 researchers have successfully used physiologically based pharmacokinetic modeling of absorption,
distribution, metabolism, and excretion of PFOA in the body in conjunction with drinking water
contaminant levels, estimates of water intake, and residential history to predict historical and current
PFOA serum levels [Shin 2011]. Researchers have also been able to simulate PFOS serum levels using
information on drinking water levels and PBPK modeling [Loccisano 2011]. Therefore, reconstruction
of historical PFOS serum levels is also feasible. However, reconstruction of PFOA and PFOS serum
levels is limited by various uncertainties. These include lack of accurate information on individual
consumption of drinking water and length of time exposed and limited information on factors that
produce inter-individual variability (e.g., gender, age) and pre-existing medical conditions (e.g.,
compromised renal function) [Loccisano 2011]. Nevertheless, the ability to predict serum PFOS and
PFOA levels based on drinking water contamination levels can substitute for, and enhance, the
information provided by PFAS biomonitoring.

Issues concerning cross-sectional study designs
Cross-sectional studies are especially suitable for assessing effect biomarkers and the prevalences of
nonfatal diseases, in particular, diseases with no clear point of onset [Checkoway 2004]. However, if the
cross-sectional study concurrently measures the exposure and the outcome (i.e., the disease or effect
biomarker), then it might be difficult to determine whether the exposure caused the outcome or whether
the outcome influenced the measured exposure level [Flanders 1992, 2016]. For example, as discussed
above, the concurrent measurement of serum PFAS levels and kidney function biomarkers might raise
the question of “reverse causation” because kidney function can affect the levels of PFAS in serum. This
17

issue can be addressed by estimating exposures based on the historical reconstruction modeling of serum
PFAS levels. In addition, it might be possible to estimate exposures during critical vulnerable periods
(e.g., in utero exposure) through the modeling of historical serum PFAS levels. However, the modeling
of historical PFAS serum levels is subject to uncertainties and data limitations, as discussed above, and
published methods are available only to model serum levels of PFOA and PFOS.
Other issues concerning cross-sectional study designs are similar to those that confront other
observational study designs, such as cohort studies. These issues include: 1) the ability to clearly define,
enumerate and recruit (without introducing selection bias) the exposed and comparison populations, 2)
the comparability of the exposed and comparison populations on risk factors other than the PFAS
exposures, 3) accurate exposure assessment, and 4) accurate measurement of effect biomarkers and
ascertainment of diseases.
Based on its review of the literature, ATSDR concludes that several health-related endpoints could be
considered for studies of the Pease population. It is also clear that exposures to the PFAS-contaminated
drinking water have occurred in the Pease population, as documented by the observed serum PFAS
levels in the NH DHHS PFAS blood testing program. Therefore, it is reasonable to conduct
epidemiological studies of the Pease population. However, whether it is feasible to study a specific
health-related endpoint depends to a great extent on the size of the exposed population that can be
recruited into a study. The usual approach to determine the necessary size of the study population for
each health-related endpoint is to conduct sample size calculations.
All epidemiological studies of environmental exposures and health outcomes have limitations and
uncertainties. Whether a study will find an association between an environmental exposure and health
effects cannot be known prior to conducting the study. No single study of the Pease population will
provide definitive answers to the community about whether their exposures to the PFAS-contaminated
drinking water caused their health problems. The ability of a study of the Pease population to provide
useful information will depend to a great extent on the success of recruiting a sufficient number of study
participants.

Feasibility of an epidemiological study of children at the Pease Tradeport
The first population that ATSDR considered for an epidemiological study was the children who attended
the two day-care centers at the Pease Tradeport. One reason to focus on children is that they are more
vulnerable to environmental exposures, in particular exposures to potential endocrine-disrupting
chemicals. In addition, there is serious concern in the community about the possible health effects to
children from the drinking water exposures, which was conveyed to ATSDR by the Pease CAP. Finally,
a study of children who attended daycare at the Pease Tradeport is the most feasible epidemiological
study to conduct. The population is less transient than an adult population and the adverse health
endpoints of interest do not require as large a sample size as adult chronic conditions.
The public health significance of conducting a study of these children consists of 1) the possibility of
early intervention if early signs of adverse health effects, including developmental delays, are observed
and 2) the relevance of a study at Pease for other populations exposed to drinking water primarily
contaminated with PFOS and PFHxS. A study of children at Pease would have scientific importance
because of key data gaps concerning PFAS exposure effects on sex hormones and on neurobehavioral,
18

immunological, and thyroid function. Animal studies support the biological plausibility of immune
effects. Animal data also suggest that PFAS might be developmental neurotoxicants that can alter
cognitive function and reduce learning ability. PFAS also have endocrine-disruptive properties and
could interfere with thyroid function and sex hormones. A study of children at Pease would be
responsive to the community’s concerns and has the potential (from the perspective of statistical power)
to provide meaningful and credible results for some of the adverse outcomes of interest. However, a
study limited to the population of children who attended the Pease Tradeport day-care centers would
likely not be sufficiently large for some of the possible adverse outcomes of interest (e.g., higher
prevalences of rare diseases or very subtle changes in biomarkers of effect that have been observed in
research conducted elsewhere).

A. Study population
The population of interest could be persons who attended day care at the Pease Tradeport before June
2014 and are in the age range of 4−17 years at the start of the study. The end of the period was selected
because the Haven well was taken out of service in May 2014. Because PFAS-contaminated drinking
water exposures could occur to children in utero and during breastfeeding if the mother worked at the
Pease Tradeport, the study would include these additional children if the exposures began prior to June
2014 and their ages are 4 – 17 years at the time the study begins.
The age range for the Pease children study was determined by taking into account the age ranges in
previous PFAS studies and the age range appropriate for the candidate endpoints. Previous
epidemiological studies of children exposed to PFAS included varying age ranges. Because of data
limitations (i.e., no PFAS serum data for those aged <12 years), the studies that used NHANES data
evaluated those aged 12–18 years or 12–19 years. Some of the C8 studies limited participant ages to
those <12 years; other C8 studies included persons up to 18 years of age. The upper age limit for many
of the Taiwan children studies of PFAS was 15 years. An age range of 4–17 years would overlap the age
ranges in these studies.
The chosen age range also reflected the focus of the study (i.e., children exposed to the PFAScontaminated drinking water while attending daycare at the Pease Tradeport). The younger age limit of 4
years was chosen because intelligence quotient (IQ) testing is available for those aged 4 years and older.
(For example, the Wechsler Preschool and Primary Scale of Intelligence test has an age band of 4 years
to 7 years, 7 months that overlaps the Wechsler test for those aged 6–16 years.) The Strengths and
Difficulties Questionnaire (SDQ), a behavioral screening questionnaire used in a Faroes study [Oulhote
2016], a Taiwan study [Lien 2016] and a Danish study [Fei 2011] has an age range of 4 – 16 years. The
upper age limit of 17 years was chosen for three reasons:
1. Age at puberty was a candidate endpoint and virtually all of the children in a C8 study achieved
puberty by age 17 years.
2. The IQ and SDQ testing instruments for children can be used for those aged ≤17 years.
3. Children aged >17 years would have been last exposed (i.e., last attended daycare) more than 10
years ago.

19

Table 5 provides the data on serum PFOS, PFOA, and PFHxS for the 379 children who participated in
the 2015 NH DHHs testing program at Pease and who were aged 1–14 years at the time of blood draw.
These children would be aged 4–17 years in 2018. The geometric mean serum PFHxS in these children
was 3.75 µg/L, approximately three times higher than the serum levels reported in the Texas [Schecter
2012] and California [Wu 2015] studies and in the NHANES data for 2013–2014.
We currently do not know how many children attended daycare at the Pease Tradeport before June 2014
and who would be in the 4–17 years age range in 2018. The Discovery Child Enrichment Center is
located at the Pease Tradeport and began operation in 1994. Its yearly enrollment is approximately 149
children ages 6 weeks to 5 years. Computerized records at this day-care center start in 1996. A
preliminary records search by the director of the Discovery Child Enrichment Center identified 695
children who attended the daycare during 1996–2015 and who would be aged of 6–18 years in 2018.
Based on the results of this search, the number of children who attended this day care prior to June 2014
and would be between the ages of 4 and 17 years in 2018 could be within the range of 250 – 450
individuals.
The Great Bay Kids’ Company is also located at the Pease Tradeport and began operation in 2010. Its
annual enrollment is approximately 270 children aged ≤12 years. Assuming that most of the children
enrolled would be ≤5 years of age, and that most of the children attend daycare for 4 years, about 300
children might have attended this daycare during the period of interest and would be aged 4–17 years in
2018.
Assuming that a minimum of about 500 children attended the two day-care centers at Pease before June
2014 and would be aged 4–17 years in 2018, and assuming a reasonable participation rate of 70%, it
would be possible to recruit 350 Pease children into the study. It would also be feasible to recruit at least
175 children in the same age range from the public schools in Portsmouth, NH, who were unexposed to
the PFAS-contaminated drinking water at the Pease Tradeport and whose parents did not work at the
Pease Tradeport or have occupational exposures to PFAS. It is reasonable to assume that participation
rates would be high because of strong interest in the community concerning the Pease Tradeport
situation. Moreover, the Pease CAP members have pledged to support recruitment efforts if and when a
study is to be conducted. Pease CAP members have strong ties and are active in the Portsmouth
community. If the actual number of children who attended the two day-care centers prior to June 2014
and would be aged 4 – 17 years in 2018 is in the range of 650 – 750, then as many as 500 children could
be recruited from the Pease population. It should also be possible to recruit at least 250 children in the
same age range from the Portsmouth public schools for the unexposed group.
A sample size of 350 exposed children and 175 unexposed children would be similar to the sample sizes
used in the Faroes study [Grandjean 2012, 2016] and in a C8 study of 320 exposed children [Stein 2013,
2014b]. However, the sample size of 350 exposed and 175 unexposed would be considerably smaller
than most of the C8 children studies and some of the other epidemiological studies of children exposed
to PFAS. Therefore, a total of 525 children, 350 exposed and 175 unexposed, should be considered a
minimum sample size, and attempts should be made to recruit a higher number of exposed and
unexposed children to improve the statistical power of the study.

20

B. Study Hypotheses
As indicated in the literature review summary, the scientific literature has little information on the health
effects of exposures to PFHxS. PFHxS is a key contaminant associated with the use of AFFF for
firefighting training and extinguishing flammable liquid fires. The study would be an important
contribution in filling this data gap and would generate knowledge relevant to other populations exposed
to drinking water contaminated by PFHxS from the use of AFFF. In addition, few studies have been
conducted to evaluate possible associations between childhood exposures to PFASs and effects on
thyroid function, uric acid and sex hormone levels, delays in reaching puberty, IQ, and immune
function. Inconsistent findings have been observed for most of these endpoints, likely in part because of
differences in exposures (e.g., drinking water and other sources, such as diet) and PFAS levels of
exposure, study population differences (e.g., age differences), and differences in methods. Moreover,
few studies have evaluated the same neurobehavioral or immune endpoint. The study would address
these issues by using methods and evaluating health effects similar to those used in previous studies of
PFAS exposures in children, in particular, methods used in the C8 studies.
Based on the literature review, the following hypotheses could be evaluated:
1. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher total cholesterol, lowdensity lipoprotein, and triglycerides, and higher prevalence of hypercholesterolemia.
2. Higher serum levels of PFOA, PFOS, or PFHxS are associated with differences in thyroid
stimulating hormone (TSH), TT4, and TT3, and a higher prevalence of hypothyroidism.
3. Higher serum levels of PFOA, PFOS, or PFHxS are associated with a higher level of uric acid
and a higher prevalence of hyperuricemia.
4. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher levels of cytokeratin18 (CK-18), a biomarker for fatty liver disease.
5. Higher serum levels of PFOA, PFOS, or PFHxS are associated with differences in testosterone,
estradiol, and sex hormone-binding globulin (SHBG).
6. Higher serum levels of PFOA, PFOS, or PFHxS are associated with delayed puberty.
7. Higher serum levels of PFOA, PFOS, or PFHxS are associated with lower IQ.
8. Higher serum levels of PFOA, PFOS, or PFHxS are associated with ADHD behaviors and
learning problems.
9. Higher serum levels of PFOA, PFOS, or PFHxS are associated with a higher prevalences of
hypersensitivity-related outcomes (e.g., asthma, rhinitis infectious diseases).
10. Higher serum levels of PFOA, PFOS, or PFHxS are associated with lower antibody responses to
rubella, mumps, and diphtheria vaccines.

21

C. Recruitment and Consent
Based on sample size calculations (see Appendix), a minimum of 350 exposed children aged 4–17 years
who attended the day-care centers at Pease before June 2014 would need to be recruited. To recruit the
children who participated in the blood testing program, NH DHHS would have to send letters to the
parents to ask that their child participate in the study. Additional children who were exposed to the
contaminated drinking water while attending the two day-care centers could be recruited via outreach to
the two day-care centers at Pease, the Portsmouth public schools, media, and community organizations
in the Portsmouth area. The Pease CAP has also offered to assist in recruitment, and CAP involvement
will be crucial in achieving high participation rates.
A minimum of 175 children aged 4–17 years, who were unexposed to the PFAS-contaminated drinking
water at the Pease Tradeport and whose mother did not work at the Pease Tradeport (or in an occupation
that involved PFAS exposure) during the pregnancy and breastfeeding of the child would be recruited
from the Portsmouth, NH, public schools. Before enrollment in the study, the child’s mother would be
interviewed to determine whether the child is eligible for the study. Recruitment would involve outreach
to the eight day-care centers in Portsmouth that were located outside the Pease Tradeport, the
Portsmouth public schools, media, and community organizations. The Pease CAP has offered to help
with the recruitment effort. The total enrollment of Portsmouth’s elementary, middle, and high schools
is projected to be 2,687 in 2016–17. To encourage participation of exposed and unexposed children, an
appropriate incentive would be provided.
The Pease blood testing program’s consent form was strictly limited to the use of the participant’s blood
sample for PFAS analyses only. The participant also consented to complete a brief questionnaire at the
time of blood draw concerning demographic information, time at Pease Tradeport, and consumption of
drinking water. The consent form did not mention the use of the blood sample for research purposes or
the possibility of re-contacting the participant for future studies. Moreover, the amount of blood drawn
from the children was only sufficient for the PFAS analyses. Therefore, ATSDR cannot directly contact
the participants in the Pease blood testing program to recruit them for a children’s study. In addition,
these participants must sign a new consent form to participate in a research study.
A parent of each child would be asked to sign a parental permission form requesting a blood sample
(about 4 teaspoons or 20 mL) from the child for the analyses of PFASs and the effect biomarkers (i.e.,
lipids, TSH, uric acid, sex hormones, and immune function parameters). The consent form would also
ask that the child be administered the Wechsler Abbreviated Scale of Intelligence (IQ) tests if aged 6
years or older or the Wechsler Preschool and Primary Scale of Intelligence for children younger than 6
years. The consent form would ask permission to access the child’s school records, including special
education records. The parent would be asked to sign a consent form to complete a questionnaire.
Children ages 7 years and older would be asked to give their assent to participate in the study.

D. Questionnaire
The parents of the child participant could be asked to complete the questionnaire. The questionnaire
could obtain demographic information, medical history of the parents and child, the child’s medications,
the dates the child’s mother worked at the Pease Tradeport (or in other occupations involving PFAS
exposures) and her reproductive history, the dates the child attended daycare at the Pease Tradeport,
water consumption of the mother and child while at Pease Tradeport (including use of water for formula,
22

juices, etc.) if applicable, bottled water consumption by the mother and child, length of time the child
was breastfed, parental information (e.g., education, primary occupation, maternal age at birth of the
participating child), the child’s height and weight, and whether the child regularly exercises, currently
smokes (and the number of cigarettes/day), or consumes alcohol (and the number of drinks/week).
Specific questions could be included in the questionnaire that address health outcomes of interest based
on the final study design. For example, for ADHD, the questionnaire could ask, “Has a doctor or health
professional ever told your child that your child has/had ADD or ADHD?” If the answer is “yes,” a
second question could ask for a list of medications being used for the condition. Parents would also be
asked if the child had learning or behavioral problems, and if so, the type of problem and the treatment
being used. Questions would be included for the hypersensitivity-related outcomes, asthma, atopic
dermatitis (or atopic eczema), and allergies. Information on the child’s vaccination history would also be
requested from the parents. The parents would also be asked when the female child first began to
menstruate.

E. Biomarkers of exposure and effect
The following biomarkers of lipids, thyroid function, kidney function, sex hormones, nonalcoholic
steatohepatitis (fatty liver), and immune function could be analyzed in the serum:
•
•
•
•
•
•

Total cholesterol, low density lipoprotein, high density lipoprotein, total triglycerides
Thyroxine (T4), T3, thyroid stimulating hormone (TSH)
Uric acid, creatinine
Cytokeratin-18 (CK-18) fragment levels (fatty liver disease)
Testosterone, estradiol, sex hormone-binding globulin (SHBG), follicle stimulating hormone,
insulin-like growth factor
Immunoglobulin G (IgG), IgA, and IgM; antibodies to measles, mumps, rubella, tetanus, and
diphtheria

Approximately 4 teaspoons of blood (20 mL) could be drawn from each participant to be analyzed for
the standard panel of PFAS compounds and the effect biomarkers. An attempt would be made to obtain
an 8-hour fasting blood sample. The parents could be asked how long the child fasted before the blood
draw. The cut points of 50 ng/dL of total testosterone and 20 pg/mL of estradiol would be used to
identify sexual maturation in boys and girls, respectively. IgG antibodies for measles, rubella, and
diphtheria would be analyzed to determine vaccine responses. Allergen-specific IgE (mold, dust mites,
dog, cat, cow’s milk, peanut, hen’s egg, and birch) could be analyzed. Serum levels of thyroid
stimulating hormone (TSH) and total T4 could be analyzed separately and also used to determine
clinical and subclinical hypothyroidism. Uric acid, total cholesterol, low-density and high-density
lipoprotein, and triglycerides could be analyzed.
For children older than 6 years, the Wechsler Abbreviated Scale of Intelligence could be administered to
the child to assess verbal IQ, performance IQ, and full-scale IQ. For children aged 4–6 years, the
Wechsler Preschool and Primary Scale of Intelligence would be administered. For each child, school
records, including special education records could be reviewed to identify learning problems and
behavioral problems. The SDQ could be administered to parents to assess emotional, conduct, and peer
relationship problems as well as problems with hyperactivity and inattention.
23

F. Exposure Assessment
As stated earlier, the analyses by NH DHHS of the data from the blood testing program at Pease
indicated that geometric mean PFHxS serum levels were higher for persons who drank ≥4 cups of water
per day than for those who drank <4 cups per day. The strongest finding was for serum PFHxS in
participants aged 0–19 years and water consumption (β = 0.31, SE=0.15, marginal effect=36.4%).
Geometric mean PFOS and PFOA serum levels were also higher among those who, while at the
Tradeport, drank ≥4 cups of water per day than for those who drank <4 cups per day [NH DHHS 2016].
Although these findings are based on a “convenience sample” (or a “volunteer sample,” i.e., not a
statistically-based sample), it is clear from these results that consumption of PFAS-contaminated
drinking water at the Pease Tradeport was a complete exposure pathway.
Study participants could submit blood samples for PFAS and biomarker analyses during 2018. For those
who participated in the 2015 blood testing program, these measurements would be used to assess their
exposures. For those who did not participate in the 2015 blood testing program but who attended
daycare at the Pease Tradeport during January 2008–May 2014, the PFAS serum levels obtained in 2018
could be used to estimate serum levels during 2015 by adjusting for PFAS elimination rates and taking
into account background PFAS exposures. For those who consumed drinking water from the Pease
Tradeport after the Haven well was taken out of service, the adjustment could also take into account the
PFAS levels in the drinking water after May 2014. The 2015 (estimated or measured) PFAS serum
levels and 2018 measured PFAS serum levels would be used in the analyses.
No water samples from the Pease Tradeport distribution system for PFAS testing are available before
2014. Using a simple mixing model that takes into account the pumping rates for each of the three wells,
the total water demand, and the concentrations of PFAS in the wells during the April and May 2014
sampling, we can estimate historical PFAS levels in the distribution system, assuming that
contamination concentrations are approximately uniform throughout the distribution system and
assuming that the contamination was present at least from 2008 through May 2014.
To estimate serum levels of PFOA and PFOS over the child’s life, the historical estimates of the
drinking water contamination could be combined using PBPK modeling with information from the
questionnaires on 1) the dates and length of time the child attended daycare at the Tradeport and the
child’s consumption of drinking water at the daycare and 2) whether the child’s mother worked at the
Pease Tradeport during pregnancy and during the period of breastfeeding and the length of the period
when the child was breastfed. PBPK modeling estimates would also incorporate information from
NHANES and from the PFAS serum levels of the unexposed comparison group to estimate background
levels of PFAS in serum. For those children whose mothers worked at the Pease Tradeport, estimates of
the mother’s serum levels during the pregnancy and breastfeeding of the child would be needed. If the
mother participated in the 2015 blood testing program at Pease, her measured PFAS serum levels could
be used in the modeling. Children’s serum levels from the 2015 NH DHHS Pease blood testing program
and serum levels obtained for this study would be used to calibrate the PBPK models.
No human PBPK model for PFHxS is currently available. However, correlation coefficients for serum
PFHxS and serum PFOS and PFOA were quite high among persons ages 2–14 years who participated in
the 2015 testing (Pearson correlation for PFHxS was 0.75, and for PFOS and PFOA was 0.73).
Therefore, it might be possible to predict historical serum levels of PFHxS based on historical estimates
for serum PFOA and PFOS.
24

G. Sample Size
The sample size for the Pease children study should include at a minimum 350 exposed children. It
should also include a minimum of 175 unexposed children randomly sampled from the Portsmouth
public schools with frequency matching to the exposed children on age, sex, and race. This minimum
sample size is based on several considerations. First, 379 children ages 1–14 years participated in the
2015 blood testing at Pease. That would be about a 75% participation rate, assuming that a minimum of
500 children attended daycare at Pease and would be in that age range in 2015. It should be possible to
recruit a similar percentage of the children who attended daycare at Pease. However, children who did
not participate in the 2015 blood testing would have to be recruited, as well as a high percentage of those
who did participate. Second, some studies conducted of PFAS exposure and children had similar or
smaller sample sizes than the 350 exposed and 175 unexposed children at Pease (e.g., Zeng [2015] and
Qin [2016] in Taiwan, Grandjean [2012] in the Faroes, Stein [2013] in a C8 study of neurobehavioral
effects, Hoffman [2010] in a NHANES study), but had sufficient statistical power to observed findings
to achieve statistical significance. Finally, sample size calculations conducted for this feasibility
assessment indicated that at least some of the health-related endpoints of interest could be evaluated,
with sufficient statistical power (i.e., statistical power ≥80%) to detect effects of exposure that are equal
to or greater than those listed in Tables 6a and 6b as well as effects observed in other PFAS studies that
occurred at PFAS serum levels similar to those in the Pease children population.
Sample size calculations were conducted using four different combinations for type 1 error (α error or
false positive error) and type 2 error (β error, false negative error, or 1 – statistical power):
1. Type 1 error = 0.05 (corresponds to a two-tail hypothesis test using a p-value cutoff of 0.05, or a
95% confidence interval, to determine statistical significance) and a type 2 error = 0.05
(corresponding to statistical power of 95%).
2. Type 1 error = 0.05 and type 2 error = 0.20 (80% power).
3. Type 1 error = 0.10 (corresponds to a one-tail hypothesis test using a p-value cutoff of 0.05, or a
90% confidence interval, to determine statistical significance) and a type 2 error = 0.10 (90%
power).
4. Type 1 error = 0.10 and type 2 error = 0.20 (80% power).
(Note: Setting the type 1 and type 2 errors to be equal indicates an equal concern for false negatives and
false positives and could be justified from a public health perspective.)
Table 6a indicates the minimum effect sizes that can be detected with a sample size of 350 Pease
children and 175 unexposed children from the Portsmouth area using the four combinations of type 1
and type 2 errors. Table 6b also includes the minimum effect sizes that can be detected with a sample
size of 500 exposed and 250 unexposed. These minimum effect sizes assume a simple comparison
between the exposed and unexposed children that is not adjusted for possible confounding risk factors or
stratified into smaller exposure groupings (e.g., low, medium, and high exposure).
Another approach to sample size calculations that might be informative was to fix the minimum
detectable effects to the effect sizes observed in previous studies for PFAS serum levels similar to those
observed in the Pease population, select the type 1 and type 2 error rates, and allow the sample size to
25

“float” instead of the minimum detectable effect. However, this approach is problematic because there
are few studies of PFAS exposures and the childhood outcomes being considered for the Pease children
study. In some instances, studies evaluating similar PFAS serum levels obtained very different effect
sizes for the same outcome. In other instances, a study with a lower PFAS serum level obtained a higher
effect size for an outcome than a study with a higher PFAS serum level. Moreover, there are no studies
of children exposed to PFAS drinking water contamination as a result of AFFF use. Therefore, there is
much uncertainty about the effect size for each health-related endpoint that would be expected for PFAS
serum levels observed among the Pease children.
With these caveats, the following sample size per stratum calculations use the findings from studies of
PFAS-exposed children. (Note: a sample size of 500 per stratum means that the study would need 500
exposed and 500 unexposed children. If the goal is to compare an outcome by exposure quartiles, then
each quartile would need 500 children. Also, a 2:1 ratio of exposed to unexposed requires a larger total
sample size than a 1:1 ratio of exposed to unexposed.) Table 6c provides a summary of the sample size
considerations for each health-related endpoint.
Lipids
Mean Total Cholesterol, LDL, HDL, triglycerides: In the Taiwan study of lipids (Zeng 2015), the
sample size of 225 children aged 12-15 was sufficient to detect total cholesterol and LDL differences of
11-12 mg/dL for PFOA serum levels similar to Pease. Table 6 indicates that with a sample size of 350
exposed and 175 unexposed, much lower mean differences in total cholesterol could be detected with
sufficient statistical power. However, the observed PFOA OR of 1.2 for hypercholesterolemia would
have required a sample size of over 1,700 per stratum with a type 1 error of 0.10 and 80% power (using
the prevalence of hypercholesterolemia in this study of 28.4%). Using a lower type 1 error and/or higher
statistical power would require even larger sample sizes to detect an OR of 1.2 for hypercholesterolemia.
The serum levels of PFOA and PFOS among the children at Pease would put them in the first quartile
(i.e., the reference level) if they had been in the C8 study (Frisbee 2010). In the lower PFOA and PFOS
quartiles, the ORs for hypercholesterolemia were between 1.2 and 1.3, requiring sample sizes of 800 –
1660 per stratum with type 1 error of 0.10 and 80% power (using the prevalence of
hypercholesterolemia in this study of 34.2%). The strongest findings in this study for total cholesterol
were observed for the top quintile of PFOS serum levels. When the top quintile PFOS serum level was
compared with the reference level, the mean difference in total cholesterol was 8.5 mg/dL and the OR
for hypercholesterolemia was 1.6. Both of these findings are within the range that could be detected with
sufficient statistical power in a Pease study with 350 exposed and 175 unexposed children. However, the
top quintile for PFOS in the C8 study contained serum levels several times higher than serum levels in
the top quintile of the Pease children.
A study using NHANES data for 1999–2008 [Geiger 2014] observed a mean difference in total
cholesterol of 4.7 mg/dL for the 2nd tertile serum levels of PFOA compared with the reference level. The
2nd tertile serum levels of PFOA in this study correspond to the PFOA serum levels among children at
Pease. To calculate a sample size to detect this mean difference, a standard deviation of 28 mg/dL
(similar to the standard deviations for total cholesterol in the Taiwan and C8 study) was used. With type
1 error of 0.10 and 80% power, the sample size required to detect a mean difference of 4.7 mg/dL would
be 439 per stratum (or with an exposed to unexposed ratio of 2, as suggested for the Pease children
study, 660 exposed and 330 unexposed would be required). In the NHANES study, the 2nd tertile PFOS
serum levels corresponded to the PFOS serum levels among Pease children. The mean difference in total
26

cholesterol for this tertile was 3.4 mg/dL, which would require 630 per stratum with type 1 error of 0.10
and 80% power.
In the NHANES study, the ORs for hypercholesterolemia corresponding to serum PFOA and PFOS
levels among children at Pease were 1.49 and 1.35, respectively. To detect an OR of 1.49 with type 1
error of 0.10 and 80% power would require 358 per stratum (or with an exposed to unexposed ratio of 2,
540 exposed and 270 unexposed).

Kidney function and uric acid
In a study of adolescents (aged 12–19 years) and kidney function using NHANES data for 2003–2010
[Kataria 2015], the top quartile for serum PFOA would correspond to the top quartile for serum PFOA
among the Pease children. The mean difference in the estimated glomerular filtration (eGFR) for the top
quartile of PFOA compared with the 1st quartile reference level was -6.6 mL/min/1.73 m2, which would
be in the range detectable, with sufficient statistical power, by the Pease study sample size of 350
exposed and 175 unexposed children.
In this study, the serum uric acid mean difference of 0.21 mg/dL was observed, comparing the top
quartile PFOA to the reference level. To detect this difference with a type 1 error of 0.10 and 80%
power would require a sample size larger than that projected for the Pease children study, i.e., 398 per
stratum (or for an exposed to unexposed ratio of two, 596 exposed and 298 unexposed children).
The serum PFOS levels in the 3rd quartile of the NHANES study would correspond to the top quartile
for serum PFOS among the Pease children. The mean difference in eGFR for the 3rd quartile PFOS level
compared to the reference level was -7.2 mL/min/1.73 m2, which would be in the range detectable with
sufficient statistical power by the Pease study sample size of 350 exposed and 175 unexposed children.
However, the mean difference in uric acid was 0.05 mg/dL which would require a sample size of more
than 5,000 per stratum.
In a Taiwan study of uric acid [Qin 2016], the sample size of 225 children aged 12–15 years was
sufficient to obtain a statistically significant OR for hyperuricemia of 1.65 for PFHxS at serum levels
much lower than among the Pease children. For PFOA, the OR for hyperuricemia was 2.2 at serum
levels much lower than observed among the Pease children. A sample size of 350 exposed and 175
unexposed children would be sufficient to detect this OR with sufficient statistical power.

Attention Deficit/Hyperactivity Disorder (ADHD) and other neurobehavioral endpoints
In a C8 study of ADHD (Stein 2011), the first quartile or reference level for PFOA and PFOS would
correspond to the serum PFOA and PFOS levels among the children at Pease. For PFHxS, the serum
levels among the children at Pease would correspond to the 3rd quartile level in the C8 study. For the 3rd
quartile of PFHxS, the OR for ADHD was 1.43, and with current medications, the OR was 1.55. The
prevalence of ADHD was 12.4%, and with current medications, 5.1%. To detect an OR of 1.43 with a
prevalence of 12.4 %, the required sample size for a type 1 error of 0.10 and 80% power would be 829
per stratum. To detect an OR of 1.55 with a prevalence of 5.1%, the required sample size for a type 1
error of 0.10 and 80% power would be 1,179 per stratum.
27

In a study that used NHANES data for 1999–2004 [Hoffman 2010], the serum PFHxS levels were about
half the levels among the children at Pease. For serum levels corresponding to the top quintile level
among the Pease children, the OR for ADHD was 1.67 (using the regression coefficient in the logistic
model). To detect this OR, a sample size of 540 per stratum would be required for type 1 error of 0.10
and 80% power. For PFOA, the serum levels corresponding to the top quintile level among the children
at Pease in the NHANES population would have an OR of 1.82 for ADHD. For this OR, the required
sample size would be 390 per stratum (or 596 exposed and 298 unexposed children) for a type 1 error of
0.10 and 80% power.
For neurobehavioral outcomes other than ADHD, some of the neurobehavioral outcome studies (e.g.,
Stein [2013, 2014b]; Wang [2015], Lien [2016]) were also in the range of the minimum sample size
suggested for the Pease children study. IQ differences in the range of 3 to 4 points could be detected
with reasonable statistical power with a sample size of 350 exposed and 175 unexposed children.
One study [Liew 2015] evaluated autism spectrum disorder and obtained an OR of 1.3 for serum
PFHxS. With a prevalence of about 1.5%, a sample size of several thousand children would be
necessary to detect this OR. To detect an OR of 2.0 with sufficient statistical power would require
sample sizes of over 1,600 exposed and 1,600 unexposed.

Sex hormones and delayed puberty
In the C8 study of sex hormones [Lopez-Espinosa 2016], the serum levels of PFOA, PFOS, and PFHxS
were considerably higher than among the children at Pease. For PFOS, the natural log estradiol percent
difference in boys of -4% (per interquartile range of the natural log of PFOS) would require at least
1,154 per stratum for type 1 error of 0.10 and 80% power. The strongest finding in this study was the
decrease in testosterone among girls associated with PFOS. The natural log testosterone percent
difference in girls was -6.6% per interquartile range of the natural log of PFOS. To detect a percent
difference this large with type 1 error of 0.10 and 80% power would require at least 290 per stratum, or
434 exposed and 217 unexposed children.
There was insufficient information to make sample size calculations for the endpoint, delayed puberty.
The C8 study that evaluated this endpoint in included thousands of boys and girls [Lopez-Espinosa
2011].

Growth hormone
In the C8 study that evaluated sex hormones, insulin-like growth factor-1 (IGF-1) was also evaluated
[Lopez-Espinosa 2016]. The difference in the natural log IGF-1 among boys and girls was -2.5% and 2.1% per interquartile range of the natural log of PFHxS, respectively. To detect these differences with
sufficient statistical power, a sample size of 350 exposed and 175 unexposed children would be
sufficient.

28

Thyroid disease and function
A C8 study [Lopez-Espinosa 2012] evaluated thyroid disease among children. The prevalence of
participant-reported thyroid disease among children in this study was very low, about 0.6% and an OR
of 1.44 was obtained for PFOA serum levels considerably higher than those in the Pease population. To
detect this OR with 80% statistical power would require a sample size of over 10,000 exposed children.
In the C8 study of thyroid function [Lopez-Espinosa 2012], the largest percent difference for natural log
TSH was 3.1%, and 2.3% for TT4. These percent changes were for PFOA and PFOS serum levels
considerably higher than the serum levels among the children at Pease. To detect a 2.3% change in TT4
would require a sample size of at least 850 per stratum (type 1 error = 0.10 and 80% power). To detect a
3.1% change in natural log TSH would require a sample size of at least 8,545 per stratum (type 1 error =
0.10 and 80% power).
In the Taiwan study of thyroid function [Lin 2013], the sample size for those aged 12–19 years was 212.
The geometric means for serum PFOA and PFOS were lower than the geometric mean serum levels
among the children at Pease. For males and females, the natural log TSH declined by 0.5 mIU/L and
0.35 mIU/L respectively, for the >90th percentile serum PFOA compared with the reference level. To
detect either of these differences with sufficient statistical power, a sample size of 350 exposed and 175
unexposed children would be sufficient.
Immune function and diseases related to immune function
For immune function, one study [Grandjean 2012] had a similar sample size (N = 532) as the minimum
proposed for the Pease children study (i.e., 350 exposed and 175 unexposed children), and two studies
had somewhat larger sample sizes that might be achievable at Pease (Stein [2016a], N = 640; and Buser
[2016], N = 637). The data reported in these studies were insufficient to conduct sample size
calculations.
For asthma, the ORs observed in the NHANES studies [Humblet 2014, Stein 2016a] were in the range
of 1.2 – 1.3 and would require much larger sample sizes than can be recruited at Pease to achieve
sufficient statistical power. However, a Taiwan study [Dong 2013] obtained ORs for asthma between
3.8 and 4.0 for PFHxS and PFOA serum levels lower than those observed in the Pease children
population. A sample size of 350 exposed and 175 unexposed would be sufficient to detect these ORs
with sufficient statistical power.
Only one study [Stein 2016a] evaluated rhinitis and observed an OR of 1.35 for serum PFOA. To detect
an OR this low with sufficient statistical power would require a sample size larger than could be
recruited from the Pease population. However, with sufficient statistical power, ORs in the range of 1.5
– 1.6 could be detected in a study of the Pease population with a sample size of 500 exposed and 250
unexposed children. These ORs would fall within the 95% CI for the finding in this study.

Other health-related endpoints
A NHANES study [Geiger 2014b] evaluated PFOS and PFOA serum levels and hypertension and
obtained ORs < 1.0. Since there is no evidence so far of an association between PFAS serum levels and
hypertension in children, this endpoint is not considered further.
29

A study conducted in the Faroes [Karlsen 2016] evaluated serum levels of PFOA, PFOS and PFHxS and
overweight/obesity in children. At age 5 years, the ORs for overweight/obesity and the third tertile
serum levels of PFOA, PFOS and PFHxS were 1.88, 0.94, and 1.22. The serum levels of the PFAS
chemicals were considerably lower than at Pease. An OR of 1.62 could be detected with 80% statistical
power with a sample size of 350 exposed and 175 unexposed children. No study has been conducted of
PFAS and liver function or fatty liver disease biomarkers in children.

Childhood cancers
For childhood cancers such as leukemias, the incidence and prevalence is very low, requiring large
sample sizes. For example, the probability of getting a leukemia at ≤15 years is 0.08% or 8 per 10,000.
For ages ≤20 years, the probability is 0.09% or 9 per 10,000. At ages ≤14 years, the incidence rate for
leukemias is 5.5 per 100,000 person-years. A study that attempted to evaluate leukemias or other
childhood cancers would have to be multi-site or national.

H. Conclusion
Very little is known about the health effects from exposure to PFHxS, a PFAS that was considerably
elevated in the serum of children tested at Pease. More information is available on the health effects of
PFOS exposure, which was also elevated in the serum of children at Pease. However, there are still
major data gaps and inconsistencies in the findings concerning the health effects of PFOS exposure,
particularly effects on immune, thyroid and kidney function, neurobehavioral endpoints, sex hormones,
and age at puberty. Based on sample size calculations, a study of children at Pease could have sufficient
statistical power to evaluate several health-related endpoints. The study could also meet the criteria of
public health significance and scientific importance, and could address some of the health concerns
voiced by the Pease CAP and the previous CAB.
The study population can be enumerated and selection bias can be minimized if recruitment is carefully
done to avoid selection bias (i.e., selection that is associated with exposure and disease status). A sample
of Portsmouth public school students would be an appropriate comparison group for the Pease children.
There is a complete exposure pathway and a well-defined exposed population. The health-related
endpoints under consideration have been evaluated in at least one epidemiological study of PFAS
exposures to children, and these endpoints can be measured accurately. Information on potential
confounding factors can be obtained via questionnaire. The issue of reverse causation and confounding
from the use of measured serum PFAS levels can be avoided by predicting serum levels using PBPK
modeling. Therefore, a children’s study at Pease could provide meaningful and credible results.
A key issue is whether a study limited to the children exposed at the Pease Tradeport would have
sufficient statistical power and precision for some of the endpoints under consideration. A minimum
sample size of 350 exposed Pease children and 175 unexposed children from the Portsmouth area would
be sufficient for several outcomes of interest. For example, Table 6 indicates that a sample size of 350
exposed and 175 unexposed children is sufficient to detect effects of reasonable size for most of the
endpoints listed in the table. In addition, some of the immune and neurobehavioral studies that had
sufficient statistical power to obtain effect estimates that achieved statistical significance had sample
sizes within the range suggested as a minimum for the Pease children study.
30

When the effect sizes seen in previous PFAS studies are considered, the suggested minimum sample size
for the Pease children study could be sufficient for several endpoints, such as mean differences in lipids,
eGFR, and IGF-1. For other outcomes, such as uric acid mean difference, the sex hormones testosterone
and estradiol, and thyroid function, the sample size of a study limited to the Pease children population
might not be sufficient. Based on sample size calculations assuming 350 Pease children and 175
unexposed children, and assuming a simple comparison of exposed versus unexposed, health endpoints
are grouped below into three categories: 1) feasible to study, 2) possible to study (but might require a
larger sample size, e.g. 500 exposed and 250 unexposed), and 3) not feasible to study using the Pease
children population, unless additional populations exposed to PFAS-contaminated drinking water are
included in the study.
Health endpoints feasible to study in children at Pease
•
•
•
•

Mean difference in lipids (total cholesterol, LDL, HDL, triglycerides)
Mean difference in estimated glomerular filtration rate (eGFR), a measure of kidney function
Insulin-like growth factor-1 (IGF-1, a measure of growth hormone deficiency)
Overweight/Obesity

Health endpoints that might be possible to study in children at Pease (although a larger sample size
may be needed)
•
•
•
•
•
•
•
•
•

Mean difference in uric acid, a measure of kidney function
Elevated total cholesterol (hypercholesterolemia)
Elevated uric acid (hyperuricemia)
IQ/neurobehavioral
Thyroid function
Sex hormones
Asthma and atopic dermatitis (immune function)
Rhinitis (stuffy, runny nose)
Antibody responses to rubella, mumps, and diphtheria vaccines

Health endpoints not feasible to study using the Pease children population (to address these health
endpoints, populations from other sites with PFAS-contaminated drinking water would need to be
included, along with the Pease children population)
•
•
•
•
•
•

Attention deficit/hyperactivity disorder (ADHD)
Autism spectrum disorder
Delayed puberty
Thyroid disease
Fatty liver disease (C-18 biomarker)
Childhood cancers

To evaluate exposure response relationships, more than two strata are necessary. For some of the
candidate outcomes that are listed above as feasible to study or possible to study, the Pease children
31

population that can be recruited to participate will not be large enough to be split into exposure tertiles
or quartiles and still have sufficient statistical power for comparisons between each of the exposure
strata and a reference (unexposed) stratum.
Data analyses similar to those used in the C8 studies could be used. The methods include linear
regression of continuous (untransformed and natural log transformed) effect biomarkers on continuous
(untransformed and natural log transformed) PFAS serum levels and categorized PFAS serum levels,
and logistic regression of categorized effect biomarkers (e.g., hypercholesterolemia) or disease
prevalence on continuous (untransformed and natural log transformed) and categorical PFAS serum
levels. Restricted cubic splines for linear and logistic regression would be conducted to obtain flexible,
smoothed exposure-response curves. Measured PFAS serum levels would be evaluated. In addition, for
PFOS and PFOA (and possibly PFHxS, if an historical reconstruction modeling method becomes
available), estimated cumulative serum levels and estimated serum levels during critical vulnerability
periods (e.g., in utero exposure) could be evaluated.
In summary, a study limited to the Pease children population will likely have a sufficient sample size for
some of the candidate endpoints if the comparisons are simply between an exposed and unexposed
group. For some of the candidate endpoints, the sample size will be insufficient for even a simple
comparison between an exposed an unexposed group. Moreover, for many of the candidate endpoints,
the Pease children population will be of insufficient size to split into tertiles or quartiles to evaluate
exposure–response trends. Therefore, the inclusion of other sites with PFAS-contaminated drinking
water could be considered.

Feasibility of an epidemiological study of adults at the Pease Tradeport
Compared with NHANES data, PFHxS serum levels were elevated among adults who participated in the
2015 NH DHHS blood testing program. However, the literature review indicated that very few studies
have been conducted that evaluated PFHxS exposures and adult health effects. PFOS serum levels were
also elevated among the adults who participated in the NH DHHS blood testing program. Although
considerably more studies found evaluated PFOS exposures and adult health effects, there remain data
gaps and inconsistencies in the findings for liver function, kidney function and kidney disease, thyroid
disease and thyroid function, autoimmune diseases and immune function, osteoporosis/osteoarthritis,
endometriosis, and most cancers.
The public health significance of conducting a study of adults at Pease is that the study will be relevant
to other adult populations exposed to drinking water primarily contaminated with PFOS and PFHxS. A
study might also provide an opportunity for early medical intervention for certain health endpoints that
might be associated with PFAS exposure but not evaluated in routine physical exams, such as alterations
in thyroid, liver, and kidney function. A study of adults at Pease would have scientific importance
because it potentially could help to fill critical data gaps mentioned above concerning the health effects
of PFHxS and PFOS exposures. Based on animal studies, there is biological plausibility that PFAS
exposures could result in alterations of immune function and might have endocrine-disruptive properties
that could lead to alterations in thyroid function. However, few epidemiological studies have evaluated
PFHxS or PFOS exposures and these health endpoints. Finally, a study of adults at Pease has the
potential to provide meaningful and credible results (from the perspective of statistical power) for some
of the adverse outcomes of interest and would be responsive to community concerns. However, a study
32

limited to Pease adults would likely not be sufficiently large to associate exposures and some adverse
health outcomes (e.g., rare diseases such as specific cancers and specific chronic diseases).
A. Study hypotheses
Based on the literature review, the following hypotheses could be evaluated:
1. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher total cholesterol, lowdensity lipoprotein and triglycerides, and a higher prevalence of hypercholesterolemia.
2.

Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher prevalences of
coronary artery disease and hypertension.

3. Higher serum levels of PFOA, PFOS, or PFHxS are associated with differences in thyroid
stimulating hormone (TSH), TT4, and TT3, and a higher prevalence of hypothyroidism.
4. Higher serum levels of PFOA, PFOS, or PFHxS are associated with a higher level of uric acid
and a higher prevalence of hyperuricemia.
5. Higher serum levels of PFOA, PFOS, or PFHxS are associated with a lower estimated
glomerular filtration rate (eGFR) and a higher prevalence of kidney disease.
6. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher levels of liver
function biomarkers alanine transaminase (ALT), γ-glutamyltransferase (GGT), and direct
bilirubin, fatty liver disease biomarker cytokeratin-18 (CK-18), and a higher prevalence of liver
disease.
7. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher prevalences of
osteoarthritis and osteoporosis.
8. Higher serum levels of PFOA, PFOS, or PFHxS are associated with a higher prevalence of
endometriosis.
9. Higher serum levels of PFOA, PFOS, or PFHxS are associated with higher prevalences of
autoimmune diseases such as ulcerative colitis, rheumatoid arthritis, lupus, and multiple
sclerosis.
10. Higher serum levels of PFOA, PFOS, or PFHxS are associated with differences in serum levels
of IgA, IgE, IgG, IgM, C reactive protein (CRP), and antinuclear antibodies (ANA) and
alterations in specific cytokines.
A study of adults could include the collection of new blood samples to analyze PFAS serum levels. The
blood samples would also be analyzed for lipids and biomarkers of kidney, liver, thyroid, and immune
function. A questionnaire could be used to ascertain kidney disease, liver disease, cardiovascular
disease, hypertension, thyroid disease, autoimmune diseases, osteoporosis, osteoarthritis, pregnancyinduced hypertension, and endometriosis. Diseases ascertained via questionnaire would be confirmed
using medical records
33

B. Study population
According to the census, Portsmouth has 21,530 residents. About 67.5 % are adults aged 19–64 years
and another 15.9% are aged 65 years and older. This would mean that there are about 14,500 adults aged
18–64 years and about 3,400 aged 65 years and over. Although the actual number is unknown, some of
the workers at the Pease Tradeport live in New Hampshire towns other than Portsmouth or in the
bordering states of Massachusetts and Maine. The Pease Tradeport has a workforce of >9,000 persons.
In the 2015 blood testing program at Pease, 1,182 adults aged ≥18 years participated. Table 5 provides
PFAS serum data for the 1,190 participants in the 2015 Pease blood testing program who will be age
≥18 years in 2018.
C. Recruitment and consent
As stated previously, the NH DHHS Pease blood testing program’s consent form was strictly limited to
use of the participant’s blood sample for PFAS analyses only. The participant also consented to
complete a brief questionnaire at the time of blood draw concerning demographics, time at Pease
Tradeport, whether the worker was a firefighter, and consumption of drinking water. The consent form
did not mention the use of the blood sample for research purposes or the possibility of re-contacting the
participant for future studies. Therefore, the blood samples were not stored for future use, and ATSDR
cannot directly contact the participants in the Pease blood testing program to recruit them for a study.
Adults would need to sign a new consent form to participate.
The consent form would request a blood sample (about 35 mL or 1.2 ounces) from the adult for the
analyses of PFASs and the effect biomarkers. (Note: 35 mL was the maximum amount of blood
obtained from adults in the C8 studies.) The consent form could also ask the participant to complete a
questionnaire covering demographics, water consumption, dates and length of time working at Pease,
occupational history, lifestyle and health behaviors, diseases diagnosed by a physician or other health
provider, and provider contact information.
To recruit adult study participants, NH DHHS would have to contact those who participated in the 2015
blood testing program. Another approach is to work with the Tenants Association at Pease (TAP) and
the Pease International Development Authority (PDA) to contact firms on their mailing lists. TAP sends
newsletters and email notices to subscribing firms at the Tradeport. The PDA list, with mailing
addresses and email addresses of all firms at the Pease Tradeport, was provided to ATSDR to help
recruit members to the Pease CAP. This list could be used to conduct outreach to recruit adult study
participants. Other methods of outreach include contacting community groups and the media.

D. Biomarkers of effect
The following biomarkers would be analyzed in the serum:
•
•
•
•
•

Total cholesterol, low density lipoprotein, high density lipoprotein, total triglycerides
Thyroxine (T4), T3, thyroid stimulating hormone (TSH)
Uric acid, creatinine
Alanine transaminase (ALT), γ-glutamyltransferase (GGT), direct bilirubin, and cytokeratin-18
(CK-18)
Immunoglobulin G (IgG), IgA, IgE and IgM; C reactive protein, and antinuclear antibodies
(ANA), and alterations in specific cytokines.
34

E. Exposure assessment
Exposure assessment could be based on the serum PFAS levels obtained in the study supplemented by
the serum PFAS levels for those who participated in the 2015 NH DHHS Pease blood testing program.
Using historical estimates of the PFAS contaminant levels in the drinking water at the Pease Tradeport
(based on water modeling methods), PBPK modeling can be used to estimate historical serum levels of
PFOA and PFOS, combining information from the questionnaire on water consumption and dates and
length of time employed at Pease Tradeport, and information on background PFAS serum levels from
NHANES and from a comparison group unexposed to PFAS-contaminated drinking water or
occupationally exposed to PFAS or AFFF. Serum levels from the 2015 NH DHHS Pease blood testing
program and serum levels obtained for this study would be used to calibrate the PBPK models. If
feasible, historical estimates of serum PFHxS can be based on historical estimates for serum PFOA and
PFOS, because serum levels of PFHxS and PFOS were highly correlated among the Pease adults who
participated in the 2015 blood testing program (Pearson correlation coefficient = 0.73).

F. Sample size considerations
A key problem for an adult study at Pease will be identifying an appropriate comparison population of
workers from the Portsmouth area with similar occupations as the Pease workforce and who were not
exposed to PFAS-contaminated drinking water or occupationally exposed to PFAS or AFFF. Another
key problem will be recruiting a sufficient number of participants to achieve reasonable statistical power
and precision of effect estimates.
Studies conducted of the adult C8 population included tens of thousands of participants. For example,
studies of thyroid disease [Winquist 2014a], cardiovascular disease and lipids [Winquist 2014b], kidney
disease [Dhingra 2016], and liver function [Darrow 2016] included 28,541 community members and
3,713 workers at the DuPont plant. Smaller studies using NHANES data (e.g., Wen [2013], Webster
[2016], Shankar [2011], Gleason [2015], and Lin [2010]) had sample sizes of 1,181–4,333 adults.
Table 7a indicates the minimum detectable effects for a study that included 1,500 participants per
stratum. For a simple comparison between exposed and unexposed, this would require a total of 3,000
participants, i.e., 1,500 exposed and 1,500 unexposed. If the study population were divided into quartiles
of PFAS serum levels, with the first quartile being the reference exposure level, then this would result in
a total sample size of 6,000 persons (i.e., 4,500 exposed and 1,500 unexposed). Four combinations of
type 1 error (α error) and type 2 error (β error) are used in the table. A type 1 error of 0.05 corresponds
to a two-tailed hypothesis test using a p-value cutoff of 0.05 to determine statistical significance, or
using a 95% confidence interval. A type 1 error of 0.10 corresponds to a one-tail hypothesis test using a
p-value cutoff of 0.05 to determine statistical significance, or using a 90% confidence interval. A type 2
errors of 0.05, 0.10, and 0.20 correspond to statistical power of 95%, 90% and 80%, respectively.
Another possible approach to sample size calculations that might be informative would be to fix the
minimum detectable effects to the effect sizes observed in previous studies for similar levels of
exposure, select the type 1 and type 2 error rates, and allow the sample size to “float” instead of the
minimum detectable effect. However, this approach is problematic because there are few studies of
PFAS exposures and the adult outcomes being considered for the Pease adult study. In some instances,
studies evaluating similar PFAS serum levels obtained very different effect sizes for the same outcome.
In other instances, a study with a lower PFAS serum level obtained a higher effect size for an outcome
35

than a study with a higher PFAS serum level. Moreover, there are no studies of adults exposed to PFAS
drinking water contamination as a result of AFFF use. Therefore, there is much uncertainty about the
effect size for each health-related endpoint that would be expected for PFAS serum levels observed
among the Pease adults. With these caveats, the following sample size per stratum calculations use the
findings from studies of PFAS-exposed adults. Table 7b provides a summary of the sample size
considerations for each health-related endpoint.

Lipids
In the lipid study conducted of the C8 adult population [Steenland 2009], PFOS serum levels
corresponding to the PFOS serum levels among adults who participated in the Pease blood testing
program would result in a 3–4 mg/dL change in total cholesterol and in LDL. Table 7a indicates that
detecting a difference of about 4 mg/dL in total cholesterol would require a sample size of about 1,500
per stratum. To detect a difference of 3 mg/dL would require a larger sample size. For LDL, a sample
size of 1,500 per stratum would be sufficient for mean differences in the 3–4 mg/dL range.
The predicted increase in total cholesterol at the highest decile for PFOA and PFOS in the C8 study was
11–12 mg/dL. To detect a difference of 11 mg/dL, a sample size in the range of 200–300 per stratum
would probably be sufficient. However, the highest decile for PFOA and PFOS in the C8 population is
considerably higher than the serum levels observed for the adult participants in the Pease 2015 blood
testing.
In a C8 study [Steenland 2009] and a Canadian study [Fisher 2013], ORs in the range of 1.35 – 1.6 were
observed for hypercholesterolemia. Although PFAS serum levels were higher in the C8 population than
the Pease population, the PFAS serum levels in the Canadian study were lower than in the Pease
population. Table 7a indicates that ORs in this range for hypercholesterolemia can be detected with
sufficient statistical power with a sample size of 1,500 per stratum.
Kidney disease/function, and uric acid
In the C8 study of chronic kidney disease [Dhingra 2016], the highest hazard ratio (HR) was observed
for the lowest quintile of exposure (compared with the reference level) and was equal to 1.36. To detect
this HR, given the low prevalence of the disease (approximately 1.4%). would require a sample size of
at least 8,600 per stratum.
In the C8 study of uric acid [Steenland 2010], serum PFOS levels that correspond to those observed
among the adult participants in the Pease blood testing program resulted in a difference of 0.14 mg/dL.
To detect this difference would require a sample size in the range of 1,600–2,100 per stratum.
The largest differences in uric acid observed in this study was 0.28 mg/dL for PFOA serum levels
≥188.7 ng/mL and 0.22 mg/dL for PFOS serum levels ≥40.5 ng/mL. These serum levels are
considerably higher than those observed for the adults at Pease. Based on sample size calculations, a uric
acid difference of 0.28 mg/dL could be detected with reasonable statistical power and a sample size in
the range of 500–600 per stratum. Table 7a indicates that much lower differences in uric acid could be
detected with reasonable statistical power using a sample size of 1,500 per stratum.

36

In the C8 study, the OR for hyperuricemia for PFOA serum levels similar to those at Pease equaled 1.02.
For the top quintile of serum PFOA in the C8 population, the OR was 1.47. Based on sample size
calculations, a sample size in the range of 450–600 would be sufficient to detect an OR of 1.47 with
reasonable statistical power. However, the top quintile serum PFOA level in the C8 study was
considerably higher than observed in the Pease population.
In a study using NHANES data [Shankar 2011], a change in uric acid of 0.40 mg/dL was observed for
serum PFOA levels similar to those observed for Pease. Based on sample size calculations, this
difference could be detected with reasonable statistical power using a sample size of about 300 per
stratum. For hyperuricemia, an OR of 1.90 was observed for serum PFOA levels similar to Pease. Based
on sample size calculations, an OR of 1.90 can be detected with reasonable statistical power using a
sample size of about 240 per stratum.
Liver function
For liver function, to detect the very subtle changes observed in the C8 studies [Gallo 2012; Darrow
2016] would require a sample size as large as the C8 study itself. The same is true for liver disease. In
the Darrow 2016 study, the highest OR observed was 1.19 for the 2nd quintile of serum PFOA. The 2nd
quintile of serum PFOA in the C8 study is higher than the serum levels at Pease. To detect an OR of
1.19 would require a sample size of at least 20,000 per stratum.
A study using NHANES data [Gleason 2015] was able to detect associations with uric acid and liver
function biomarkers at serum PFAS levels similar to those observed at Pease and with a total sample
size of 4,333 persons. This study evaluated quartiles of serum PFAS, so each stratum had a sample size
of about 1,083 persons. Another study that used NHANES data [Lin 2010] also was able to detect
associations with liver function biomarkers with a total sample size of 2,216 persons. This study also
evaluated quartiles, so each stratum had a sample size of about 554 persons.
Cardiovascular disease
The C8 study that evaluated coronary artery disease did not find an elevation in risk [Winquist 2014b].
However, a study that used NHANES data [Shankar 2012] obtained an OR of 2.01 for cardiovascular
disease for the 4th quartile PFOA serum levels. These PFOA serum levels, ≥6 ng/mL, would correspond
to the 5th quintile of PFOA serum levels among Pease adults. The prevalence of cardiovascular disease
in this study was 13%. To detect an OR of 2.01, a sample size of about 250/stratum would probably be
sufficient.
Hypertension
One study evaluated hypertension in a community population and observed an OR <1.0 [Winquist
2014b]. The prevalence of hypertension in this study was about 38%. With a sample size of 1,500 per
stratum and a prevalence of 38%, ORs between 1.21 and 1.31 could be detected with sufficient
statistical power.
Thyroid disease/function
For thyroid disease, the C8 study evaluated self-reported disease and self-reported disease that was
confirmed by medical records [Winquist 2014a]. For serum PFOA levels similar to those at Pease, the
37

hazard ratios were in the range of 1.2–1.3. For all self-reported thyroid disease (prevalence = 11.3%), a
sample size of about 2,100 per stratum would probably be sufficient to detect a hazard ratio of 1.3. The
prevalence for confirmed disease was 6.5%, so that a sample size of about 3,500 per stratum would
probably be necessary to detect an HR of 1.3.
A study that used NHANES data evaluated thyroid disease [Melzer 2010]. For confirmed thyroid
disease (prevalence = 2.4% in this study), the ORs were slightly above 1.1 for PFOS and PFOA serum
levels similar to those at Pease. To detect this OR would require a sample size equivalent to the C8
population. The highest OR observed was 1.89 among men in the top quartile of PFOS and PFOA. To
detect this odds ratio, a sample size of about 1,400 per stratum would probably be sufficient.
The C8 study that evaluated thyroid function biomarkers [Knox 2011] observed very subtle changes that
would require a study of equivalent size (52,296) to detect associations with sufficient statistical power.
On the other hand, a study that used NHANES data [Wen 2013] to evaluate thyroid function observed
larger changes that could be detected with a total sample size of <1,200 (or <300 per quartile stratum).

Immune function and autoimmune diseases
Only one published study [Stein 2016b] evaluated serum immune biomarkers at baseline (i.e., crosssectionally) and PFAS serum levels. The study evaluated de-identified archived blood samples from 75
adults aged 21-49. Given the very small sample size, this should be considered a pilot study. The PFHxS
serum levels in this study were considerably lower than in the Pease adult population and a few positive
findings were observed but the confidence intervals for these findings were extremely wide indicating
little precision and a high degree of uncertainty in the effect estimates. Given the strong animal
evidence of effects on the immune system from PFAS exposures [NTP 2016], a cross-sectional
evaluation of PFAS serum levels and immune biomarkers in a Pease adult study could provide important
information on the effects of PFAS exposures on immune function in humans.
The prevalences of ulcerative colitis, rheumatoid arthritis, lupus, and multiple sclerosis in a C8 study
[Steenland 2013] were ≤ 1.2%. As indicated in Table 7a, ORs ≤ 2.0 cannot be detected with sufficient
statistical power for these endpoints with a sample size of 1,500 per stratum. For lupus and multiple
sclerosis, ORs <3.5 cannot be detected with sufficient statistical power with a sample size of 1,500 per
stratum.

Osteoarthritis and Osteoporosis
Two studies evaluated osteoarthritis. In a C8 study [Innes 2011], an OR of about 1.4 was observed for
serum PFOA levels considerably higher than those at Pease. However, in an NHANES study [Uhl
2013], an OR of 1.5 was observed for serum PFOA levels similar to those at Pease. Table 7a indicates
that ORs in the range of 1.4 – 1.6 can be detected with sufficient statistical power with a sample size of
1,500 per stratum.
An NHANES study evaluated osteoporosis in women [Khalil 2016] and obtained an OR > 10 for serum
PFHxS levels lower than those at Pease. With 750 women per stratum, an OR of 1.58 can be detected
with sufficient statistical power.
38

Endometriosis
An NHANES study [Campbell 2016] obtained ORs of 1.47 and 2.86 for serum PFHxS and PFOA,
respectively. The serum levels for these two PFAS were similar to those in the Pease population. Table
7a indicates that with a sample size of 750 per stratum, ORs in the range of 1.55 – 1.85 can be detected
with sufficient statistical power.
Pregnancy-induced hypertension
Several C8 studies evaluated pregnancy-induced hypertension. One study observed an OR of 1.6 for
serum PFOS. However, the PFOS serum levels in the C8 study were higher than those at Pease. Table
7a indicates that ORs in the range of 1.6 – 1.9 can be detected with sufficient statistical power for a
sample size of 750 pregnancies per stratum.
Cancer incidence
For kidney cancer, Table 7a indicates that ORs <3.8 cannot be detected with sufficient statistical power
with a sample size of 1,500 per stratum. Even for a cancer with a much higher prevalence than kidney
cancer, e.g., prostate cancer, ORs < 2.0 cannot be detected with sufficient statistical power with a sample
size of 750 men per stratum.

F. Conclusion
A sample size of about 1,500 per stratum (or a total sample size of 6,000 if quartiles are evaluated)
would have sufficient statistical power to detect several of the health-related endpoints, as indicated by
Tables 7a and 7b. For some endpoints, such as mean difference in uric acid, hyperuricemia, and
cardiovascular disease, smaller sample sizes of about 500 per stratum might be sufficient. For other
endpoints, such as ulcerative colitis, rheumatoid arthritis, and chronic kidney and liver disease, sample
sizes larger than 1,500 per stratum would be necessary. Based on the sample size calculations that
assume a sample size of 1,500 adults employed at the Pease Tradeport and 1,500 adults from the
Portsmouth area who were never employed at the Pease Tradeport, and assuming a simple comparison
of exposed versus unexposed, health endpoints are grouped below into three categories: 1) feasible to
study, 2) possible to study (but might require a larger sample size from the Pease population), and 3) not
feasible to study using the Pease adult population unless additional populations exposed to PFAScontaminated drinking water are included in the study.
Health endpoints feasible to study in adults at Pease
•
•
•
•
•
•
•
•

Mean difference in lipids (total cholesterol, LDL, HDL, triglycerides)
Elevated total cholesterol (hypercholesterolemia)
Mean difference in uric acid, a measure of kidney function
Elevated uric acid (hyperuricemia)
Thyroid disease (unconfirmed)
Cardiovascular disease
Hypertension
Osteoarthritis and osteoporosis
39

•

Mean differences in serum immunoglobin (IgA, IgE, IgG, IgM), and C-reactive protein (an indicator
of inflammation); increase in antinuclear antibodies (an indicator of autoimmune reaction);
alterations in specific cytokines

Health endpoints that may be possible to study in adults at Pease (although a larger sample size may
be needed)
•
•
•
•
•

Liver function and CK-18 (fatty liver disease biomarker)
Thyroid disease (confirmed)
Thyroid function
Endometriosis
Pregnancy-induced hypertension

Health endpoints not feasible to study using the Pease adult population (in order to address these
health endpoints, populations from other sites with PFAS-contaminated drinking water would need to be
included along with the Pease adult population)
•
•
•
•
•
•
•

Liver disease
Kidney disease
Ulcerative colitis
Rheumatoid arthritis
Lupus
Multiple sclerosis
Kidney cancer (and other adult cancers)

To evaluate exposure–response trends, the study participants would need to be split into tertiles or
quartiles based on their serum PFAS levels. For some of the candidate health endpoints that are listed
above as feasible to study or possible to study, the Pease adult population that can be recruited to
participate will not be large enough to be split into exposure tertiles or quartiles and still have sufficient
statistical power for comparisons between each of the exposure strata and a reference (unexposed)
stratum. For example, if the study population is to be divided into quartiles, and assuming that a sample
size of 1,500 per stratum would be sufficient for many of the endpoints of interest, then it would be
necessary to recruit 4,500 adults (aged ≥18 years at the start of the study) from the Pease workforce and
a representative group (i.e., employed in similar occupations as the Pease workforce) of 1,500 adults
from the Portsmouth area who were not exposed at Pease.
Data analyses similar to those used in the C8 studies would be used. The methods include linear
regression of continuous (untransformed and natural log-transformed) effect biomarkers on continuous
(untransformed and natural log-transformed) PFAS serum levels and categorized PFAS serum levels;
and logistic regression of categorized effect biomarkers (e.g., hypercholesterolemia) or disease
prevalence on continuous (untransformed and natural log-transformed) and categorical PFAS serum
levels. Restricted cubic splines for linear and logistic regression would be conducted to obtain flexible,
smoothed exposure-response curves. Measured PFAS serum levels would be evaluated. In addition, for
PFOS and PFOA (and possibly PFHxS if an historical reconstruction modeling method becomes
available), estimated cumulative serum levels would be evaluated.
40

In summary, a study limited to the Pease adult population could likely have a sufficient sample size for
some of the candidate endpoints if the comparisons are simply between an exposed and unexposed
group. Recruitment of at least 1,500 adults from Pease should be feasible, given that the 2015 blood
testing program at Pease was able to recruit at least 1,182 adults aged >18 years who worked at Pease.
However, a study limited to the Pease adult population might not have a sufficient sample size to
evaluate exposure–response relationships. Moreover, a study limited to the Pease worker population
might not have sufficient variability in serum PFAS levels to evaluate exposure–response trends
effectively. Sufficient variability in PFAS serum levels might be achieved by including other
populations with residential exposures to PFAS-contaminated drinking water.

Feasibility of an epidemiological study of former military service and civilian
workers at the former Pease Air Force base
Drinking water contamination at a military base involves potential residential exposures to those living
and training at the base and potential exposures to those working at the base. At the former Pease Air
Force Base, starting in the 1970s, AFFF foam was used for fire training and to extinguish flammable
liquid fires. The PFAS contamination in the Haven well water supply likely occurred sometime during
the period from the start of AFFF usage and the closing of the base and would have resulted in
exposures to those living and working at the base.
To evaluate the incidence and mortality of specific cancers, a large population of adults would need to
be followed for a sufficient number of years to account for the long induction periods of most cancers
and to have sufficient statistical power. For example, the Camp Lejeune mortality study of U.S. Marines
and Navy personnel followed a cohort of 154,932 from 1979 to 2008 for a total of over 4 million personyears [Bove 2014]. To evaluate cancer incidence for the Camp Lejeune cohort, ATSDR will conduct
follow-up using state and federal cancer registries for the period 1996–2016 (1996 is the earliest date
that >90% of the state registries were in operation), for a total of over 3 million person-years. For the
civilian worker cohort at Camp Lejeune, 8,085 workers will be followed over the period 1996–2016 for
cancer incidence, for a total of 121,875 person-years. This is similar in size to a study of cancer
incidence among workers at a PFAS manufacturing plant [Raleigh 2014]. A recent study of firefighters
followed a pooled cohort of 29,993 from San Francisco, Chicago, or Philadelphia from 1985 through
2009, for a total of 403,152 person-years [Daniels 2014]. A C8 study of cancer incidence that relied on
self-reported cancers that were confirmed by medical records and cancer registry review included
32,254 who contributed over 1 million person-years of follow-up [Barry 2013].
In October 1989, 3,465 military personnel were assigned to Pease Air Force Base, accompanied by
4,746 dependents. The Air Force estimates that 537 civilian employees were employed on base at that
time [USAF 1990]. From 1970 to 1990, an average of 3,000 personnel and their families were assigned
to the base at any one time. Before 1970, the base supported a maximum of 5,000 personnel [USAF
1994]. One important consideration about including Pease service personnel and civilian workers in a
cancer incidence and mortality study is that drinking water at the base was also contaminated by TCE
from the Haven well during some of the years the base operated. Service personnel and civilian workers
stationed at the base before 1986 should not be included because of this contamination. Because the base
closed by 1991, the number of service personnel and civilian workers at Pease AFB that could be
included in a study would be insufficient to evaluate cancers with sufficient statistical power.
41

Because of the relatively small numbers of personnel assigned to Pease Air Force Base, we
conclude that it is not feasible to conduct a study of cancer incidence and mortality that is limited
to the Pease military service personnel and civilian worker cohorts stationed at the base from 1986
onward. For a study to be feasible, it would require a larger population size, for example, by including
service personnel and civilian workers from other military bases with PFAS-contaminated drinking
water as a result of the use of AFFF. Exposures to other drinking water contaminants, such as TCE,
other chlorinated organic chemicals, and benzene, must also be taken into account when considering
candidate military bases and defining the cohorts.
Cohorts of service personnel and civilian workers can be identified at military bases from personnel data
maintained at the Defense Manpower Data Center. Personnel data are available from 1971, although
information on military unit, which is needed to determine the base where the individual was stationed,
does not begin until the second quarter of 1975. For civilian workers, data are available starting in the
last quarter of 1972, with data missing for the first quarter of 1973. The data contain the location of the
workplace (codes for state, city, and ZIP code). The Defense Manpower Data Center data contain Social
Security number, name, date of birth, and sex to facilitate follow-up.
Military service personnel constitute a highly mobile population after their tours of duty are completed.
For a mortality study, this is not a problem, because the NDI is available to obtain information on causes
of death. However, there is no national cancer registry to ascertain cancer incidence. Therefore, a study
of military service personnel and civilian workers would require gaining the participation of all or most
of the state cancer registries and the Department of Veterans Affairs Central Cancer Registry (VACCR).
The Camp Lejeune Cancer Incidence Study is one model for such a study. This study is attempting to
recruit at least two-thirds of the state cancer registries and VACCR to cover >90% of the Camp Lejeune
and Camp Pendleton cohorts. The study will send the personal identifiers for each cohort member to
each registry for matching with the registry’s data. For any matches that occur, the registry will send to
ATSDR the cancer information that is linked to personal identifier (e.g., Social Security number or a
unique identification number linked to the Social Security number). This will allow assessment of
exposures and other covariates and cancers at the individual level.
The most appropriate military sites for inclusion would be those with water systems that are not complex
so that simple mixing models can be used to estimate PFAS-contaminant levels throughout the
distribution system. In addition, candidate sites should have information on the history of AFFF use at
the base including major incidents such as spills, fires, etc.

Other study designs and health-related endpoints
1. Adverse birth outcomes
To evaluate adverse birth outcomes such as SGA, preterm birth, and specific congenital malformations
with sufficient statistical power, several thousand births should be studied. For example, to detect an OR
of 1.5 for SGA (5th percentile) with 80% power would require 1,775 births per stratum. For SGA (10th
percentile) and preterm birth, with 80% power, an OR of 1.5 can be detected with a sample size of about
960–990 births per stratum. For rare birth defects, such as neural tube defects, to detect an OR of 2.5
with 80% power would require a sample size of about 22,000 births per stratum. For oral clefts, to detect
an OR of 2.0 would require about 15,000 births per stratum.
42

Birth weight, SGA and preterm birth can be evaluated using birth certificate data. For birth defects, a
population-based registry must be used to identify cases.
An adverse birth outcome study is not feasible at Pease because there were too few births to mothers
who worked at the Tradeport during their pregnancy. The most appropriate candidate populations for a
study of adverse birth outcomes would be one or more large municipalities with residential exposures to
PFAS-contaminated drinking water where a simple mixing model could be used to estimate contaminant
levels throughout the distribution system, i.e., a system that is not complex but instead has relatively
uniform contaminant levels throughout the distribution system.

2. Registry
Creating a registry of exposed children and adults at the Pease Tradeport involves following the health
status over a period of time and is similar to an epidemiological, longitudinal study of an exposed
cohort. The difference is that an epidemiological study would usually include a comparison, unexposed
cohort. A registry, like a longitudinal epidemiological study, can be resource-intensive. A decision
would also have to be made concerning the length of the follow-up. As in any longitudinal effort,
individuals will drop out over time, resulting in interpretation difficulties (e.g., selection bias resulting
from loss to follow-up). In any event, before a registry or longitudinal study can be contemplated, an
initial cross-sectional study must first be conducted, similar to the children’s study and adult study
discussed above.

3. Multi-site studies
The results of sample size calculations indicated that the exposed populations at the Pease Tradeport and
the former Pease Air Force Base were of insufficient size for some of the health-related endpoints of
interest to the community. Moreover, Pease CAP members have expressed interest in linking the Pease
communities with other communities that have been exposed to PFAS-contaminated drinking water. A
national database exists that can be used to identify other communities with PFAS-contaminated
drinking water. Data on PFAS contamination of public drinking water supplies are available for large
systems (serving >10,000 retail customers) and a small sample of small systems (n = 800 or 0.5% of a
total of 144,165 systems serving <10,000 retail customers) via the Third Unregulated Contaminant
Monitoring Rule (UCMR-3) database maintained by the EPA [US EPA 2016b].
UCMR-3 monitoring for PFAS is required at the entry point to the distribution system for each well and
at any interconnection that is in operation. Water utilities had to sample twice during a 12-month period
from 2013–2015 with sampling events occurring 5–7 months apart. The UCMR dataset contains
sampling data from January 2, 2013 through March 1, 2016. Table A1 in the Appendix lists the utilities
ranked by the maximum level of combined PFOS and PFHxS detected in the system. The highest level
was detected in the system serving the Mariana Islands. Among the U.S. water systems, the top 10
systems for combined PFOS and PFHxS were Artesian Water Company in Delaware; Security Water
System in Colorado Springs, CO; Horsham and Warminster systems in Pennsylvania; Oatman Water
Company in Arizona; Issaquah Water System in Washington; Hyannis Water System in Massachusetts;
Suffolk County Water Authority in New York; Warrington Township Water in Pennsylvania; and
United Water in Pennsylvania, which serves various municipalities.

43

Although the UCMR database can be used to identify potential sites for further consideration for health
studies, it has several limitations. First, most small systems are not included in the database. Second, the
data represent levels of contamination at the entry points to the distribution system of the water utility
(e.g., contaminant levels in a supply well) and generally do not represent the levels of contamination
reaching particular residences served by the utility. To estimate the population receiving contaminated
drinking water and the levels of PFAS in their drinking water, the UCMR data must be supplemented
with information on the configuration and operation of the utility’s system. For a system that mixes all
its sources of water before to entering the distribution system, a simple mixing model can be used to
estimate the contaminant levels in the drinking water serving the residences by taking into account the
contaminant levels in each source and the contribution of each source to the total supply. This is the
situation at the Pease Tradeport, where water from each of the supply wells is mixed at the treatment
plant before entering the distribution system. However, many utilities have more complex systems in
which each of the supply wells (or surface water sources) primarily serve particular areas of the
distribution system. For these systems, additional information is needed (for example, on the operation
of the supply wells, tank levels, and the water demand in each area of the distribution system), and
complex modeling methods must be used.

Conclusions
The ability of a study of the Pease population to provide useful information will depend to a great extent
on the success of recruiting sufficient number of study participants. The feasibility assessment
concluded that it is possible to evaluate some health-related endpoints if a sufficient number of children
and adults from the Pease population participate. Other health-related endpoints would require larger
numbers of exposed individuals and would require the inclusion of populations from other sites who
were exposed to PFAS-contaminated drinking water. The feasibility assessment concluded that a third
study design, a mortality and cancer incidence study of former military service and civilian worker
personnel, would not be feasible solely with the population at Pease.
The feasibility of successfully evaluating particular health-related endpoints (or effect biomarkers) could
change depending on final study design and goals.

44

References
Agency for Toxic Substances & Disease Registry (ATSDR). Public Health Assessment for Pease Air
Force Base, Portsmouth, Rockingham County, New Hampshire. Atlanta: US Department of Health and
Human Services; September 30, 1999.
https://www.atsdr.cdc.gov/HAC/pha/PHA.asp?docid=1198&pg=0
Agency for Toxic Substances and Disease Registry (ATSDR). Public Health Assessment: Willow Grove
Naval Air Station/Air Reserve Station (a.k.a. Naval Air Station Joint Reserve Base and Air Force
Reserve Station) Montgomery County, Pennsylvania. Atlanta: US Department of Health and Human
Services; 2002a.
Agency for Toxic Substances and Disease Registry (ATSDR). Public Health Assessment: Naval Air
Warfare Center [A/K/A Naval Air Development Center (8 Waste Areas)] Warminster Township, Bucks
County, Pennsylvania. Atlanta: US Department of Health and Human Services; 2002b.
Agency for Toxic Substances and Disease Registry (ATSDR). Letter Health Consultation:
Perfluorochemicals (Pfcs) Contamination in Public and Private Drinking Water Sources Near the
Former Naval Air Warfare Center Warminster, Bucks County, Pennsylvania. Atlanta: US Department of
Health and Human Services; 2016.
Alexander BH, Olsen GW, Burris JM, Mandel JH, Mandel JS. Mortality of employees of a
perfluorooctanesulphonyl fluoride manufacturing facility. Occup Environ Med 2003;60:722–729.
Alexander BH, Olsen GW. Bladder cancer in perfluorooctanesulfonyl fluoride manufacturing workers.
Ann Epidemiol 2007;17:471-478.
Andersen CS, Fei C, Gamborg M, Nohr EA, Sørensen TI, Olsen J. Prenatal exposures to perfluorinated
chemicals and anthropometric measures in infancy. Am J Epidemiol 2010;172(11):1230-7.
Anderson EL, Howe LD, Jones HE, Higgins JPT, Lawlor DA, Fraser A. The prevalence of nonalcoholic fatty liver disease in children and adolescents: A systematic review and meta-analysis. PLoS
One 2015;10:e0140908.
Apelberg BJ, Witter FR, Herbstman JB, Calafat AM, Halden RU, Needham LL, Goldman LR. Cord
serum concentrations of perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) in relation to
weight and size at birth. Environ Health Perspect 2007;115(11):1670-6.
Avanasi R1, Shin HM, Vieira VM, Savitz DA, Bartell SM. Impact of Exposure Uncertainty on the
Association between Perfluorooctanoate and Preeclampsia in the C8 Health Project Population. Environ
Health Perspect 2016;124(1):126-32.
Bach CC, Bech BH, Brix N, Nohr EA, Bonde JP, Henriksen TB. Perfluoroalkyl and polyfluoroalkyl
substances and human fetal growth: a systematic review. Crit Rev Toxicol 2015a;45(1):53-67.
Bach CC, Liew Z, Bech BH, Nohr EA, Fei C, Bonefeld-Jorgensen EC, Henriksen TB, Olsen J.
Perfluoroalkyl acids and time to pregnancy revisited: An update from the Danish National Birth Cohort.
Environ Health 2015b;14:59.
45

Bach CC, Vested A, Jørgensen KT, Bonde JP, Henriksen TB, Toft G. Perfluoroalkyl and
polyfluoroalkyl substances and measures of human fertility: a systematic review. Crit Rev Toxico.
2016a;46(9):735-55.
Bach CC, Bech BH, Nohr EA, Olsen J, Matthiesen NB, Bonefeld-Jørgensen EC, Bossi R, Henriksen
TB. Perfluoroalkyl Acids in Maternal Serum and Indices of Fetal Growth: The Aarhus Birth Cohort.
Environ Health Perspect 2016b;124(6):848-54.
Barry V, Winquist A, Steenland K. Perfluorooctanoic Acid (PFOA) Exposures and Incident Cancers
among Adults Living Near a Chemical Plant. Environ Health Perspect 2013;121:1313–1318.
Berg V, Nøst TH, Hansen S, Elverland A, Veyhe AS, Jorde R, Odland JØ, Sandanger TM. Assessing
the relationship between perfluoroalkyl substances, thyroid hormones and binding proteins in pregnant
women; a longitudinal mixed effects approach. Environ Int 2015;77:63-9.
Berg V, Nøst TH, Pettersen RD, Hansen S, Veyhe AS, Jorde R, Odland JØ, Sandanger TM. Persistent
Organic Pollutants and the Association with Maternal and Infant Thyroid Homeostasis: A Multipollutant
Assessment. Environ Health Perspect 2017;125:127-133.
Bonefeld-Jorgensen EC, Long M, Bossi R, Ayotte P, Asmund G, Krüger T, Ghisari M, Mulvad G, Kern
P, Nzulumiki P, Dewailly E. Perfluorinated compounds are related to breast cancer risk in Greenlandic
Inuit: a case control study. Environ Health 2011;10:88.
Bonefeld-Jorgensen EC, Long M, Fredslund SO, Bossi R, Olsen J. Breast cancer risk after exposure to
perfluorinated compounds in Danish women: a case-control study nested in the Danish National Birth
Cohort. Cancer Causes Control 2014;25:1439-48.
Bove FJ, Ruckart PZ, Maslia M, Larson TC. Evaluation of mortality among Marines and navy personnel
exposed to contaminated drinking water at USMC base Camp Lejeune: A retrospective cohort study.
Environ Health 2014a; 13:10.
Bove FJ, Ruckart PZ, Maslia M, Larson TC. Mortality study of civilian employees exposed to
contaminated drinking water at USMC base Camp Lejeune: A retrospective cohort study. Environ
Health 2014b;13:68.
Braun JM, Kalkbrenner AE, Just AC, Yolton K, Calafat AM, Sjödin A, Hauser R, Webster GM, Chen
A, Lanphear BP. Gestational exposure to endocrine-disrupting chemicals and reciprocal social,
repetitive, and stereotypic behaviors in 4- and 5-year-old children: the HOME study. Environ Health
Perspect 2014;122(5):513-20.
Buck Louis GM, Peterson CM, Chen Z, Hediger ML, Croughan MS, Sundaram R, Stanford JB,
Fujimoto VY, Varner MW, Giudice LC, Kennedy A, Sun L, Wu Q, Kannan K. Perfluorochemicals and
endometriosis: The ENDO Study. Epidemiology 2012;23:799-805.
Buck Louis GM, Sundaram R, Schisterman EF, Sweeney AM, Lynch CD, Gore-Langton RE, Maisog J,
Kim S, Chen Z, Barr DB. Persistent environmental pollutants and couple fecundity: the LIFE study.
Environ Health Perspect 2013;121(2):231-6.
46

C8 Science Panel (Tony Fletcher, Kyle Steenland, David Savitz). Status Report: PFOA and immune
biomarkers in adults exposed to PFOA in drinking water in the mid Ohio valley. March 2009.
Callan AC, Rotander A, Thompson K, Heyworth J, Mueller JF, Odland JØ, Hinwood AL. Maternal
exposure to perfluoroalkyl acids measured in whole blood and birth outcomes in offspring. Sci Total
Environ 2016;569-570:1107-13.
Campbell S, Raza M, Pollack AZ. Perfluoroalkyl substances and endometriosis in US women in
NHANES 2003-2006. Reprod Toxicol 2016;65:230-235.
Checkoway H, Pearce N, Kriebel D. Research Methods in Occupational Epidemiology, Second Edition.
Oxford U. Press 2004.
Chen MH, Ha EH, Wen TW, Su YN, Lien GW, Chen CY, Chen PC, Hsieh WS. Perfluorinated
compounds in umbilical cord blood and adverse birth outcomes. PLoS One 2012;7(8):e42474.
Christensen KY, Maisonet M, Rubin C, Holmes A, Calafat AM, Kato K, Flanders WD, Heron J,
McGeehin MA, Marcus M. Exposure to polyfluoroalkyl chemicals during pregnancy is not associated
with offspring age at menarche in a contemporary British cohort. Environ Int 2011;37:129-135.
Community Advisory Board – Haven Well, City of Portsmouth, NH. Final Report on PFC
Contamination in Pease Tradeport Wells. December 21, 2015.
http://cityofportsmouth.com/CAB%20Final%20Report%20December%2014%202015.pdf
Consonni D, Straif K, Symons JM, Tomenson JA, van Amelsvoort LG, Sleeuwenhoek A, Cherrie JW,
Bonetti P, Colombo I, Farrar DG, Bertazzi PA. Cancer risk among tetrafluoroethylene synthesis and
polymerization workers. Am J Epidemiol 2013;178:350-358.
Daniels RD, Kubale TL, Yiin JH, Dahm MM, Hales TR, Baris D, Zahm SH, Beaumont JJ, Waters KM,
Pinkerton LE. Mortality and cancer incidence in a pooled cohort of US firefighters from San Francisco,
Chicago and Philadelphia (1950–2009). Occup Environ Med 2014;71:388-397.
Darrow LA, Stein CR, Steenland K. Serum perfluorooctanoic acid and perfluorooctane sulfonate
concentrations in relation to birth outcomes in the Mid-Ohio Valley, 2005-2010. Environ Health
Perspect 2013;121(10):1207-13.
Darrow LA, Howards PP, Winquist A, Steenland K. PFOA and PFOS serum levels and miscarriage risk.
Epidemiology 2014;25(4):505-12.
Darrow LA, Groth AC, Winquist A, Shin HM, Bartell SM, Steenland K. Modeled perfluorooctanoic
acid (PFOA) exposure and liver function in a Mid-Ohio Valley Community. Environ Health Perspect
2016;124:1227-1233.
Dhingra R, Lally C, Darrow LA, Klein M, Winquist A, Steenland K. Perfluorooctanoic acid and chronic
kidney disease: Longitudinal analysis of a Mid-Ohio Valley community. Environ Res 2016a;145:85-92.

47

Dhingra R, Winquist A, Darrow LA, Klein M, Steenland K. A study of reverse causation: examining the
associations of perfluorooctanoic acid serum levels with two outcomes. Environ Health Perspect 2016b
Aug 16. [Epub ahead of print]
Domazet SL, Grøntved A, Timmermann AG, Nielsen F, Jensen TK. Longitudinal associations of
exposure to perfluoroalkylated substances in childhood and adolescence and indicators of adiposity and
glucose metabolism 6 and 12 years later: The European Youth Heart Study. Diabetes Care 2016;
39(10):1745-51.
Dong GH, Tung KY, Tsai CH, Liu MM, Wang D, Liu W, Jin YH, Hsieh WS, Lee YL, Chen PC. Serum
polyfluoroalkyl concentrations, asthma outcomes, and immunological markers in a case-control study of
Taiwanese children. Environ Health Perspect 2013; 121(4): 507-513, 513e501-508.
Emmett EA, Zhang H, Shofer FS, Freeman D, Rodway NV, Desai C, Shaw LM. Community Exposure
to Perfluorooctanoate: Relationships Between Serum Levels and Certain Health Parameters. J Occup
Environ Med 2006;48:771-9.
Fei C, McLaughlin JK, Tarone RE, Olsen J. Perfluorinated chemicals and fetal growth: a study within
the Danish National Birth Cohort. Environ Health Perspect 2007;115(11):1677-82.
Fei C, McLaughlin JK, Tarone RE, Olsen J. Fetal growth indicators and perfluorinated chemicals: a
study in the Danish National Birth Cohort. Am J Epidemiol 2008;168(1):66-72.
Fei C, McLaughlin JK, Lipworth L, Olsen J. Maternal levels of perfluorinated chemicals and
subfecundity. Hum Reprod 2009;24(5):1200-5.
Fisher M, Arbuckle TE, Wade M, Haines DA. Do perfluoroalkyl lsubstances affect metabolic function
and plasma lipids?—Analysis of the 2007–2009, Canadian Health Measures Survey (CHMS) Cycle1.
Environ Res 2013;121:95-103.
Fitz-Simon N, Fletcher T, Luster MI, Steenland K, Calafat AM, Kato K, Armstrong B. Reductions in
serum lipids with a 4-year decline in serum perfluorooctanoic acid and perfluorooctanesulfonic acid.
Epidemiology 2013;24:569-76.
Flanders WD, Lin L, Pirkle JL, Caudill SP. Assessing the direction of causality in cross-sectional
studies. Am J Epidemiol 1992;135:926-935.
Flanders WD, Klein M, Mirabelli MC. Conditions for valid estimation of causal effects on prevalence in
cross-sectional and other studies. Ann Epidemiol 2016;26:389-394.
Forns J, Iszatt N, White RA, Mandal S, Sabaredzovic A, Lamoree M, Thomsen C, Haug LS, Stigum H,
Eggesbø M. Perfluoroalkyl substances measured in breast milk and child neuropsychological
development in a Norwegian birth cohort study. Environ Int 2015;83:176-82.
Frisbee SJ, Brooks Jr AP, Maher A, et al. The C8 health project: design, methods, and participants.
Environ Health Perspect 2009;117:1873-1882.

48

Frisbee SJ, Shankar A, Knox SS, Steenland K, Savitz DA, Fletcher T, Ducatman AM. Perfluorooctanoic
acid, perfluorooctanesulfonate, and serum lipids in children and adolescents: Results from the C8 Health
Project. Arch Pediatr Adolesc Med 2010;164:860-869.
Fu Y, Wang T, Fu Q, Wang P, Lu Y. Associations between serum concentrations of perfluoroalkyl acids
and serum lipid levels in a Chinese population. Ecotoxicol Environ Saf 2014;106:246-52.
Gallo V, Leonardi G, Genser B, Lopez-Espinosa MJ, Frisbee SJ, Karlsson L, Ducatman AM, Fletcher T.
Serum perfluorooctanoate (PFOA) and perfluorooctane sulfonate (PFOS) concentrations and liver
function biomarkers in a population with elevated PFOA exposure. Environ Health Perspect
2012;120:655–660.
Geiger SD, Xiao J, Shankar A. Positive association between perfluoroalkyl chemicals and
hyperuricemia in children. Am J Epidemiol 2013;177:1255-1262.
Geiger SD, Xiao J, Ducatman A, Frisbee S, Innes K, Shankar A. The association between PFOA, PFOS
and serum lipid levels in adolescents. Chemosphere 2014a;98:78-83.
Geiger SD, Xiao J, Shankar A. No association between perfluoroalkyl chemicals and hypertension in
children. Integr Blood Press Control 2014b;7:1-7.
Gleason JA, Post GB, Fagliano JA Associations of perfluorinated chemical serum concentrations and
biomarkers of liver function and uric acid in the US population (NHANES), 2007-2010. Environ Res
2015;136:8-14.
Goudarzi H, Miyashita C, Okada E, Kashino I, Kobayashi S, Chen CJ, Ito S, Araki A, Matsuura H, Ito
YM, Kishi R. Effects of prenatal exposure to perfluoroalkyl acids on prevalence ofallergic diseases
among 4-year-old children. Environ Int 2016;94:124-32.
Grandjean P, Andersen EW, Budtz-Jørgensen E, Nielsen F, Mølbak K, Weihe P, Heilmann C. Serum
Vaccine Antibody Concentrations in Children Exposed to Perfluorinated Compounds. JAMA 2012;307:
391–397.
Grandjean P, Heilmann C, Weihe P, Nielsen F, Mogensen UB, Budtz-Jørgensen E. Serum vaccine
antibody concentrations in adolescents exposed to perfluorinated compounds. Environ Health Perspect
2016 Aug 9. [Epub ahead of print]
Granum, Haug LS, Namork E, Stølevik SB, Thomsen C, Aaberge IS, van Loveren H, Løvik M, Nygaard
UC. Pre-natal exposure to perfluoroalkyl substances may be associated with altered vaccine antibody
levels and immune-related health outcomes in early childhood. J Immunotoxicol 2013; 10:4, 373-379.
Grice MM, Alexander BH, Hoffbeck R, Kampa DM. Self-reported medical conditions in
perfluorooctanesulfonyl fluoride manufacturing workers. J Occup Environ Med 2007;49:722-729.
Gump BB, Wu Q, Dumas AK, Kannan K. Perfluorochemical (PFC) exposure in children: associations
with impaired response inhibition. Environ Sci Technol 2011;45(19):8151-9.

49

Hamm MP, Cherry NM, Chan E, Martin JW, Burstyn I. Maternal exposure to perfluorinated acids and
fetal growth. J Expo Sci Environ Epidemiol 2010;20(7):589-97.
Hardell E, Kärrman A, van Bavel B, Bao J, Carlberg M, Hardell L. Case–control study on perfluorinated
alkyl acids (PFAAs) and the risk of prostate cancer. Environ Int 2014;63:35-39.
Hoffman K, Webster TF, Weisskopf MG, Weinberg J, Vieira VM. Exposure to polyfluoroalkyl
chemicals and attention deficit/hyperactivity disorder in U.S. children 12-15 years of age. Environ
Health Perspect 2010;118(12):1762-7.
Humblet O, Diaz-Ramirez LG, Balmes JR, Pinney SM, Hiatt RA. Perfluoroalkyl Chemicals and Asthma
among Children 12-19 Years of Age: NHANES (1999-2008). Environ Health Perspect 2014; 122(10):
1129-1133.
Innes KE, Ducatman AM, Luster MI, Shankar A. Association of osteoarthritis with serum levels of the
environmental contaminants perfluorooctanoate and perfluorooctane sulfonate in a large Appalachian
population. Am J Epidemiol 2011;174:440-50.
Islami F, Moreira DM, Boffeta P, Freedland SJ. A systematic review and meta-analysis of tobacco use
and prostate cancer mortality and incidence in prospective cohort studies. Eur Urol 2014;66:1054-1064.
Jain RB. Association between thyroid profile and perfluoroalkyl acids: data from NHNAES 2007-2008.
Environ Res 2013;126:51-9.
Jensen TK, Andersen LB, Kyhl HB, Nielsen F, Christesen HT, Grandjean P. Association between
perfluorinated compound exposure and miscarriage in Danish pregnant women. PLoS ONE 2015;10(4):
e0123496.
Ji K, Kim S, Kho Y, Paek D, Sakong J, Ha J, Kim S, Choi K. Serum concentrations of major
perfluorinated compounds among the general population in Korea: Dietary sources and potential impact
on thyroid hormones. Environ Int 2012;45:78-85.
Jørgensen KT, Specht IO, Lenters V, Bach CC, Rylander L, Jönsson BA, Lindh CH, Giwercman A,
Heederik D, Toft G, Bonde JP. Perfluoroalkyl substances and time to pregnancy in couples from
Greenland, Poland and Ukraine. Environ Health 2014;13:116.
Karlsen M, Grandjean P, Weihe P, Steuerwald U, Oulhote Y, Valvi D. Early-life exposures to persistent
organic pollutants in relation to overweight in preschool children. Reprod Toxicol 2016 Aug 3. [Epub
ahead of print]
Kataria A, Trachtman H, Malaga-Dieguez L, Trasande L. Association between perfluoroalkyl acids and
kidney function in a cross-sectional study of adolescents. Environ Health 2015;14:89.
Kato S, Itoh S, Yuasa M, Baba T, Miyashita C, Sasaki S, Nakajima S, Uno A, Nakazawa H, Iwasaki Y,
Okada E, Kishi R. Association of perfluorinated chemical exposure in utero with maternal and infant
thyroid hormone levels in the Sapporo cohort of Hokkaido Study on the Environment and Children's
Health. Environ Health Prev Med 2016 Apr 30. [Epub ahead of print]
50

Kennedy GL, Symons JM. Carcinogenicity of Perfluoroalkyl Compounds. In J.C. Dewitt ed.
Toxicological Effects of Perfluoroalkyl and Polyfluoroalkyl Substances, Molecular and Integrative
Toxicology, Chapter 12. Springer (Switzerland 2015).
Khalil N. Chapter 13. Epidemiological Findings. In J.C. DeWitt ed. Toxicological Effects of
Perfluoroalkyl and polyfluoroalkyl Substances, Molecular and Integrative Toxicology. Springer
International, Switzerland 2015.
Khalil N, Chen A, Lee M, Czerwinski SA, Ebert JR, DeWitt JC, Kannan K. Association of
Perfluoroalkyl Substances, Bone Mineral Density, and Osteoporosis in the U.S. Population in NHANES
2009-2010. Environ Health Perspect 2016;124:81-7.
Kielsen K, Shamim Z, Ryder LP, Nielsen F, Grandjean P, Budtz-Jorgensen E, Heilmann C. Antibody
response to booster vaccination with tetanus and diphtheria in adults exposed to perfluorinated alkylates.
J Immunotoxicol 2016: 13:2, 270-273.
Kim DH, Kim UJ, Kim HY, Choi SD, Oh JE. Perfluoroalkyl substances in serum from South Korean
infants with congenital hypothyroidism and healthy infants--Its relationship with thyroid hormones.
Environ Res 2016;147:399-404.
Kishi R, Nakajima T, Goudarzi H, Kobayashi S, Sasaki S, Okada E, Miyashita C, Itoh S, Araki A, Ikeno
T, Iwasaki Y, Nakazawa H. The Association of Prenatal Exposure to Perfluorinated Chemicals with
Maternal Essential and Long-Chain Polyunsaturated Fatty Acids during Pregnancy and the Birth Weight
of Their Offspring: The Hokkaido Study. Environ Health Perspect 2015;123(10):1038-45.
Knox SS, Jackson T, Frisbee SJ, Javins B, Ducatman AM. Perfluorocarbon exposure, gender and
thyroid function in the C8 Health Project. J Toxicol Sci 2011;36:403-410.
Kristensen SL, Ramlau-Hansen CH, Ernst E, Olsen SF, Bonde JP, Vested A, Halldorsson TI, Becher G,
Haug LS, Toft G. Long-term effects of prenatal exposure to perfluoroalkyl substances on female
reproduction. Hum Reprod 2013;28(12):3337-48.
Lau C. Perfluorinated Compounds: An Overview. In J.C. Dewitt ed. Toxicological Effects of
Perfluoroalkyl and Polyfluoroalkyl Substances, Molecular and Integrative Toxicology, Chapter 12.
Springer (Switzerland 2015).
Lauritzen HB, Larose TL, Øien T, Sandanger TM, Odland JO, van de Bor M, Jacobsen GW. Maternal
serum levels of perfluoroalkyl substances and organochlorines and indices of fetal growth: a
Scandinavian case-cohort study, Pediatr Res 2017;81(1-1):33-42.
Lee ES, Han S, Oh JE. Association between perfluorinated compound concentrations in cord serum and
birth weight using multiple regression models. Reprod Toxicol 2016;59:53-9.
Lenters V, Portengen L, Rignell-Hydbom A, Jönsson BA, Lindh CH, Piersma AH, Toft G, Bonde JP,
Heederik D, Rylander L, Vermeulen R. Prenatal Phthalate, Perfluoroalkyl Acid, and Organochlorine
Exposures and Term Birth Weight in Three Birth Cohorts: Multi-Pollutant Models Based on Elastic Net
Regression. Environ Health Perspect 2016;124(3):365-72.
51

Lien GW, Huang CC, Shiu JS, Chen MH, Hsieh WS, Guo YL, Chen PC. Perfluoroalkyl substances in
cord blood and attention deficit/hyperactivity disorder symptoms in seven-year-old children.
Chemosphere. 2016;156:118-27.
Liew Z, Ritz B, Bonefeld-Jørgensen EC, Henriksen TB, Nohr EA, Bech BH, Fei C, Bossi R, von
Ehrenstein OS, Streja E, Uldall P, Olsen J. Prenatal exposure to perfluoroalkyl substances and the risk of
congenital cerebral palsy in children. Am J Epidemiol 2014;180(6):574-81.
Liew Z, Ritz B, von Ehrenstein OS, Bech BH, Nohr EA, Fei C, Bossi R, Henriksen TB, BonefeldJørgensen EC, Olsen J. Attention deficit/hyperactivity disorder and childhood autism in association with
prenatal exposure to perfluoroalkyl substances: a nested case-control study in the Danish National Birth
Cohort. Environ Health Perspect 2015;123(4):367-73.
Lin CY, Lin LY, Chiang CK, Wang WJ, Su YN, Hung KY, Chen PC. Investigation of the associations
between low-dose serum perfluorinated chemicals and liver enzymes in US adults. Am J Gastroenterol
2010;105:1354-63.
Lin CY, Wen LL, Lin LY, Wen TW, Lien GW, Hsu SH, Chien KL, Liao CC, Sung FC, Chen PC, Su
TC. The associations between serum perfluorinated chemicals and thyroid function in adolescents and
young adults. J Hazard Mater 2013;244-245:637-644.
Lin LY, Wen LL, Su TC, Chen PC, Lin CY. Negative association between serum perfluorooctane
sulfate concentration and bone mineral density in US premenopausal women: NHANES, 2005-2008. J
Clin Endocrinol Metab 2014;99:2173-80.
Loccisano AE, Campbell Jr. JL, Andersen ME, Clewell III HJ. Evaluation and prediction of
pharmacokinetics of PFOA and PFOS in the monkey and human using a PBPK model. Regul Toxicol
Pharmacol 2011;59:157-175.
Looker C, Luster MI, Calafat AM, Johnson VJ, Burleson GR, Burleson FG, Fletcher T. Influenza
Vaccine Response in Adults Exposed to Perfluorooctanoate and Perfluorooctanesulfonate. Toxicol Sci
2014; 138(1): 76–88.
Lopez-Espinosa MJ, Fletcher T, Armstrong B, Genser B, Dhatariya K, Mondal D, Ducatman A,
Leonardi G. Association of Perfluorooctanoic Acid (PFOA) and Perfluorooctane Sulfonate (PFOS) with
age of puberty among children living near a chemical plant. Environ Sci Technol 2011;45:8160-8166.
Lopez-Espinosa MJ, Mondal D, Armstrong B, Bloom MS, Fletcher T. Thyroid function and
perfluoroalkyl acids in children living near a chemical plant. Environ Health Perspect 2012;120:10361041.
Lopez-Espinosa MJ, Mondal D, Armstrong BG, Eskenazi B, Fletcher T. Perfluoroalkyl substances, sex
hormones, and insulin-like growth factor-1 at 6–9 years of age: a cross-sectional analysis within the C8
Health Project. Environ Health Perspect 2016;124:1269-1275.
Lundin JI, Alexander BH, Olsen GW, Church TR. Ammonium perfluorooctanoate production and
occupational mortality. Epidemiol 2009;20:921-928.
52

Maisonet M, Terrell ML, McGeehin MA, Christensen KY, Holmes A, Calafat AM, Marcus M. Maternal
concentrations of polyfluoroalkyl compounds during pregnancy and fetal and postnatal growth in British
girls. Environ Health Perspect 2012;120(10):1432-7.
Maisonet M, Näyhä S, Lawlor DA, Marcus M. Prenatal exposures to perfluoroalkyl acids and serum
lipids at ages 7 and 15 in females. Environ Int 2015a;82:49-60.
Maisonet M, Calafat AM, Marcus M, Jaakkola JJ, Lashen H. Prenatal exposure to perfluoroalkyl acids
and serum testosterone concentrations at 15 years of age in female ALSPAC study participants. Environ
Health Perspect 2015b;123:1325-1330.
Mattsson K, Rignell-Hydbom A, Holmberg S, Thelin A, Jönsson BA, Lindh CH1, Sehlstedt A, Rylander
L. Levels of perfluoroalkyl substances and risk of coronary heart disease: Findings from a populationbased longitudinal study. Environ Res 2015;142:148-54.
Melzer D, Rice N, Depledge MH, Henley WE, Galloway TS. Association between serum
perfluorooctanoic acid (PFOA) and thyroid disease in the U.S. National Health and Nutrition
Examination Survey. Environ Health Perspect 2010;118:686–692.
Mogensen UB, Grandjean P, Heilmann C, Nielsen F, Weihe P, Budtz-Jorgensen E. Structural equation
modeling of immunotoxicity associated with exposure to perfluorinated alkylates. Environ Health
2015;14:47.
Mora AM, Oken E, Rifas-Shiman SL, Webster TF, Gillman MW, Calafat AM, Ye X, Sagiv SK.
Prenatal exposure to perfluoroalkyl substances and adiposity in early and mid-childhood. Environ
Health Perspect 2016 Jun 28. [Epub ahead of print]
National Toxicology Program (NTP). Systematic Review of Immunotoxicity Associated With Exposure
to Perfluorooctanoic acid (PFOA) or Perfluorooctane Sulfonate (PFOS). U.S. DHHS, May 11, 2016.
Nelson JW, Hatch EE, Webster TF. Exposure to polyfluoroalkyl chemicals and cholesterol, body
weight, and insulin resistance in the general U.S. population. Environ Health Perspect 2010;118:197202.
New Hampshire Division of Public Health Services, Department of Health and Human Services (NH
DHHS). Pease PFC Blood Testing Program: April 2015 – October 2015. Pease Tradeport, Portsmouth,
NH, June 2016. http://www.dhhs.nh.gov/dphs/documents/pease-pfc-blood-testing.pdf
Nolan LA, Nolan JM, Shofer FS, Rodway NV, Emmett EA. The relationship between birth weight,
gestational age and perfluorooctanoic acid (PFOA)-contaminated public drinking water. Reprod Toxicol
2009;27(3-4):231-8.
Nolan LA, Nolan JM, Shofer FS, Rodway NV, Emmett EA. Congenital anomalies, labor/delivery
complications, maternal risk factors and their relationship with perfluorooctanoic acid (PFOA)contaminated public drinking water. Reprod Toxicol 2010;29(2):147-55.

53

Ode A, Kallen K, Gustafsson P, Rylander L, Jonsson BA, Olofsson P, Ivarsson SA, Lindh CH, RignellHydbom A. Fetal Exposure to Perfluorinated Compounds and Attention Deficit Hyperactivity Disorder
in Childhood. PLoS ONE 2014; 9(4): e95891.
Ogden CL, Carroll MD, Lawman HG, Fryar CD, Kruszon-Moran D, Kit BK, Flegal KM. Trends in
Obesity Prevalence Among Children and Adolescents in the United States, 1988-1994 Through 20132014. JAMA 2016;315:2292-2299.
Olsen GW, Burlew MM, Marshall JC, Burris JM, Mandel JH. Analysis of episodes of care in a
perfluorooctanesulfonyl fluoride production facility. J Occup Environ Med 2004;46:837-846.
Oulhote Y, Steuerwald U, Debes F, et al. Behavioral difficulties in 7-year old children in relation to
developmental exposure to perfluorinated alkyl substances. Environ Int 2016;97:237-245.
Qin XD, Qian Z, Vaughn MG, Huang J, Ward P, Zeng XW, Zhou Y, Zhu Y, Yuan P, Li M, Bai Z, Paul
G, Hao YT, Chen W, Chen PC, Dong GH, Lee YL. Positive associations of serum perfluoroalkyl
substances with uric acid and hyperuricemia in children from Taiwan. Environ Pollution 2016;212:519524.
Raleigh KK, Alexander BH, Olsen GW, Ramachandran G, Morey SZ, Church TR, Logan PW, Scott LL,
Allen EM. Mortality and cancer incidence in ammonium perfluorooctanoate production workers. Occup
Environ Med 2014;71:500-506.
Savitz DA, Stein CR, Bartell SM, Elston B, Gong J, Shin HM, Wellenius GA. Perfluorooctanoic acid
exposure and pregnancy outcome in a highly exposed community. Epidemiology 2012a;23(3):386-92.
Savitz DA, Stein CR, Elston B, Wellenius GA, Bartell SM, Shin HM, Vieira VM, Fletcher T.
Relationship of perfluorooctanoic acid exposure to pregnancy outcome based on birth records in the
mid-Ohio Valley. Environ Health Perspect 2012b;120(8):1201-7.
Schecter A, Malik-Bass N, Calafat AM, Kato K, Colacino JA, Gent TL, Hynan LS, Harris TR, Malla S,
Birnbaum L. Polyfluoroalkyl compounds in Texas children from birth through 12 years of age. Environ
Health Perspect 2012;120:590-594.
Shankar A, Xiao J, Ducatman A. Perfluoroalkyl chemicals and elevated serum uric acid in US adults.
Clin Epidemiol 2011;3:251-258.
Shankar A, Xiao J, Ducatman A. Perfluorooctanoic acid and cardiovascular disease in US adults. Arch
Intern Med 2012;172:1397-403.
Shin HM, Vieira VM, Ryan PB et al. Retrospective exposure estimation and predicted versus observed
serum perfluorooctanoic acid concentrations for participants in the C8 Health Project. Environ Health
Perspect 2011;119:1760-1765.
Shrestha S, Bloom MS, Yucel R, Seegal RF, Wu Q, Kannan K, Rej R, Fitzgerald EF. Perfluoroalkyl
substances and thyroid function in older adults. Environ Int 2015;75:206-14.

54

Starling AP, Engel SM, Richardson DB, Baird DD, Haug LS, Stuebe AM, Klungsøyr K, Harmon Q,
Becher G, Thomsen C, Sabaredzovic A, Eggesbø M, Hoppin JA, Travlos GS, Wilson RE, Trogstad LI,
Magnus P, Longnecker MP. Perfluoroalkyl substances during pregnancy and validated preeclampsia
among nulliparous women in the Norwegian Mother and Child Cohort Study. Am J Epidemiol
2014;179(7):824-33.
Steenland K, Tinker S, Frisbee S, Ducatman A, Vaccarino V. Association of perfluorooctanoic acid and
perfluorooctane sulfonate with serum lipids among adults living near a chemical plant. Am J Epidemiol
2009;170:1268-1278.
Steenland K Association of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) with
uric acid among adults with elevated community exposure to PFOA. Environ Health Perspect
2010;118:229–233.
Steenland K, Woskie S. Cohort mortality study of workers exposed to perfluorooctanoic acid. Am J
Epidemiol. 2012;176:909–917.
Steenland K, Zhao L, Winquist A, Parks C. 2013. Ulcerative colitis and perfluorooctanoic acid (PFOA)
in a highly exposed population of community residents and workers in the mid-Ohio valley. Environ
Health Perspect 2013; 121(8):900-5.
Steenland K, Zhao L, Winquist A. A cohort incidence study of workers exposed to perfluorooctanoic
acid (PFOA). Occup Environ Med 2015;72:373-380.
Stein CR, Savitz DA, Dougan M. Serum levels of Perfluorooctanoic Acid and Perfluorooctane Sulfonate
and pregnancy outcome. Am J Epidemiol. 2009;170:837–846.
Stein CR, Savitz DA. Serum perfluorinated compound concentration and attention deficit/hyperactivity
disorder in children 5-18 years of age. Environ Health Perspect 2011;119:1466-1471.
Stein CR, Savitz DA, Bellinger DC. Perfluorooctanoate and neuropsychological outcomes in children.
Epidemiology 2013;24:590-599.
Stein CR, Savitz DA, Elston B, Thorpe PG, Gilboa SM. Perfluorooctanoate exposure and major birth
defects. Reprod Toxicol. 2014a;47:15-20.
Stein CR, Savitz DA, Bellinger DC. Perfluorooctanoate exposure in a highly exposed community and
parent and teacher reports of behavior in 6-12-year-old children. Paediatr Perinat Epidemiol
2014b;28:146-156.
Stein CR, McGovern KJ, Pajak AM, Maglione PJ, Wolff MS. Perfluoroalkyl and Polyfluoroalkyl
Substances and Indicators of Immune function in children aged 12–19 y: National Health and Nutrition
Examination Survey. Pediatr Res. 2016a; 79(2):348-57.
Stein CR, Ge Y, Wolff MS et al. Perfluoroalkyl substance serum concentrations and immune response to
FluMist vaccination among healthy adults. Environ Res 2016b;149:171-178.

55

Strøm M, Hansen S, Olsen SF, Haug LS, Rantakokko P, Kiviranta H, Halldorsson TI. Persistent organic
pollutants measured in maternal serum and offspring neurodevelopmental outcomes--a prospective study
with long-term follow-up. Environ Int 2014;68:41-8.
Toft G, Jönsson BA, Bonde JP, Nørgaard-Pedersen B, Hougaard DM, Cohen A, Lindh CH, Ivell R,
Anand-Ivell R, Lindhard MS. Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers
of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in Danish Boys (1980-1996).
Environ Health Perspect 2016;124(1):151-6.
U.S. Air Force (USAF) (1990). Final Environmental Impact Statement for the Closure of Pease
AirForce Base. May 1990.
U.S. Air Force (USAF) (1994). Revised Community Relations Plan. Pease AFB. June 1994.
U.S. Environmental Protection Agency (EPA). Provisional Health Advisories for Perfluorooctanoic
Acid (PFOA) and Perfluorooctyl Sulfonate (PFOS); 2009. Available from
http://water.epa.gov/action/advisories/drinking/upload/2009_01_15_criteria_drinking_phaPFOA_PFOS.pdf
U.S. Environmental Protection Agency (EPA). Fact Sheet – PFOS & PFOA Drinking Water Health
Advisories. Ground Water and Drinking Water; 2016a. Available from:
https://www.epa.gov/sites/production/files/201606/documents/drinkingwaterhealthadvisories_pfoa_pfos_updated_5.31.16.pdf
U.S. Environmental Protection Agency (EPA). Third Unregulated Contaminant Monitoring Rule, July
2016b. Available from:
https://www.epa.gov/dwucmr/occurrence-data-unregulated-contaminant monitoring-rule#3
Uhl SA, James-Todd J, Bell ML. Association of Osteoarthritis with Perfluorooctanoate and
Perfluorooctane Sulfonate in NHANES 2003–2008. Environ Health Perspect 2013;121:447-452.
Vélez MP, Arbuckle TE, Fraser WD. Maternal exposure to perfluorinated chemicals and reduced
fecundity: the MIREC study. Hum Reprod. 2015;30(3):701-9.
Verner MA, Loccisano AE, Morken NH, Yoon M, Wu H, McDougall R, Maisonet M, Marcus M, Kishi
R, Miyashita C, Chen MH, Hsieh WS, Andersen ME, Clewell HJ 3rd, Longnecker MP. Associations of
perfluoroalkyl substances (PFAS) with lower birth weight: an evaluation of potential confounding by
glomerular filtration rate using a physiologically based pharmacokinetic model (PBPK). Environ Health
Perspect 2015;123(12):1317-24.
Vested A, Ramlau-Hansen CH, Olsen SF, Bonde JP, Kristensen SL, Halldorsson TI, Becher G, Haug
LS, Ernst EH, Toft G. Associations of in utero exposure to perfluorinated alkyl acids with human semen
quality and reproductive hormones in adult men. Environ Health Perspect 2013;121(4):453-8.
Vesterholm Jensen D, Christensen J, Virtanen HE, Skakkebæk NE, Main KM, Toppari J, Veje CW,
Andersson AM, Nielsen F, Grandjean P, Jensen TK. No association between exposure to perfluorinated
compounds and congenital cryptorchidism: a nested case-control study among 215 boys from Denmark
and Finland. Reproduction 2014;147(4):411-7.
56

Vieira VM, Hoffman K, Shin HM, Weinberg JM, Webster TF, Fletcher T. Perfluorooctanoic Acid
Exposure and Cancer Outcomes in a Contaminated Community: A Geographic Analysis. Environ
Health Perspect 2013;121:318–323.
Vuong AM, Yolton K, Webster GM, Sjödin A, Calafat AM, Braun JM, Dietrich KN, Lanphear BP,
Chen A. Prenatal polybrominated diphenyl ether and perfluoroalkyl substance exposures and executive
function in school-age children. Environ Res 2016;147:556-64.
Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample
size, median, range and/or interquartile range. BMC Med Res Methodol 2014;14:135.
Wang IJ, Hsieh WS, Chen CY, Fletcher T, Lien GW, Chiang HL, Chiang CF, Wu TN, Chen PC. The
effect of prenatal perfluorinated chemicals exposures on pediatric atopy. Environ Res 2011; 111:785–
791.
Wang Y, Rogan WJ, Chen PC, Lien GW, Chen HY, Tseng YC, Longnecker MP, Wang SL. Association
between maternal serum perfluoroalkyl substances during pregnancy and maternal and cord thyroid
hormones: Taiwan maternal and infant cohort study. Environ Health Perspect 2014;122:529-34.
Wang Y, Rogan WJ, Chen HY, Chen PC, Su PH, Chen HY, Wang SL. Prenatal exposure to
perfluroalkyl substances and children's IQ: The Taiwan maternal and infant cohort study. Int J Hyg
Environ Health 2015;218(7):639-44.
Washino N, Saijo Y, Sasaki S, Kato S, Ban S, Konishi K, Ito R, Nakata A, Iwasaki Y, Saito K,
Nakazawa H, Kishi R. Correlations between prenatal exposure to perfluorinated chemicals and reduced
fetal growth. Environ Health Perspect 2009;117(4):660-7.
Watkins DJ, Josson J, Elston B, Bartell SM, Shin H-M, Vieira VM, Savitz DA, Fletcher T, Wellenius
GA. Exposure to perfluoroalkyl acids and markers of kidney function among children and adolescents
living near a chemical plant. Environ Health Perspect 2013;121:625-630.
Webster GM, Venners SA, Mattman A, Martin JW. Associations between perfluoroalkyl acids (PFASs)
and maternal thyroid hormones in early pregnancy: a population-based cohort study. Environ Res
2014;133:338–347.
Webster GM, Rauch SA, Marie NS, Mattman A, Lanphear BP, Venners SA. Cross-sectional
associations of serum perfluoroalkyl acids and thyroid hormones in U.S. adults: variation according to
TPOAb and iodine status (NHANES 2007-2008). Environ Health Perspect. 2016;124(7):935-42.
Wen LL, Lin LY, Su TC, Chen PC, Lin CY. Association between serum perfluorinated chemicals and
thyroid function in U.S. adults: The National Health and Nutrition Examination Survey 2007–2010. J
Clin Endocrinol Metab 2013;98:E1456-E1464.
Whitworth KW, Haug LS, Baird DD, Becher G, Hoppin JA, Skjaerven R, Thomsen C, Eggesbo M,
Travlos G, Wilson R, Longnecker MP. Perfluorinated compounds and subfecundity in pregnant women.
Epidemiology 2012a;23(2):257-63.

57

Whitworth KW, Haug LS, Baird DD, Becher G, Hoppin JA, Skjaerven R, Thomsen C, Eggesbo M,
Travlos G, Wilson R, Cupul-Uicab LA, Brantsaeter AL, Longnecker MP. Perfluorinated compounds in
relation to birth weight in the Norwegian Mother and Child Cohort Study. Am J Epidemiol.
2012b;175(12):1209-16.
Winquist A, Steenland K. Perfluorooctanoic acid exposure and thyroid disease in community and
worker cohorts. Epidemiology 2014a;25:255-264.
Winquist A, Steenland K. Modeled PFOA exposure and coronary artery disease, hypertension, and high
cholesterol in community and worker cohorts. Environ Health Perspect 2014b; 122:1299-1305.
Wu K, Xu X, Peng L, Liu J, Guo Y, Huo X. Association between maternal exposure to
perfluorooctanoic acid (PFOA) from electronic waste recycling and neonatal health outcomes. Environ
Int 2012;48:1-8.
Wu XM, Bennett DH, Calafat AM, Kato K, Strynar M, Andersen E, Moran RE, Tancredi DJ, Tulve NS,
Hertz-Picciotto I. Serum concentrations of perfluorinated compounds (PFC) among selected populations
of children and adults in California. Environ Res 2015;136:264-273.
Zeng XW, Qian Z, Emo B, Vaughn M, Bao J, Qin XD, Zhu Y, Li J, Lee YL, Dong GH. Association of
polyfluoroalkyl chemical exposure with serum lipids in children. Sc Total Environ 2015;512-513;364370.
Zhang Y, Beesoon S, Zhu L, Martin JW. Biomonitoring of perfluoroalkyl acids in human urine and
estimates of biological half-life. Environ Sci Technol 2013;47:10619–10627.
Zhou Y, Hu LW, Qian ZM, Chang JJ, King C, Paul G, Lin S, Chen PC, Lee YL, Dong GH. Association
of perfluoroalkyl substances exposure with reproductive hormone levels in adolescents: By sex status.
Environ International 2016;94:189-195.
Zhu Y, Qin XD, Zeng XW, Paul G, Morawska L, Su MW, Tsai CH, Wang SQ, Lee YL, Dong GH.
Associations of serum perfluoroalkyl acid levels with T-helper cell-specific cytokines in children: By
gender and asthma status. Sci Total Environ 2016;559:166-73.

58

Tables

59

Table 1. Serum levels of selected PFAS in µg/L, children aged <12 years in Pease and comparison
populations.
Pease Tradeport*
TX†
CA‡
Geometric
PFAS
Median Geometric 95% CI
Max.
Median Max. Median
mean
mean
PFOA
3.63
3.43
3.23 – 3.64 12.00
2.85
13.50
4.50
4.46
PFOS
8.27
8.11
7.59 – 8.67 30.80
4.10
93.90
6.15
6.28
PFHxS
4.24
3.83
3.48 – 4.22 31.70
1.20
31.20
1.25
1.30
PFNA
0.90
0.92
0.86 – 0.98
5.20
1.20
55.80
1.70
1.84
*
Pease: N=366, aged <12 years; sampling occurred in 2015.
†
TX reference group: N=300, age ≤12 years (Schecter et al 2012), sampling occurred in 2009.
‡
CA reference group: N=68, ages 2-8 years (Wu et al 2015), sampling occurred in 2007-2009.

Max.
19.50
26.70
9.80
11.20

Table 2. Serum levels of selected PFAS in µg/L, aged ≥12 years in Pease and NHANES comparison
values.
Pease Tradeport*
NHANES 2013-2014†
Geometric
Geometric
95th
PFAS
Median
Mean
95% CI
Max.
Median
Mean
95% CI
percentile
PFOA
3.10
2.99
2.87 – 3.11
PFOS
9.17
8.74
8.37 – 9.13
PFHxS
4.16
4.21
3.98 – 4.46
PFNA
0.70
0.68
0.65 – 0.70
*
N=1,212 ages ≥12 years, sampled in 2015
† N=2,168 ages ≥12 years.

32.00
95.60
116.00
4.90

60

2.07
5.20
1.40
0.70

1.94
4.99
1.35
0.68

1.76 – 2.14
4.50 – 5.52
1.20 – 1.60
0.61 – 0.72

5.57
18.50
5.60
2.00

Table 3. Summary of the PFAS literature on adults.
PFOS
PFHxS
Cancer
Prostate
+
+
Bladder
+
*
Colorectal
+
*
Breast
I
I
Pancreatic
I
*
Testicular
*
*
Kidney
*
*
Thyroid
*
*
Liver
*
*
Leukemia
*
*
non-Hodgkin lymphoma
*
*
Multiple myeloma
*
*
Ovarian
*
*
Other diseases
Kidney disease/kidney function
*
*
Hyperuricemia
+
I
Liver disease/liver function
+
+
Cardiovascular Disease,
+
+
hypertension, hypercholesterolemia
Thyroid disease/function
+
+
Autoimmune diseases
*
*
Osteoarthritis, osteoporosis and
+
+
bone mineral density
Immune response
+
+
Reproductive outcomes
+
+

PFOA
+
+
I
+
+
+
+
+
+
+
+
*
+
+
+
+
+
+
+
+
+
+

“+” One or more studies suggesting increased risk of an adverse outcome (e.g., OR or RR ≥ 1.20)
“*” no studies were conducted (for liver cancer and PFOS, and multiple myeloma and PFOA, there were
too few deaths (≤2) to evaluate).
“I” inconclusive – the findings have not suggested an increased risk (e.g., an OR or RR <1.20)

61

Table 4. Summary of the PFAS literature on children.
PFOS
PFHxS PFOA
Adverse birth outcomes
+
+
+
Lipids
+
I
+
Thyroid function
+
*
+
Thyroid disease
I
*
+
Uric acid
+
+
+
Sex hormones
+
+
+
Delay in reaching puberty
+
I
+
Neurobehavioral outcomes
+
+
+
Immune function
+
+
+
Hypertension
I
*
I
Adiposity/BMI/Overweight
+
+
+
“+” One or more studies suggesting increased risk of an adverse outcome (e.g., OR or RR ≥ 1.20)
“*” no studies were conducted.
“I” inconclusive – the findings have not suggested an increased risk (e.g., an OR or RR <1.20)

Note: adverse birth outcomes are not included in this table because these outcomes are not feasible to
study at Pease. Although the number of children potentially exposed to the PFAS-contaminated drinking
water while attending daycare at the Pease Tradeport can be estimated, there is a lack of information on
the number of children potentially exposed in utero to the PFAS-contaminated drinking water because
their mothers were employed at the Pease Tradeport during the pregnancy. To evaluate adverse birth
outcomes with sufficient statistical power would require the inclusion of several hundreds of exposed
births.

62

Table 5. Pease serum levels in µg/L for PFOS, PFOA and PFHxS, based on ages in 2018.

Population age
(during 2018)
4 – 17 (N=379)

PFOS

PFOA

PFHxS

Mean
SD
Median
Geometric mean
Maximum
Top quartile
Top quintile

9.66
6.01
8.09
7.98
30.80
12.70
14.20

3.85
2.14
3.40
3.29
12.00
5.00
5.41

5.47
4.62
4.10
3.75
31.70
7.60
8.79

Mean
SD
Median
Geometric mean
Maximum
Top quartile
Top quintile

11.59
9.63
9.30
8.82
95.60
14.20
16.10

3.76
2.74
3.12
3.02
32.00
4.65
5.20

7.05
8.75
4.21
4.26
116.00
8.60
10.00

≥18, not a firefighter (n=1,092) Mean
SD
Median
Geometric mean
Maximum
Top quartile
Top quintile

10.95
8.32
8.97
8.53
78.00
13.70
15.50

3.75
2.78
3.10
2.99
32.00
4.62
5.10

6.52
7.44
4.08
4.04
61.40
8.10
9.45

Firefighter (n=98)

18.70
17.41
11.75
12.80
95.60
23.10
28.64

3.93
2.14
3.40
3.37
12.10
5.26
5.90

12.95
16.64
8.14
7.74
116.00
14.70
17.46

≥18 (n=1,190)

Mean
SD
Median
Geometric mean
Maximum
Top quartile
Top quintile

63

Table 6a. Minimum detectable effects for a Pease children study with 350 exposed and 175
unexposed.*
Endpoint
Total cholesterol
(mean difference)
Hypercholesterolemia
Hyperuricemia
Uric acid (mean
difference)
eGFR (mean
difference)#
ADHD¶
ADHD + meds¶
Atopic dermatitis
Asthma
Rhinitis
Hypertension
Overweight/Obese

α and β = .05
9.8 mg/dL

α = .05, β=.20
7.6 mg/dL

α and β = .10
8.0 mg/dL

α = .10, β=.20
6.8 mg/dL

OR = 2.00
OR = 2.30
0.40 mg/dL

OR = 1.73
OR = 1.96
0.31 mg/dL

OR = 1.77
OR = 2.00
0.33 mg/dL

OR =1 .63
OR = 1.83
0.28 mg/dL

9.3

7.2

7.5

6.4

OR = 2.47
OR = 3.50
OR = 2.49
OR = 2.56
OR = 2.08
OR = 2.12
OR = 2.18

OR = 2.09
OR = 2.80
OR = 2.10
OR = 2.16
OR = 1.79
OR = 1.80
OR = 1.85

OR = 2.13
OR = 2.89
OR = 2.15
OR = 2.21
OR = 1.83
OR = 1.85
OR = 1.90

OR = 1.94
OR = 2.52
OR = 1.95
OR = 2.00
OR = 1.69
OR = 1.69
OR = 1.75

Table 6b. Minimum detectable effects for a Pease children study with 500 exposed and 250
unexposed.*
Endpoint
Total cholesterol
(mean difference)
Hypercholesterolemia
Hyperuricemia
Uric acid (mean
difference)
eGFR (mean
difference)#
ADHD¶
ADHD + meds¶
Atopic dermatitis
Asthma
Rhinitis
Hypertension
Overweight/Obese

α and β = .05
8.2 mg/dL

α = .05, β=.20
6.4 mg/dL

α and β = .10
6.7 mg/dL

α = .10, β=.20
5.7 mg/dL

OR = 1.78
OR = 2.04
0.34 mg/dL

OR = 1.57
OR = 1.75
0.26 mg/dL

OR = 1.60
OR = 1.79
0.27 mg/dL

OR =1 .50
OR = 1.65
0.23 mg/dL

7.7

6.0

6.3

5.4

OR = 2.18
OR = 2.98
OR = 2.20
OR = 2.26
OR = 1.85
OR = 1.88
OR = 1.93

OR = 1.85
OR = 2.40
OR = 1.86
OR = 1.91
OR = 1.62
OR = 1.64
OR = 1.69

OR = 1.90
OR = 2.48
OR = 1.91
OR = 1.96
OR = 1.65
OR = 1.68
OR = 1.73

OR = 1.73
OR = 2.19
OR = 1.74
OR = 1.78
OR = 1.54
OR = 1.56
OR = 1.60

*

Some health-related endpoints are not included in the table because there was insufficient information
to calculate minimum detectable effects. For sex hormones, insulin-like growth factor – 1, and thyroid
function, see the appendix for a description of the assumptions used in the sample size calculations and
the resulting calculations.
#

mL/min/1.73 m2
64

¶

The prevalence of an ADHD diagnosis reported by a study participant in the C8 study (Stein 2011) was
12.4%. In this study, the prevalence of an ADHD diagnosis reported by a study participant who also
reported currently using a medication commonly used to treat ADHD was 5.1%.

65

Table 6c. Summary of information used to categorize the feasibility of studying health-related endpoints for a Pease children study.
Health-related
Endpoint

Lipids (total
cholesterol)

Minimum
Detectable Effect
Size: 350 exposed,
175 unexposed
6.8 mg/dL

Estimated glomerular
filtration rate (eGFR)

5.5 mL/min/1.73 m2

Insulin-like growth
hormone-1 (IGF-1)

See appendix for
sample size
calculations and
assumptions
required for the
calculations.
OR=1.62

Overweight/Obesity

Other Sample Size Considerations

Conclusion

A Taiwan study (Zeng 2015) obtained mean differences of 11-12
mg/dL for total cholesterol and low density lipoprotein at PFOA
serum levels similar to Pease.
A NHANES study (Kataria 2015) observed a mean difference of
6.6 mL/min/1.73 m2 for PFOA serum levels similar to those at
Pease. For PFOS, the mean difference was 7.2 mL/min/1.73 m2
A C8 study (Lopez-Espinosa 2016) observed a reduction of IGF-1
for PFHxS serum levels similar to those at Pease that could be
detected with sufficient power by a sample size of 350 exposed
and 175 unexposed.

Feasible to study at Pease

Feasible to study at Pease.

A Faroes study (Karlsen 2016) observed and OR of 1.88 for PFOA Feasible to study at Pease.
serum levels below those at Pease. This OR could be detected
with a sample size of 350 exposed and 175 unexposed children.
The prevalence of obesity in children is 17% (Ogden 2016)

Hypercholesterolemia OR=1.63

A NHANES study (Geiger 2014) obtained ORs of 1.49 and 1.35
for serum PFOA and PFOS levels similar to those at Pease. To
detect an OR of 1.49 with 80% power requires a minimum of 540
exposed and 270 unexposed

Uric acid

A NHANES study (Kataria 2015) obtained a mean difference of
0.21 mg/dL for PFOA serum levels similar to Pease. However, for
PFOS, the mean difference was 0.05 mg/dL.

0.28 mg/dL

Feasible to study at Pease

66

Possible to study at Pease
although a sample size of at
least 500 exposed and 250
unexposed would be
necessary (see table 6b).
Possible to study at Pease
although a larger sample size
than 500 exposed and 250
unexposed would be
necessary.

Health-related
Endpoint

Minimum
Detectable Effect
Size
OR=1.83

Other Sample Size Considerations

Conclusion

A Taiwan study (Qin 2016) obtained an OR of 1.65 for PFHxS
serum levels much lower than at Pease. For PFOA serum levels
lower than at Pease, an OR of 2.2 was obtained.

IQ

3 point mean
difference

A Taiwan study (Wang 2015) obtained IQ mean differences of ≤2
points for PFOS serum levels higher than at Pease. A C8 study
(Stein 2013) did not find a decrease in IQ with PFOA exposure
and did not evaluate PFOS or PFHxS.

Neurobehavioral

Could not be
calculated due to
insufficient
information

Sex hormones

See appendix for
sample size
calculations and
assumptions
required for the
calculations.

Some studies had sample sizes achievable at Pease while others
had much larger sample sizes. The effects observed were not large
(e.g., an OR for learning problems was 1.2 for PFHxS and lower
for the other PFAS, and ORs for hyperactivity and coordination
problems were <1.5 for each of the PFAS). The few studies that
have been conducted evaluated different neurobehavioral tests.
At PFOS serum levels much higher than at Pease, a C8 study
(Lopez-Espinosa 2016) observed reductions in estradiol that
would require a sample size of over a thousand of exposed to
achieve sufficient statistical power. However, the observed
reductions in testosterone would require a sample size of between
500 and 1,000 exposed.

Possible to study at Pease
although a sample size of at
least 500 exposed and 250
exposed may be necessary to
evaluate the effect of serum
PFHxS. (For serum PFOA,
the Pease sample size of 350
exposed and 175 unexposed
may be sufficient)
Possible to study at Pease
although a sample size larger
than 500 exposed and 250
unexposed would be
necessary.
Similar conclusion as for IQ:
Possible to study at Pease
although a sample size larger
than 500 exposed and 250
unexposed would be
necessary.
Possible to study at Pease
although a sample size larger
than 500 exposed and 250
unexposed would be
necessary.

Hyperuricemia

67

Health-related
Endpoint

Minimum
Detectable Effect
Size
Thyroid function
See appendix for
sample size
calculations and
assumptions
required for the
calculations.
Liver function/CK-18 Could not be
calculated due ot
insufficient
information

Atopic dermatitis

OR=1.95

Asthma

OR=2.00

Rhinitis

OR=1.69

Other Sample Size Considerations

Conclusion

A C8 study (Lopez-Espinosa 2012) observed small differences for
PFOS and PFOA serum levels considerably higher than at Pease.
To detect these differences would require a sample size of over a
thousand exposed. On the other hand, a Taiwan study (Lin 2013)
observed differences that could be detected with sufficient power
with a sample size of 350 exposed and 175 unexposed.
No PFAS study has evaluated liver function or fatty liver disease
biomarkers in children. In a meta-analysis (Anderson 2015), the
prevalence of non-alcoholoic fatty liver disease was 11% among
those aged >5 - ≤15 years. Higher prevalences occur among
children who are overweight. Childhood studies that evaluated
biomarker CK-18 have used sample sizes smaller than those
anticipated at Pease, but these studies included a high proportion
of children with fatty liver disease.
A Taiwan study (Wang 2011) obtained an OR of 2.19 for PFOS
serum levels similar to Pease. However, the study evaluated
children aged 2 years. No other PFAS study evaluated atopic
dermatitis
Two NHANES studies (Humblet 2014, Stein 2016) observed ORs
between 1.2 and 1.3 which would require a sample size of over
2,000 exposed. However, a Taiwan study (Dong 2013) obtained
ORs between 3.8 and 4.0 for PFHxS and PFOA serum levels
lower than at Pease.
A NHANES study (Stein 2016a) evaluated rhinitis and obtained
an OR of 1.35 for serum PFOA similar to those at Pease. To
detect this OR would require over a thousand exposed. However,
ORs between 1.5 and 1.6 could be detected with sufficient
statistical power with a sample size of 500 exposed and 250
unexposed. These are ORs that are reasonable to detect and fall
within the 95% CI for the finding in the NHANES study.

Possible to study at Pease.

68

Possible to study at Pease

Possible to study at Pease.

Possible to study at Pease.

Possible to study at Pease

Health-related
Endpoint

Minimum
Detectable Effect
Size
Could not be
calculated due to
insufficient
information

Other Sample Size Considerations

Conclusion

Three studies that have been conducted of these endpoints had
sample sizes that could be achievable at Pease. Only two studies
(Granum 2013, Stein 2016) have evaluated the same endpoint –
rubella.

Possible to study at Pease
although a sample size larger
than 500 exposed and 250
exposed may be necessary.

Attention
deficit/hyperactivity
disorder (ADHD)

ORs: 1.9 – 2.5

Autism spectrum
disorder (ASD)

ORs > 4.0

Not feasible to study using the
Pease population alone (for
ADHD confirmed by current
medications)
Not feasible to study using the
Pease population alone.

Delayed puberty

Could not be
calculated due to
insufficient
information
OR > 8.0

A C8 study (Stein 2011) obtained an OR of 1.55 (ADHD + meds)
for PFHxS serum levels similar to Pease. A NHANES study
(Hoffman 2010) observed an OR of 1.67 for PFHxS serum levels
similar to Pease.
One study (Liew 2015) obtained an OR of 1.3 for serum PFHxS
levels lower than at Pease. To detect this OR would require
>10,000 exposed.
Only one study evaluated delayed puberty among children. This
was a C8 study (Lopez-Espinosa 2011) that evaluated several
thousand children. It is likely that sample sizes much larger than at
Pease would be necessary.
A C8 study (Lopez-Espinosa 2012) obtained an OR of 1.44 for
PFOA serum levels considerably higher than those in the Pease
population. To detect this OR with 80% statistical power would
require a sample size of over 10,000 exposed children.
No PFAS study has evaluated childhood cancers. Given the
incidence and prevalence of cancers such as leukemia, a sample
size of many thousands of exposed would be necessary.

Not feasible to study using the
Pease population alone.

Antibody response to
childhood vaccines

Thyroid disease

Childhood cancers

Not feasible to study using the
Pease population alone.

Not feasible to study using the
Pease population alone.

The minimum detectable effect size is based on a sample size of 350 children exposed and 175 children unexposed, and specifying statistical
power of 80% (or a type 2 or “β” error of .20) and a type 1 (“α”) error of .10 (see table 6a). This minimum detectable effect size is compared
to the adverse effect sizes observed in other PFAS studies. Where possible, the focus is on adverse effect sizes in the PFAS studies observed
for PFAS serum levels similar to those among the Pease children. An endpoint is considered feasible to study at Pease if an adverse effect
size observed in PFAS study can be detected with sufficient statistical power (i.e., statistical power of ≥80%) by a sample size achievable at
Pease, i.e., a sample size of 350 exposed children at Pease and 175 children unexposed to the PFAS-contaminated drinking water at Pease. If
69

only one PFAS study has been conducted on a health-related endpoint, then the endpoint was considered feasible to study at Pease if an odds
ratio of <2.0 could be detected with statistical power of 80%.
Note: The studies mentioned in the column of the table labeled “Other Sample Size Considerations” are included only to give a sense of the
adverse effect sizes that might occur in a Pease study. Due to the paucity of studies for each health-related endpoint, there is considerable
uncertainty concerning the effect sizes that might be expected to occur in a Pease study.
OR: Odds ratio. The odds ratio roughly approximates the risk ratio. The risk ratio is the proportion of the exposed population with a disease
divided by the proportion of the unexposed population with a disease.
Note: Hypertension is not included in this table because there is no evidence so far of an association between PFAS serum levels and
hypertension in children. Liver function and fatty acid biomarkers are not included in this table because no study has been done to evaluate
these biomarkers and PFAS exposure in children. Adverse birth outcomes are not included in this table because these outcomes are not
feasible to study at Pease. Although the number of children potentially exposed to the PFAS-contaminated drinking water while attending
daycare at the Pease Tradeport can be estimated, there is a lack of information on the number of children potentially exposed in utero to the
PFAS-contaminated drinking water because their mothers were employed at the Pease Tradeport during the pregnancy. To evaluate adverse
birth outcomes with sufficient statistical power would require the inclusion of several hundreds of exposed births.
Note: The health-related endpoints listed in this table satisfy the criteria of scientific importance and public health significance as discussed
on page 8 of the text.

70

Table 7a. Minimum detectable effects for an adult epidemiological study, 1,500 per stratum.*

Endpoint
Chronic kidney disease
Thyroid disease, unconfirmed
Thyroid disease, confirmed
Total cholesterol (mean
difference)
LDL (mean difference)
Hypercholesterolemia
Uric acid (mean difference)
Hyperuricemia
Elevated ALT (>45 IU/L, men;
>34 IU/L, women)
Elevated GGT (>55 IU/L, men;
>38 IU/L, women)
Elevated direct bilirubin
(>0.03 mg/dL)
ALT (mean difference)
GGT (mean difference)
Direct bilirubin (mean
difference)
Liver disease
Cardiovascular disease
Hypertension
Ulcerative colitis
Rheumatoid arthritis
Lupus
Multiple Sclerosis
Osteoporosis
Osteoarthritis
Endometriosis
(750 per stratum)
Pregnancy-induced hypertension
(750 per stratum)
Kidney cancer

α and β = .05
OR=2.54
OR=1.48
OR=1.63
5.5 mg/dL

α = .05, β=.20
OR=2.14
OR=1.36
OR=1.48
4.3 mg/dL

α and β = .10
OR=2.20
OR=1.38
OR=1.50
4.5 mg/dL

α = .10, β=.20
OR=2.00
OR=1.32
OR=1.42
3.8 mg/dL

4.5 mg/dL
OR=1.42
0.21 mg/dL
OR=1.35
OR=1.49

3.5 mg/dL
OR=1.32
0.17 mg/dL
OR=1.27
OR=1.37

3.7 mg/dL
OR=1.34
0.18 mg/dL
OR=1.28
OR=1.39

3.1 mg/dL
OR=1.28
0.15 mg/dL
OR=1.24
OR=1.33

OR=1.44

OR=1.33

OR=1.35

OR=1.29

OR=2.80

OR=2.34

OR=2.40

OR=2.16

2.65 IU/L
5.92 IU/L
0.079 mg/dL

2.06 IU/L
4.60 IU/L
0.060 mg/dL

2.15 IU/L
4.80 IU/L
0.064 mg/dL

1.83 IU/L
4.09 IU/L
0.055 mg/dL

OR=2.24
OR=1.45
OR=1.31
OR=4.13
OR=2.70
OR=6.87
OR=5.30
OR=1.73
OR=1.58
OR=1.92

OR=1.92
OR=1.34
OR=1.24
OR=3.24
OR=2.25
OR=4.97
OR=3.97
OR=1.55
OR=1.44
OR=1.69

OR=1.97
OR=1.36
OR=1.25
OR=3.38
OR=2.32
OR=5.24
OR=4.15
OR=1.58
OR=1.46
OR=1.73

OR=1.80
OR=1.30
OR=1.21
OR=2.94
OR=2.10
OR=4.33
OR=3.50
OR=1.48
OR=1.39
OR=1.61

OR=1.84

OR=1.63

OR=1.66

OR=1.55

OR=5.60

OR=4.27

OR=4.45

OR=3.80

*

Some health-related endpoints are not included in the table because there was insufficient information
to calculate minimum detectable effects. For thyroid function, see the appendix for a description of the
assumptions used in the sample size calculations and the resulting calculations.

71

Table 7b. Summary of information used to categorize the feasibility of studying health-related endpoints for a Pease adult study.
Health-related
Endpoint

Minimum
Detectable Effect
Size: 1,500 exposed
and 1,500
unexposed
Lipids (total cholesterol) 3.8 mg/dL

Hypercholesterolemia

OR=1.28

Uric acid

0.15 mg/dL

Hyperuricemia

OR=1.24

Thyroid disease
(unconfirmed)

OR=1.32

Cardiovascular disease

OR=1.30

Hypertension

OR=1.21

Osteoarthritis

OR=1.39

Osteoporosis

OR=1.48

Other Sample Size Considerations

Conclusion

A C8 study (Steenland 2009) observed a 3 – 4 mg/dL
change in total cholesterol and LDL for PFOS serum
levels similar to those at Pease.
A Canadian study (Fisher 2013) obtained an OR of 1.57
for PFHxS serum levels similar to those at Pease.
A NHANES study (Shankar 2011) observed a mean
difference of 0.40 mg/dL for serum PFOA levels similar
to those at Pease.
A NHANES study (Shankar 2011) obtained an OR of
1.90 for serum PFOA levels similar to those at Pease.
A C8 study (Winquist 2014a), hazard ratios ≤1.3 were
obtained for PFOA serum levels similar to those at
Pease. (Only PFOA was evaluated in this study.)
A NHANES study (Shankar 2012) obtained an OR of
2.01 for PFOA serum levels similar to those at Pease.
Only one community study (a C8 study, Winquist
2014b), evaluated hypertension and obtained an OR <
1.0 for serum PFOA (the only PFAS evaluated).
However, the sample size achievable at Pease is capable
of detecting very low ORs with sufficient statistical
power.
A NHANES study (Uhl 2013) obtained an OR of 1.5 for
serum PFOA levels similar to those at Pease.
A NHANES study (Khalil 2016) obtained an OR > 10
among women, for serum PFHxS levels lower than those
at Pease.

Feasible to study at Pease

72

Feasible to study at Pease
Feasible to study at Pease

Feasible to study at Pease
Feasible to study at Pease

Feasible to study at Pease
Feasible to study at Pease

Feasible to study at Pease
Feasible to study at Pease

Health-related
Endpoint
Serum Immune
Biomarkers

Minimum
Detectable Effect
Size
Could not be
calculated due to
insufficient
information

Liver function:
Elevated ALT
OR=1.33
Elevated GGT
OR=1.29
Elevated direct bilirubin OR=2.16

Other Sample Size Considerations

Conclusion

Only one published study (Stein 2016b) has been
conducted that evaluated serum immune biomarkers at
baseline (i.e., cross-sectionally). This study had a sample
size of 75 adults. A cross-sectional evaluation of PFAS
serum levels and immune biomarkers in a Pease adult
study could provide important information on the effects
of PFAS exposures on immune function in humans.

Feasible to study at Pease

A NHANES study (Gleason 2015) evaluated PFAS
serum levels similar to those at Pease. For elevated ALT,
ORs between1.2 and 1.5 were obtained. For elevated
GGT, ORs between 1.0 and 1.3 were obtained. For
elevated direct bilirubin, ORs between 1.1 and 1.7 were
obtained. No study has evaluated PFAS exposures and
fatty liver disease biomarker CK-18

Possible to study at Pease, but may
require a larger sample size than
1,500 exposed and 1,500
unexposed to evaluate PFOS and
PFHxS serum levels and ALT and
GGT. Direct bilirubin is probably
not feasible to study using the
Pease population alone.
Possible to study at Pease, but will
require a larger sample size than
1,500 exposed and 1,500
unexposed.
Possible to study at Pease.

CK-18

Not studied

Thyroid disease
(confirmed)

OR=1.42

A C8 study (Winquist 2014a), hazard ratios ≤1.3 were
obtained for PFOA serum levels similar to those at
Pease. (Only PFOA was evaluated in this study.)

Thyroid function

See appendix for
sample size
calculations and
assumptions required
for the calculations.

A C8 study (Knox 2011) observed very subtle changes
that would require a study of equivalent size (52,296) to
detect associations with sufficient statistical power. On
the other hand, a NHANES study (Wen 2013) observed
larger changes (at PFAS serum levels similar to those at
Pease) that could be detected with a sample size
achievable at Pease.

73

Health-related
Endpoint
Endometriosis

Minimum
Detectable Effect
Size
OR=1.61
(750 exposed & 750
unexposed)

Other Sample Size Considerations

Conclusion

A NHANES study (Campbell 2016) obtained ORs of
1.47 and 2.86 for serum PFHxS and PFOA, respectively.
The serum levels for these two PFAS were similar to
those in the Pease population.
A C8 study (Stein 2009, Darrow 2013) obtained an OR
of 1.6 for serum PFOS levels higher than at Pease.

Possible to study at Pease if
sufficient numbers of women can
be recruited.

Pregnancy-induced
hypertension

OR=1.55 (750
exposed pregnancies
and 750 unexposed
pregnancies

Liver disease

OR=1.80

A C8 study (Darrow 2016) and a NHANES study
(Melzer 2010) observed no elevation in liver disease.
However, the C8 study evaluated only PFOA and the
NHANES study evaluated PFOA and PFOS but not
PFHxS.

Not feasible to study using the
Pease population alone.

Kidney disease

OR=2.00

Not feasible to study using the
Pease population alone.

Ulcerative colitis

OR=2.94

Rheumatoid arthritis

OR=2.10

Lupus

OR=4.33

Multiple sclerosis

OR=3.50

A C8 study (Dhingra 2016a) evaluated only PFOA and
obtained ORs of 1.26 and 1.36 for the retrospective and
prospective analyses, respectively, at the second quintile
PFOA serum level. (Smaller ORs were observed at
higher PFOA serum levels.)
A C8 study (Steenland 2013) observed RRs between 2.8
and 3.1 at the highest serum PFOA levels, considerably
higher than those at Pease. At lower PFOA serum levels,
the RRs were <2.2
A C8 study (Steenland 2013) observed RRs between 1.3
and 1.7 for serum PFOA.
A C8 study (Steenland 2013) observed RRs <1.3 for
serum PFOA.
A C8 study (Steenland 2013) observed RRs between 1.1
and 1.6 for serum PFOA
74

Possible to study at Pease but may
require a larger sample size than
1,500 exposed and 1,500
unexposed in order to achieve a
sufficient number of pregnancies.

Not feasible to study using the
Pease population alone.

Not feasible to study using the
Pease population alone.
Not feasible to study using the
Pease population alone.
Not feasible to study using the
Pease population alone.

Health-related
Endpoint
Kidney cancer

Minimum
Detectable Effect
Size
OR=3.80 for kidney
cancer

Other Sample Size Considerations

Conclusion

A C8 study of a community population (Vieira 2013)
observed an RR of 1.70 for those served by the Little
Hocking water system.

Not feasible to study using the
Pease population alone. (Due to the
very low background prevalences
of other adult cancers, it is not
feasible to study cancers using the
Pease population alone.)

The minimum detectable effect size is based on a sample size of 1,500 adults exposed and 1,500 adults unexposed, and specifying statistical
power of 80% (or a type 2 or “β” error of .20) and a type 1 (“α”) error of .10 (see table 6a). This minimum detectable effect size is compared
to the adverse effect sizes observed in other PFAS studies. Where possible, the focus is on adverse effect sizes in the PFAS studies observed
for PFAS serum levels similar to those among the Pease adults. An endpoint is considered feasible to study at Pease if an adverse effect size
observed in PFAS study can be detected with sufficient statistical power (i.e., statistical power of ≥80%) by a sample size of 1,500 exposed
and 1,500 unexposed. If only one PFAS study has been conducted on a health-related endpoint, then the endpoint was considered feasible to
study at Pease if an odds ratio of <2.0 could be detected with statistical power of 80%.
Note: the studies mentioned in the column of the table labeled “Other Sample Size Considerations” are included only to give a sense of the
adverse effect sizes that might occur in a Pease study. Due to the paucity of studies for each health-related endpoint, there is considerable
uncertainty concerning the effect sizes that might be expected to occur in a Pease study.
OR: odds ratio. The odds ratio roughly approximates the risk ratio (RR). The risk ratio is the proportion of the exposed population with a
disease divided by the proportion of the unexposed population with a disease.
A hazard ratio can be interpreted in the same way as a risk ratio.
Note: The health-related endpoints listed in this table satisfy the criteria of scientific importance and public health significance as discussed
on page 8 of the text.

75

Appendix

76

Literature review
The literature review focused on the epidemiological results for PFOA, PFOS and PFHxS since these
were the major contaminants detected in the Haven Well during the April and May 2014 sampling as
well as the elevated PFAS in the serum of those tested in the NH DHHS Pease testing program. The
purpose of the literature review was to identify the health-related endpoints that have been evaluated in
at least one epidemiological study, and to assess the extent of the epidemiological research on the health
effects of PFHxS and PFOS. The findings of the studies included in the literature review were also used
to inform sample size calculations.
Literature searches using PubMed were conducted to identify epidemiological studies that evaluated
measured or estimated serum levels of PFOS, PFOA and PFHxS. The key words used in the search were
perfluorooctane sulfonate, PFOS, perfluorooctanoic acid, PFOA, perfluorohexane sulfonate, PFHxS,
perfluoroalkyl substances, PFAS, perfluorinated compounds, PFC, and perfluorinated chemicals. The
PubMed search identified epidemiological studies through October 31, 2016.

Cancers
The C8 science panel in 2012 reviewed the literature on PFAS and cancers and concluded that there was
a “probable link” between exposure to PFOA and testicular and kidney cancer
(http://www.c8sciencepanel.org/pdfs/Probable_Link_C8_Cancer_16April2012_v2.pdf). No other
cancers were considered to have a probable link with PFOA exposure. The panel noted that PFOA
caused liver, testicular and pancreatic tumors (adenomas) in rodent studies.
A review of the literature by DuPont researchers noted that PFOS causes liver adenomas in rodent
studies (Kennedy and Symons 2015) but concluded that the evidence of associations between
community drinking water exposures to PFOA and kidney and testicular cancers was “limited”. The
review also concluded that studies of populations exposed to low levels of PFOA and PFOS had
equivocal results with no consistent associations across studies. Studies of workers exposed to higher
levels of PFOS and PFOA were also viewed as lacking consistent associations. Their overall conclusion
was: “Based on the evidence reported to date, the prospect for developing a carcinogenic outcome
following exposure to PFOA and PFOS is remote.”
ATSDR’s literature search identified fifteen studies and the results from these studies are summarized in
Table A2. Based on its assessment of the epidemiological literature, ATSDR concluded that for most
cancers there was either limited information or no information concerning associations with PFAS
exposures. In particular, very few studies have evaluated PFHxS exposures and cancers. Although more
information is available for PFOS exposure and cancers, the information is still very limited. Clearly
more research is needed to investigate whether PFHxS and PFOS exposures are associated with
increased risks of specific cancers. More information is available on PFOA exposure and cancers
because of the studies conducted of the C8 population (workers and community members) and of
workers at the 3M Cottage Grove plant. However, the available information is still too limited to
determine whether a causal association exists between PFOA exposure and specific cancers. While
cancer research at Pease is not feasible, additional research on the effects of PFOA exposure on specific
cancers has the potential to provide the evidence necessary to assess whether PFOA is a cause of one or
more specific cancers.
77

Other adult diseases (Table A3)
The C8 science panel reviewed the literature for adult non-cancer diseases and found probable links for
PFOA and ulcerative colitis (an autoimmune disease), hypercholesterolemia (high cholesterol), thyroid
disease (hyperthyroidism in females, hypothyroidism in males), and pregnancy-induced hypertension
(PIH). The panel concluded that the evidence was not sufficient for a probable link between PFOA and
other autoimmune diseases (e.g., lupus, rheumatoid arthritis and Crohn’s disease), stroke, hypertension,
coronary artery disease, diabetes, chronic kidney or liver disease, asthma, chronic obstructive pulmonary
disease, osteoarthritis or Parkinson disease (http://www.c8sciencepanel.org/prob_link.html). Another
review of the literature noted that PFOA was linked to uric acid levels and that PFAS exposure was
associated with elevated liver enzymes, osteoarthritis, kidney disease and immunotoxicity in some
studies, but the findings across studies were inconsistent (Khalil 2015). The following literature review
of other adult diseases focuses on studies of populations (e.g., studies utilizing NHANES data) and
highly exposed communities (e.g., the C8 population). In these studies, the exposure assessment is based
on serum PFAS levels, either measured or predicted based on physiologically-based pharmacokinetic
(PBPK) models. We use the same classification scheme for the other adult diseases as was used above
for cancers.
1. Kidney function/kidney disease
Two studies of the C8 population were conducted to evaluate kidney disease or kidney function. The
first study of chronic kidney disease used both a retrospective and prospective longitudinal approach
(Dhingra 2016a). In the retrospective approach, there were 397 confirmed cases arising from a cohort of
32,254. Of the 397 cases, 187 were non-diabetic. In the prospective approach, the cohort was restricted
to those who were kidney disease-free at the baseline interview (2005-2006) and evaluated 212
confirmed cases (106 non-diabetic) that occurred after the baseline interview. Analyses were also
conducted restricting the cohort and the kidney disease cases to those without diabetes. The study
evaluated yearly modeled PFOA serum concentrations and cumulative exposure. In the full cohort, the
estimated hazard ratio (HR) in the top quintile of cumulative serum PFOA in the retrospective analysis
was 1.24 (95% CI: 0.88, 1.75), and the trend was not monotonic. In the prospective analysis, the HR in
the top quintile was 1.12 (95% CI: 0.72, 1.75) and the trend was also not monotonic. Similar findings
were obtained for the non-diabetic population. When exposures were lagged, the HRs were reduced.
In a second C8 study, the phenomenon of “reverse causation” was assessed in a cross-sectional
evaluation of impaired kidney function (estimated glomerular filtration rate, eGFR) and earlier
menopause (Dhingra 2016b). Self-reported menopause was evaluated among 9,192 women aged 30-65
and kidney function among 29,499 adults. Although there was a non-monotonic negative trend for eGFR
across measured serum PFOA quintiles (β ± S.E.= -0.98 ± 0.27 for the top quintile), neither modeled
serum PFOA nor modeled cumulative exposure showed associations with eGFR. This suggested that the
finding for eGFR was due to reverse causation, i.e., that reduced kidney function as measured by eGFR
caused the increased measured serum PFOA. This result would occur because reduced kidney function
slowed the excretion of PFOA. The study also found a significant increasing trend of reported early
menopause with increasing measured PFOA category, but when using modeled serum or modeled
cumulative exposure PFOA instead of measured, this trend disappeared. Again, this suggested reverse
causation, i.e., that early menopause caused the increased PFOA serum levels. The study found that
measured serum PFOA levels increased on average 4% per year for the first seven years after
menopause and then stopped increasing. he authors emphasized that caution is necessary when using
exposure biomarkers in cross-sectional studies.
78

A study using NHANES data for the 2003-2010 cycles evaluated estimated GFR and serum uric acid
among adolescents aged 12-19 years (Kataria 2015). PFOA and PFOS serum levels were associated
with lower eGFR and higher serum uric acid. However, given the findings in the C8 study, these
associations may, at least in part, be due to the phenomenon of reverse causation.
In summary, there is little clear evidence that PFAS affects kidney function or increases the risk of
kidney disease. However, because so few studies have been conducted, there is insufficient information
to determine whether PFAS exposures affect the kidney. Therefore additional research is needed. In
order to avoid the issue of reverse causation, future cross-sectional studies should not rely solely on
serum PFAS levels to assess exposures.
Three studies have evaluated uric acid in adults and PFAS exposures. In a C8 cross-sectional study of
54,951 adults, both PFOA and PFOS serum levels were associated with increased uric acid (Steenland
2010). Elevated uric acid (hyperuricemia) was also evaluated, and the OR in the top quintile was 1.47
(95% CI: 1.37, 1.58) with a monotonic trend. PFOS also was associated with hyperuricemia, but to a
lesser extent than PFOA. A second cross-sectional study used NHANES data for 1999-2000 and 20032006 (serum uric acid was not included in the 2001-2002 NHANES cycle) and found that PFOA and
PFOS were associated with hyperuricemia (Shankar 2011). Again, PFOA had the stronger relationship
with a fourth quartile OR of 1.97 (95% CI: 1.44, 2.70) and a monotonic trend compared to a fourth
quartile OR of 1.48 (95% CI: 0.99, 2.22) for PFOS and a non-monotonic trend. A third cross-sectional
study evaluated NHANES data for the 2007-2010 cycles and found that both PFOA and PFOS serum
levels were associated with increasing serum uric acid (Gleason 2015). However, the study found that
only serum PFOA had a monotonic increased risk of hyperuricemia.
Because these studies relied on serum PFAS levels to assess exposure and were cross-sectional, it is
possible that the phenomenon of reverse causation may explain at least part of these findings, i.e., that a
reduction in kidney function or chronic kidney disease causes hyperuricemia as well as increased PFAS
serum levels due to a reduction in excretion. Nevertheless, the studies consistently found an increased
risk of hyperuricemia associated with serum levels of PFOA, and to a lesser extent PFOS. Further
research is needed that supplements serum PFAS measurements with modeled serum PFAS estimates to
account for possible reverse causation.
2. Liver disease/liver function
Two C8 studies and one study that evaluated NHANES data have assessed PFAS exposures and liver
function. The first C8 cross-sectional study included 47,092 adults and measured alanine transaminase
(ALT), γ-glutamyltransferase (GGT) and direct bilirubin (Gallo 2012). Both PFOA and PFOS serum
levels were associated with increased ALT in linear models and with high ALT in logistic models. The
trends were not monotonic. The second C8 study evaluated liver disease among 32,254 adults including
3,713 DuPont workers and liver biomarkers among 30,723 adults including 1,892 DuPont workers
(Darrow 2016). The liver biomarker part of the study was cross-sectional. The study avoided the issue of
reverse causation by using modeled estimates of yearly serum levels of PFOA. Estimated cumulative
exposure and the estimated PFOA serum level during 2005-2006 were evaluated. PFOA was associated
with increasing levels of ALT and with abnormal ALT. PFOA was not associated with liver disease. An
earlier study conducted in the C8 study area by Emmett 2006 was limited by a small sample size (371
residents including 20 children ages 2-10 and 29 individuals aged 11-20). Another study using
NHANES data for the 1999-2006 cycles found no association with liver disease, with 4th quartile ORs
for PFOA and PFOS of 0.61 and 0.95, respectively (Melzer 2010).
79

A study using NHANES data for 2007−2010 found associations between serum PFAS and liver function
parameters (Gleason 2015). PFHxS was associated with increased ALT, PFOS was associated with
increased total bilirubin, PFOA was associated with increased ALT, GGT and total bilirubin, and PFNA
was associated with increased ALT. An earlier study using NHANES data for 1999-2004 found positive
associations between PFOA and ALT and natural log GGT and PFHxS and PFNA and total bilirubin
(Lin 2010). The association between PFOA and liver enzymes was more evident in obese subjects, as
well as subjects with insulin resistance and/or metabolic syndromes.
In summary, the two studies that evaluated PFAS exposures and liver disease found no association.
Consistent associations across three studies were found for PFOA and increased ALT. PFOS was also
associated with increased ALT in a C8 study. The C8 study that evaluated modeled estimated PFOA
serum levels avoided the issue of reverse causation and observed an association between PFOA and
increased ALT. Even though the number of studies is small, PFOA has consistently been associated with
increased ALT. Because few studies have been conducted, further research is needed to evaluate PFAS
exposures and liver function, supplementing serum PFAS measurements with modeled serum PFAS
estimates to account for the issue of reverse causation.

3. Coronary Artery Disease, hypertension, hypercholesterolemia (high cholesterol)
One C8 study evaluated incident coronary artery disease and hypertension and modeled PFOA serum
levels (Winquist and Steenland 2014b). There was no association with hypertension or coronary artery
disease – in both analyses, the HRs were higher for the lower quintiles than the higher quintiles, no HR
was higher than 1.26, and most HRs were ≤1.1. For coronary artery disease, the HR for the top quintile
of cumulative exposure to PFOA was 1.07 (95% CI: 0.93, 1.23) with a non-monotonic trend. For
hypertension, the HR for the top quintile of cumulative exposure to PFOA was 0.98 (95% CI: 0.91,
1.06).
A case-control study of coronary heart disease was conducted in Sweden with 253 cases and 253
matched controls (Mattsson 2015). The adjusted ORs for serum PFOS for the 3rd and 4th quartile were
1.30 (95% CI: 0.74, 2.26) and 1.07 (95% CI: 0.60, 1.92), respectively. The strongest finding in this
study was for perfluoroheptanoic acid (PFHpA) with adjusted ORs for the 3rd and 4th quartile of 2.58
(95% CI: 1.39, 4.78) and 1.73 (95% CI: 0.94, 3.16). All of the other PFAS had adjusted ORs in the 4th
quartile <1.0.
A study that evaluated NHANES data for 1999-2006 found a slight increase in heart disease for PFOA
(4th quartile OR=1.08, 95% CI: 0.70, 1.69) and no association with PFOS (4th quartile OR=0.91) (Melzer
2010). Another study that evaluated NHANES data (1999-2003) found PFOA associated with
cardiovascular disease (Shankar 2012). For the top quartile of PFOA, the OR for cardiovascular disease
was 2.01 (95% CI: 1.12, 3.60) with a monotonic trend. Elevated ORs were observed for PFOA and both
coronary heart disease and stroke but the trend was not monotonic, with 4th quartile ORs of 2.24 (95%
CI: 1.02, 4.94) and 4.26 (95% CI: 1.84, 9.89), respectively.
Several studies have evaluated hypercholesterolemia or serum lipids and have found consistent positive
associations with PFAS. In the C8 cross-sectional study of 46,294 adults, both PFOS and PFOA were
associated with increasing total cholesterol and LDL (Steenland 2009). The predicted increase in total
cholesterol from the lowest to highest decile of PFOS and PFOA was 11–12 mg/dL. For
hypercholesterolemia, the OR for the top quartile of PFOA was 1.38 (95% CI: 1.28, 1.50) with a
80

monotonic trend, while for PFOS, the OR was 1.51 (95% CI: 1.40, 1.64), with a monotonic trend. PFOA
was also associated with triglycerides. In a longitudinal study of 560 adults from the C8 population who
were followed for 4.4 years, individuals with the greater declines in serum PFOA and PFOS had greater
LDL decreases (Fitz-Simon 2013). For an individual whose serum PFOA fell by half, the predicted fall
in LDL was 3.6% (95% CI: 1.5%, 5.7%). A stronger finding was observed for PFOS: a decline in serum
PFOS by half was predicted to decrease LDL by 5% (95% CI: 2.5%, 7.4%). Incident
hypercholesterolemia was found to be associated with PFOA in a C8 study, especially among men aged
40-60. The HR for the top quintile of cumulative exposure was 1.44 (95% CI: 1.28, 1.62) although the
trend was not monotonic (Winquist and Steenland 2014b).
A study of cholesterol using NHANES data for the 2003-2004 cycle found positive associations for
PFOS, PFOA and PFNA but a negative association with PFHxS (Nelson 2010). The highest quartile of
PFOS exposure had total cholesterol levels 13.4 mg/dL (95% CI: 3.8–23.0) higher than in the lowest
quartile. For PFOA, PFNA, and PFHxS, effect estimates were 9.8 (95% CI, –0.2 to 19.7), 13.9 (95% CI,
1.9–25.9), and –7.0 (95% CI, –13.2 to –0.8), respectively. A study using 2007-2009 cross-sectional data
from the Canadian Health Measures Survey found a positive association between PFHxS and
hypercholesterolemia (Fisher 2013). For the top quartile of PFHxS, the weighted OR was 1.57 (95% CI:
0.93, 2.64) with a monotonic trend. The finding for PFHxS contradicted the NHANES study which
found that PFHxS was associated with a decline in cholesterol. In the Canadian study, the top quartile
for PFOA had an OR of 1.50 (95% CI: 0.86, 2.62) with a non-monotonic trend.
In a small study conducted in China, 133 individuals were evaluated for PFAS exposure and high serum
lipid levels (Fu 2014). The study did not evaluate PFHxS, but did evaluate PFOA, PFOS, PFNA,
perfluorodecanoic acid (PFDA) and perfluoroundecanoic acid (PFUdA). For high total cholesterol, there
was a monotonic trend for PFDA with the top quartile OR of 3.84 (95% CI: 0.87, 16.95). Nonmonotonic trends for high total cholesterol were observed for PFOS (4th quartile OR=2.27, 95% CI:
0.47, 10.92) and PFUdA (4th quartile OR=3.70, 95% CI: 0.76, 18.03). For high LDL, monotonic trends
were observed for PFOS (top quartile OR=2.27, 95% CI: 0.50, 10.37) and PFUdA (top quartile
OR=4.16, 95% CI: 0.96, 18.00). The wide confidence intervals in this study were due to the small
sample size.
A study of 891 pregnant women in Norway found a monotonic trend for PFOA and increasing total
cholesterol with a regression coefficient of 2.58 (95% CI: -4.32, 9.47) per natural log PFOA (ng/ml)
(Starling 2014). Non-monotonic trends were also observed for total cholesterol and PFOS and PFHxS
with regression coefficients (per natural log-ng/ml) of 8.96 (95% CI: 1.70, 16.22) and 3.00 (95% CI: 1.75, 7.76), respectively. Monotonic trends were observed for PFOS, PFHxS, and PFUdA and HDL, but
no monotonic trends were observed for PFAS and LDL. The strongest finding for LDL was for PFOS
with a regression coefficient of 6.48 (95% CI: -0.07, 13.03). Smaller effects were observed between
LDL and PFOA and PFHxS with regression coefficients of 2.25 (95% CI: -3.97, 8.48) and 1.92 (95%
CI: -2.50, 6.33), respectively. No associations were observed for triglycerides.
In summary, because of the small number of studies and conflicting findings, more research is needed to
evaluate whether PFAS exposures affect the risk of cardiovascular disease or hypertension. Several
studies have evaluated PFOA and PFOS and lipids and the findings consistently indicate that an
association exists with increased lipids.

81

4. Thyroid function/disease
A C8 study that evaluated thyroid disease in retrospective and prospective analyses included 32,254
adults aged ≥20 years (28,541 community members and 3,713 workers) who completed baseline
questionnaires in 2005-2006 and follow-up questionnaires during 2008-2010 and 2010-2011 (Winquist
and Steenland, 2014a). About 2/3 of the thyroid diseases were hypothyroidism. In the retrospective
analysis, the HR for the top quintile of cumulative exposure to PFOA and functional thyroid disease was
1.28 (95% CI: 1.06, 1.53) with a non-monotonic trend. The finding was much stronger in females with a
5th quintile HR of 1.37 (95% CI: 1.11, 1.68) and a monotonic trend compared to the results in males (5th
quintile HR=1.05, 95% CI: 0.66, 1.66, and no trend). PFOA was more strongly associated with
hypothyroidism with a HR for the top quintile of 1.40 (95% CI: 1.12, 1.75) and a non-monotonic trend.
Elevated HRs were observed for both males and females and hypothyroidism, but there was a monotonic
trend for females. Elevated HRs were observed for females and hyperthyroidism, but the trend was not
monotonic. HRs were not elevated for males and hyperthyroidism. Similar findings were observed when
the community cohort was evaluated separately. However there was a monotonic trend for females and
hyperthyroidism, and the HRs for males and hypothyroidism were higher than for females although the
trend was not monotonic.
In the prospective analysis, the HR for the top quintile of cumulative PFOA exposure and functional
thyroid disease was 1.12 (95% CI: 0.82, 1.52) with a non-monotonic trend. There was no association for
females, but there was a monotonic trend for males with an HR for the top quintile of 1.85 (95% CI:
0.93, 3.68). The highest HRs were for hyperthyroidism among males and females although based on
relatively small numbers, particularly among males, and the trends were not monotonic. For
hypothyroidism, there was no association among females, but there was a monotonic trend for males
with an HR for the top quintile of 2.02 (95% CI: 0.87, 4.65). Similar findings were observed when the
community cohort was evaluated separately. The authors concluded that there was an association
between PFOA exposure and thyroid disease, especially for hyperthyroidism among women in the
retrospective analyses and for hypothyroidism among men in the prospective analyses.
A study that evaluated NHANES data for 1999-2006 found associations between PFOA and PFOS and
reported ever had thyroid disease and reported current thyroid disease with medication (Melzer 2010).
None of the exposure-response trends were monotonic. For the 4th quartile PFOA, the ORs for ever had
thyroid disease were 1.68 (95% CI: 1.14, 2.49) for women and 1.50 (95% CI: 0.66, 3.39) for men. For
current thyroid disease with medication, the ORs for 4th quartile PFOA were 2.24 (95% CI: 1.38, 3.65)
for women and 2.12 (95% CI: 0.93, 4.82) for men. For the 4th quartile PFOS, the ORs for ever had
thyroid disease were 1.15 (95% CI: 0.78, 1.70) for women and 1.78 (95% CI: 0.58, 5.52) for men. For
current thyroid disease with medication, the ORs for 4th quartile PFOA were 1.27 (95% CI: 0.82, 1.97)
for women and 2.68 (95% CI: 1.03, 6.98) for men.
In summary, although only two studies have evaluated thyroid disease in adults and PFAS exposure,
both had positive findings. In particular, the C8 study found elevated HRs for thyroid diseases in the
prospective analyses. However, there also appears to be effect modification by gender. Because of the
few studies that have evaluated PFAS exposure and thyroid disease, more research is necessary, in
particular, studies designed to evaluate effect modification by gender.
Thyroid function biomarkers were evaluated in a C8 cross-sectional study that included 52,296 adults
with a year or more exposure to PFOA (Knox 2011). The biomarkers evaluated were thyroxine (T4), T3
uptake, and thyroid stimulating hormone (TSH). Both PFOA and PFOS were associated with an
82

elevation in serum thyroxine and a reduction in T3 uptake. Interactions between gender and PFOS were
observed for T3 uptake and thyroxine and between PFOA and gender for T3 uptake.
Three studies evaluated NHANES data and thyroid function. The first study evaluated NHANES data
for 2007-2008 and found that TSH and TT3 levels increased with PFOA and TT4 levels increased with
PFHxS (Jain 2013). This study is not included in Table A3 because no confidence intervals were
presented, the age range included adolescents (i.e., ages 12-18), and these NHANES data were evaluated
by a second study that evaluated NHANES data for 2007-2010 (Wen 2013). The latter study found
mixed results by gender (Wen 2013). PFHxS was associated with an increase in TT4 for women (β =
0.26, 95% CI: 0.11, 0.41) but not for men (β = -0.03, 95% CI: -0.18, 0.64). On the other hand PFHxS
was associated with a decline in T4 among men (β = -0.016, 95% CI: -0.029, -0.003) but not for women
(β = 0.003, 95% CI: -0.024, 0.030). PFHxS was associated with an increase in TT3 for women β = 4.07,
95% CI: 2.23, 5.92) but not for men (β = -0.08, 95% CI: -1.70, 1.56). PFOA was associated with TT3
among women (β = 6.63, 95% CI: 0.55, 12.72) but not for men (β = 0.78, 95% CI: -3.05, 4.60).
The third NHANES study evaluated NHANES data for 2007-2008 but focused on potential susceptible
subgroups with thyroid “stressors”, i.e., with low iodine status, with high thyroid peroxidase antibody
(TPOAb) or with both (Webster 2016). The key findings were that all 4 PFAS evaluated (PFOA, PFOS,
PFHxS and PFNA) were associated with increased fT3, increased fT3/fT4, increased TSH, and
increased TT3 in the group with joint exposure to high TPOAb and low iodine. PFOS and PFHxS were
also associated with decreased fT4 in the group with high TPOAb and low iodine. The findings were
considerably weaker for those with normal iodine and TPOAb status and those with either low iodine
alone or high TPOAb alone.
A group of 87 men and women residing in the upper Hudson River area of New York who were
originally recruited for a study of PCB exposure were evaluated for PFOA and PFOS exposure and
thyroid function (Shrestha 2015). Natural log PFOS was positively associated with fT4 (β=0.054, 95%
CI: 0.002, 0.106) and T4 (β = 0.766, 95% CI: 0.327, 1.205), which corresponded to 4% and 9%
increases in fT4 and T4 per interquartile range difference in PFOS. A positive association also was
observed for log PFOS and TSH (β = 0.129, 95% CI: -0.023, 0.281). An association was also observed
for natural log PFOA and T4 (β = 0.380, 95% CI: -0.070, 0.830). When both PFOS and PFOA were
included in the models, the association between PFOS and T4 persisted, but the association between
PFOS and fT4 was attenuated.
A cohort of 633 individuals aged >12 years from Siheung, Korea was evaluated for PFAS exposure and
thyroid function (Ji 2012). Slight declines in T4 (e.g., for PFOS, β = -0.021, 95% CI: -0.048, 0.005) and
slight increases in TSH (e.g., for PFNA, β = 0.110, 95% CI: -0.035, 0.255) were observed. The strongest
findings were for perfluorotridecanoic acid (PFTrDA) and decreased T4 and increased TSH.
In summary, based on findings from these studies, there appears to be gender differences in the effects
of PFAS exposures on thyroid function. This is not surprising given the evidence of effect modification
by gender for the associations between PFOS and PFOA and thyroid diseases. In general, TSH and TT4
increased with PFOA, PFOS and PFHxS exposure. TT3 also appeared to increase but not in the C8
study. A susceptible population with low iodine status and high TPOAb status was identified in one
study. Although several studies have been conducted of PFAS exposure and adult alterations in thyroid
function, there are inconsistencies in the findings for TT3 and TT4. Additional research is needed to
resolve these inconsistencies. In particular, studies should be designed to evaluate effect modification by
gender as well as by vulnerable subpopulations such as those who have thyroid “stressors”.
83

Several studies evaluated thyroid function among pregnant women primarily because of the potential
effect on the fetus of maternal thyroid hormone disruption. A study in Taiwan of 285 women in their
third trimester observed the strongest findings for PFNA, perfluorododecanoic acid (PFDoDA), and
perfluoroundecanoic acid (PFUnDA) and decreased maternal free T4 and TT4. The effects of PFHxS on
free T4 and TT4 were considerably weaker than these three PFAS but much stronger than PFOA or
PFOS (Wang 2014). In a Canadian study of 152 women, those with normal TPOAb had little effects
from PFAS exposure on fT4, TT4 or TSH (Webster 2014). However, women with high TPOAb had
increases of 46% to 69% in TSH for interquartile range increases of PFOA, PFOS, and PFNA.
(Interquartile range increase in PFHxS was associated with only a 2% increase in TSH in these women.)
All four PFAS evaluated were associated with a 3% − 7% decrease in fT4 among women with high
TPOAb. No associations were found for TT4.
In a study of 375 women in Norway, 4th quartile PFOS levels were associated with a 0.35 (95% CI: 0.21,
0.50) mean difference in TSH, corresponding to a 24% higher mean concentration of TSH (Berg 2015).
A multipollutant assessment of persistent organic pollutants (POPs) including PCBs, DDT,
hexachlorobenzene, and PFAS was conducted with a similar group of Norway women, and PFOS was
again found to be associated with increased TSH with a monotonic trend controlling for other POPs
(Berg 2017). A study of 392 women from Hokkaido, Japan found a negative correlation between PFOS
and TSH and a smaller positive correlation with fT4 (Kato 2016). There was little correlation between
PFOA and TSH or fT4.
In general, the studies of maternal thyroid function and PFAS exposure suggested associations between
PFAS and increased TSH. For the other thyroid biomarkers, the results were mixed.
5. Autoimmune diseases
A C8 study evaluated autoimmune diseases retrospectively and prospectively in a cohort of 32,254
adults including 3,713 workers (Steenland 2013). Self-reported autoimmune diseases were confirmed
via medical records. For the retrospective analyses, the follow-up was from 1952 through the interviews
conducted in 2008-2011. For the 4th quartile cumulative PFOA exposure, the RRs for ulcerative colitis
for unlagged and 10-year lagged exposures were 2.86 (95% CI: 1.65, 4.96) and 3.05 (1.56, 5.96),
respectively. The trends for ulcerative colitis were monotonic. For multiple sclerosis, the 4th quartile
exposure RRs, unlagged and 10-year lagged, were 1.26 (0.65, 2.42) and 1.32 (0.61, 2.84), respectively,
but the trends were not monotonic. For rheumatoid arthritis, an elevated RR for 4th quartile PFOA
exposure was observed only for the 10-year lagged exposure (RR=1.35, 95% CI: 0.87, 2.11), but the
trend was not monotonic. In the prospective analyses, autoimmune cases that occurred between the
baseline 2005-2006 interviews and the 2008-2011 follow-up interviews were too few to evaluate
multiple sclerosis, lupus, type 1 diabetes or Crohn’s disease.
For ulcerative colitis, the RRs for the 4th quartile cumulative PFOA exposure, unlagged and 10-year
lagged, were 1.62 (95% CI: 0.57, 4.61) and 1.51 (95% CI: 0.43, 4.30). The wide confidence intervals
were due to the small number (n=30) of cases that occurred during the follow-up period. The trends
were not monotonic. The RRs for rheumatoid arthritis for the 4th quartile PFOA cumulative exposure
were <1.0.
On the basis of this study, the C8 Science Panel decided that there was a probable link between PFOA
exposure and ulcerative colitis. Of note, the RRs were elevated in both the retrospective and prospective
analyses. A study of ulcerative colitis conducted of the 3,713 workers at the DuPont West Virginia plant
84

found a RR of 6.57 (95% CI: 1.47, 29.4) for the top quartile of cumulative PFOA exposure and a 10year exposure lag. The trend was monotonic and the findings were based on 28 total cases. For
rheumatoid arthritis and no exposure lag, the RR for the top quartile was 4.45 (95% CI: 0.99, 19.9) with
a monotonic trend based on 23 total cases (Steenland 2015, see Table A2).
The C8 studies indicate an association between PFOA exposure and ulcerative colitis. For the other
autoimmune diseases, the information is inadequate to determine whether PFOA is associated with
increased risk. For the other PFAS, there is a lack of information on the risk for autoimmune diseases.
Research is needed to determine whether PFAS exposures increase the risk of autoimmune diseases,
especially since there is toxicological evidence that PFAS exposures affect the immune system (NTP
2016).
6. Osteoarthritis, osteoporosis and bone mineral density
Two studies evaluated osteoarthritis. In a C8 study, 49,432 adults were included and 3,731 reported a
physician diagnosed case of osteoarthritis in the baseline survey (Innes 2011). For the top quartile of
serum PFOA levels, the OR for osteoarthritis was 1.42 (95% CI: 1.26, 1.59) with a monotonic trend. No
association (ORs < 1.00) was observed for serum PFOS. Higher ORs were observed for PFOA and
osteoarthritis among those aged <55 years and those who were not obese. In the NHANES study, data
for 2003-2008 were evaluated (Uhl 2013). For PFOA, the OR for the top quartile was 1.55 (95% CI:
0.99, 2.43) with a non-monotonic trend. The elevated ORs were entirely due to the effect in females,
since the ORs for males were <1.00. For the top quartile of serum PFOS, the OR was 1.77 (95% CI:
1.05, 2.96) with a non-monotonic trend. Elevated ORs were observed in both males and females.
Two studies evaluated bone mineral density using NHANES data. The first study used NHANES data
for 2005-2008 to evaluate serum PFOS and PFOA and lumbar spine and hip bone mineral density
among men, women in menopause and premenopausal women (Lin 2014). Self-reported fractures were
also evaluated. A unit increase in natural log serum PFOS was associated with a decrease in total lumbar
spine bone mineral density by 0.022 g/cm2 (95% CI: -0.038, -0.007) in women not in menopause, but
the trend was not monotonic and the decline was only observed among those with PFOS serum levels
>75th percentile. No associations were observed for PFOA, and no associations were observed for total
hip bone mineral density. For all types of self-reported fractures, the ORs for women in menopause were
1.53 (95% CI: 0.63, 3.74) for PFOA and 1.59 (95% CI: 0.88, 2.86) for PFOS. No associations were
found for all types of fractures among men or premenopausal women. Premenopausal women had
elevated ORs for PFOA and hip fracture (OR=1.59, 95% CI: 0.57, 4.46) and spine fracture (OR=1.83,
95% CI: 0.59, 5.61). Men had an elevated OR for PFOA and spine fracture (OR=1.54, 95% CI: 0.85,
2.79).
The second NHANES study used data for 2009-2010 to evaluate bone mineral density and osteoporosis
(Khalil 2016). Both sexes had declines in femur bone mineral density for each of the PFAS evaluated
(PFOA, PFOS, PFHxS and PFNA). The strongest association was for PFOS among postmenopausal
women (natural log-PFOS β = -0.033, 95% CI: -0.049, -0.015). For femur neck mineral density, declines
were seen for PFOS, PFHxS and PFNA in both sexes and for PFOA among premenopausal women. The
strongest association was for PFOS among postmenopausal women (natural log-PFOS β = -0.033, 95%
CI: -0.049, -0.017). For lumbar spine bone mineral density, declines were observed for PFOA and PFOS
among men and postmenopausal women, PFNA and both sexes, and PFHxS and postmenopausal
women. The strongest association was for PFNA among postmenopausal women (natural log-PFNA β =
-0.043, 95% CI: -0.073, -0.013). For osteoporosis, the 4th quartile ORs for PFOA, PFOS, PFHxS and
85

PFNA were 2.59 (95% CI: 1.01, 6.67), 1.07 (95% CI: 0.36, 3.19), 13.20 (95% CI: 2.72, 64.15) and 3.23
(95% CI: 1.44, 7.21), respectively. None of the trends were monotonic except possibly for PFOA where
the ORs for the 2nd and 3rd quartiles were essentially the same and could be considered monotonic.
In summary, some positive findings were observed in the two studies that evaluated bone mineral
density and the two studies that evaluated osteoarthritis. Only one study evaluated osteoporosis and
positive findings were observed as well. Because only a few studies have been conducted, additional
research is needed.
7. Immune Response
Four studies have evaluated immune response or immune biomarkers. A report by the C8 Science Panel
on a cross-sectional study that has not yet been published evaluated immune biomarkers such as IgG,
IgM, IgA, IgE, total antinuclear antibodies (ANA) and C reactive protein (CRP) among the C8 adult
population (C8 Science Panel 2009). The study included 56,315 adults. The Panel reported that “several
statistically significant associations between levels of immunoglobulins and C8 were found: For IgA the
pattern of association indicated a significant decreasing trend with increasing PFOA; this was also
apparent for IgE but only in females. For IgG there was not a consistent trend with PFOA. ANA shows a
positive significant relationship with increasing PFOA. CRP showed a strong downward trend with
increasing PFOA,”
A second C8 study evaluated influenza vaccine response in 403 adults who did not have influenza
within the last 3 months and who provided pre- and post-vaccination blood samples to determine virusspecific antibody titers (Looker 2014). Associations between PFOA and PFOS and self-reported
influenza and colds in the past 12 months as reported on questionnaires (n = 755) were also assessed.
Elevated PFOA serum concentrations (4th vs 1st quartiles of exposure) were associated with reduced
antibody titer rise which may correlate with an increased risk of not attaining the antibody threshold
considered to offer long-term protection. Small negative associations (regression coefficients for the
geometric mean antibody titer ranging from -0.03 to -0.22) were observed comparing the highest and
lowest quartiles of PFOA and antibody titer rise and ratios for influenza B and Influenza A/H3N2 and
antibody titer rise for Influenza A/H1N1; there was a monotonic exposure response relationship for
H1N1. For PFOS, small negative associations were observed comparing the highest and lowest quartiles
and antibody titer rise and ratios for Influenza A/H3N2 (-0.04 and -0.03, respectively). People exposed
to the highest quartile PFOA and PFOS were less likely to seroconvert following vaccinations for
Influenza B (ORs = 0.71 and 0.87, respectively). Additionally, people exposed to the highest quartile of
PFOS were less likely to seroconvert following vaccinations for Influenza A/H1N1 (OR = 0.94) and
people exposed to the highest quartile of PFOA were less likely to seroconvert following vaccinations
for Influenza A H3N2 (OR = 0.62). OR for cold or flu ranged from 1.09-1.20 for people exposed to the
highest quartile of PFOS.
An exploratory study measured serum-PFAS concentrations in 12 adults whose antibody responses were
followed for 30 days after a booster vaccination with diphtheria and tetanus (Kielsen 2016). Participants
were healthy volunteers from a hospital in Denmark. Diphtheria antibody concentrations postvaccination were decreased by 8.2 to 18.2% for a doubling of exposures to several PFASs, including
PFOA, PFOS, and PFHxS. Tetanus antibody concentrations were decreased by 4.4 to 10.8% for a
doubling of exposures to several PFASs, including PFOS and PFHxS (but not PFOA). The authors note
that “serum PFAS concentrations showed significant negative associations with the rate of increase in
86

the antibody responses.” The authors also noted that this effect was particularly strong for the longerchain PFAS such as PFDA and PFNA.
Another exploratory study included 78 adults and evaluated the immune response to vaccination with
FluMist intranasal live attenuated influenza vaccine and PFAS exposure (Stein 2016b). Between 9% and
25% of the adults seroconverted after vaccination. PFAS exposure was associated with seroconversion
but the small numbers that had seroconverted resulted in extremely wide confidence intervals. The
strongest association observed between PFAS and immune marker response was for PFHxS and lower
mean interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) levels. The second and third tertile
regression coefficients for PFHxS and IFN-γ were -40 (95% CI: -76, -3.7), and -40 (95% CI: -84, 2.69).
For TNF-α, the second and third tertile regression coefficients were -5.3 (95% CI: -9.2, -1.3) and -4.8
(955 CI: -9.4, -0.10). However the authors concluded that the findings did not support an association
between PFAS exposure and reduced immune response to FluMist vaccination, although the study had
severe limitations including small sample size and the limited antibody response to FluMist.
Although there were positive findings in the four studies that have been conducted, each study had
serious limitations. Therefore additional research is needed with improved study designs and sufficient
statistical power to evaluate PFAS exposures and immune response.

8. Reproductive outcomes
Table A4 provides details on the studies of subfertility and infertility. Six studies evaluated subfertility
(time to pregnancy [TTP]) and/or infertility. Positive associations were found for subfertility/delayed
TTP for PFOA in three studies (Bach 2015a, Fei 2009, Whitworth 2012a) and PFOS in four studies
(Bach 2015a, Fei 2009, Jorgenson 2014, Whitworth 2012a), and slightly positive associations were
found for PFOA, PFHxS, and PFOS in one study (Velez 2015) and for PFNA in one study (Jorgenson
2014). Positive associations were found for infertility and PFOA in two studies (Fei 2009, Velez 2015),
for PFOS in two studies (Fei 2009, Jorgenson 2014), and for PFHxS and PFNA in one study each
(Jorgenson 2014, Velez 2015), and slightly positive associations were found for PFOS and PFOA in one
study each (Jorgenson 2014, Velez 2015). An additional study assessed adult male semen quality,
testicular volume, and reproductive hormone levels in men who were exposed in utero to PFOS and
PFOA and found positive associations with these chemicals and outcomes (Vested 2013).
A systematic review assessed fertility by evaluating studies on reproductive hormones (10 studies in
men and 3 in women) and TTP (2 studies in men and 8 in women) as well as nine studies of semen
characteristics (Bach 2016a). In men, there were inconsistent results across PFAS studies of semen
volume, sperm concentration, total sperm count, motility, and morphology; levels of testosterone, free
androgen index/free testosterone, estradiol, SHBG, LH, FSH, and inhibin B; and TTP. In women,
studies showed mostly positive associations for PFOS and PFOA and infertility and fecundability in
parous women, but were inconsistent for PFAS and reproductive hormones. The review concluded that
there was not strong evidence for an association between PFAS exposures and reproductive outcomes in
men or women. Another review of the epidemiological literature on reproductive outcomes came to a
similar conclusion (Khalil 2015).
Table A5 provides details on the studies of pre-eclampsia and pregnancy-induced hypertension. Five
studies evaluated pre-eclampsia. Positive associations were found for pre-eclampsia and PFOA (Savitz
2012a) and PFOS (Stein 2009) and slightly positive associations were found with PFOA in two studies
87

(Avanasi 2015, Nolan 2010) and PFOS in one study (Starling 2014). Three studies evaluated pregnancyinduced hypertension (PIH). Positive associations were found for PIH and PFOA in three studies
(Darrow 2013, Nolan 2010, Savitz 2012b) and PFOS in one study (Darrow 2013). The C8 Science Panel
concluded that there was a probable link between PFOS exposure and pregnancy-induced hypertension.
Two studies evaluated PFAS exposure and endometriosis. The first study included a sample of 495
women aged 18-44 scheduled for laparoscopy/laparotomy in clinics located in Salt Lake City and San
Francisco (Buck Louis 2012). The second sample was a population-based sample of 131 women
matched to the first sample on age and residence within a 50-mile radius of the participating clinics.
Forty-one percent (n=190) of the women scheduled for laparoscopy (“operative sample”) had newly
diagnosed endometriosis and 11% (n=14) of the population-based sample had endometriosis. The ORs
for PFOS, PFOA, PFHxS and PFNA among the operative sample were 1.25 (95% CI: 0.87, 1.80), 1.62
(95% CI: 0.99, 2.66), 0.85 (0.42, 1.73), and 1.99 (95% CI: 0.91, 4.33). Among those cases with stage 3
or stage 4 disease, the ORs for PFOS, PFOA, PFHxS and PFNA were 1.50 (95% CI: 0.82, 2.74), 1.86
(95% CI: 0.81, 4.24), 1.24 (0.47, 3.31), and 0.99 (95% CI: 0.27, 3.65).
The second study evaluated NHANES data for 2003-2006 (Campbell 2016). Seven percent of the
sample of 753 women reported physician-diagnosed endometriosis (n=54). Fourth quartile ORs for
PFOA, PFOS, PFHxS and PFNA were 2.86 (95% CI: 0.63, 12.91), 3.48 (95% CI: 1.00, 12.00), 1.47
(95% CI: 0.40, 5.41), and 3.24 (95% CI: 0.81, 12.91). None of the trends were monotonic.
Both studies of PFAS exposures and endometriosis had positive findings. However, since only two
studies have been conducted, more research is needed to determine whether PFAS exposures increase
the risk of endometriosis.

Summary of the Literature Review for adult diseases
For most adult diseases there is little or no information on the effects of exposures to PFHxS. Although
there is more information for PFOS, it is still inadequate to determine whether exposures increase the
risk for most of the adult diseases. PFOA has the most information, primarily because of the C8 studies.
Still additional research is needed to determine whether PFOA exposures increase the risk of several
adult cancers and non-cancers including colorectal cancer, multiple myeloma, kidney function/kidney
diseases, liver function, autoimmune diseases other than ulcerative colitis, and immune function.

Health Effects of PFAS in Children
Tables A6 and A7 provides details of studies of adverse birth outcomes and congenital malformations.
Table A8 provides details of studies of other adverse outcomes in children aged ≥2 years.
1. Adverse birth outcomes
Table A6 provides details on the studies that evaluated adverse birth outcomes. Twenty studies
evaluated birth weight and the results are presented in Table A6. Additionally, two meta-analysis of
birthweight found an overall decrease in birthweight associated with PFOA and PFOS (Verner 2015,
Bach 2015b). Ten studies evaluated preterm birth. Five studies evaluated small for gestational age
(SGA). Ten studies evaluated birth length, seven evaluated head circumference, one evaluated
88

abdominal circumference, and eight evaluated gestational age as a continuous variable. As evident from
Table A6, virtually all of the studies of these adverse birth outcomes evaluated PFOA and PFOS but
only a minority evaluated PFHxS.
Table A7 presents the results of five studies that evaluated birth defects. Positive associations with
PFOA and club foot, heart defect, and circulatory defect were found in one study with very small
numbers of cases (Nolan 2010). One study found a positive association between PFOS and PFOA
(slight) and cryptorchidism (Jensen 2014) while another study found no associations with
cryptorchidism and hypospadias (Toft 2016). One study relying on maternally-reported cases found
positive associations with PFOA and defects of the brain, limb, eye, and heart (Stein 2014a).
A study on congenital hypothyroidism found a “large” difference in PFOA concentrations between cases
and controls and mean concentrations of PFOA and PFNA in cases were “significantly higher” than in
controls, but the study was based on small numbers (Kim 2016).
One study evaluated cerebral palsy (Liew 2014). Using case-cohort sampling of the Danish National
Birth Cohort during 1996-2002, the study evaluated 156 cases and 550 controls. Maternal serum PFAS
were associated with cerebral palsy for boys, in particular PFOS (RR per ln unit increase=1.7, 95% CI:
1.0, 2.8) and PFOA (RR=2.1, 95% CI: 1.2, 3.6). However risks were not elevated in girls except for
PFHxS (RR=1.1, 95% CI: 0.6, 1.9).
Five studies evaluated miscarriage and two studies evaluated stillbirth. Positive associations were found
for miscarriage and PFOS in two studies (Jensen 2015, Darrow 2014); PFNA, PFDA, and PFHxS in one
study (Jensen 2015); and PFOA (slight) in one study (Darrow 2014). No associations were found for
stillbirth.
Because of inconsistencies in the findings across studies, more research is needed to evaluate the effect
of PFAS exposures on birth weight, SGA, head circumference and other fetal growth parameters,
reduced gestational age, and preterm birth. Few studies have been conducted of PFAS exposures and
miscarriage, stillbirth, birth defects, congenital hypothyroidism or cerebral palsy, therefore additional
research is necessary with improved study designs and sufficient statistical power.
2. Lipids
Table A8 presents the results of the five studies, including a C8 study, that evaluated lipids (Frisbee
2010, Nelson 2010, Geiger 2014a, Maisonet 2015a, Zeng 2015). All five studies found increases in
lipids with increasing exposures to PFOS and/or PFOA. In one NHANES study (Nelson 2010), PFHxS
was associated with increased lipids. Overall, the findings of increased lipids from exposures to PFAS
has been consistent.
3. Thyroid function
There have been several studies conducted of infants and most have observed that prenatal PFAS
exposures disrupted thyroid function. Only two studies have been conducted of older children, and their
results are presented in Table A8. The C8 study found increases in a thyroid hormone (TSH [thyroid
stimulating hormone]) with increasing serum levels of PFOS and PFOA (Lopez-Espinosa 2012). PFHxS
was not evaluated. PFOA, but not PFOS, was associated with an increased risk of thyroid disease.
89

In a Taiwan study (Lin 2013), findings were inconsistent for PFOS and PFOA and thyroid hormones
when boys and girls were evaluated separately (effect was stronger in males). PFHxS was not evaluated.
Because of the few studies that evaluated PFAS exposures and thyroid function among children older
than infants, more research is needed.
4. Uric acid
Table A8 presents the results of studies that evaluated uric acid. A study in Taiwan found elevated uric
acid levels associated with PFOA, PFHxS, and PFOS (Qin 2016). Two studies that evaluated NHANES
data found elevations in uric acid and in the risk of hyperuricemia for PFOA and PFOS (Geiger 2013,
Kataria 2015). More research is needed to follow up these findings. Because these studies were crosssectional, there is concern about the possibility of reverse causation (e.g., impaired kidney function
could cause both elevated uric acid and a reduction in the elimination of PFAS via the kidney resulting
in higher serum PFAS levels). Future studies should attempt to predict serum PFAS levels and/or
evaluate this outcome longitudinally (prospectively).
5. Sex hormones
Four studies evaluated sex hormones and are presented in Table A8. The C8 study found declines for
testosterone in boys and girls with increasing serum levels of PFOA and PFOS and decline in boys with
increasing serum levels of PFHxS (Lopez-Espinosa 2016). In the larger of the two Taiwan studies,
declines in testosterone levels were observed in both sexes with increasing serum levels of PFOA, in
boys only with increasing serum levels of PFOS, and in girls only with increasing serum levels of
PFHxS (Zhou 2016). On the other hand, a study of girls in the UK observed increases rather than
decreases in testosterone with increasing serum levels of PFOS, PFOA and PFHxS (Maisonet 2015b).
Overall, there is some evidence that PFAS exposure may decrease testosterone levels, but the findings
have not been consistent across the few studies that have been conducted. More research is needed to
determine whether and how PFAS exposures affect sex hormone levels.

6. Delay in reaching puberty
Three studies evaluated delays in reaching puberty and are presented in Table A8. In the C8 study, both
PFOA and PFOS were associated with delays in puberty (Lopez-Espinosa 2011). PFHxS was not
evaluated. In a study conducted in Denmark, PFOS and PFOA were also associated with delay in
reaching puberty (Kristensen 2013). However, a study conducted in the UK found that PFOA was
associated with an earlier age at puberty while PFOS was associated with delayed puberty, and the
results were conflicting for PFHxS (Christensen 2011). More research is needed to evaluate whether
PFAS exposure can cause delays in reaching puberty.
7. Neurobehavioral outcomes
Neurobehavioral outcomes are presented in Table A8. Two studies evaluated IQ. The C8 study found
only slight differences in IQ and the results were not consistent for PFOA, the only PFAS evaluated
(Stein 2013). A study conducted in Taiwan found deficits in IQ for PFOS, but not for PFOA or PFHxS
(Wang 2015).
90

There is some evidence from the C8 study (Stein 2011) and studies conducted in Denmark (Liew 2015),
Sweden (Ode 2014), and using NHANES data (Hoffman 2010) that PFAS may be associated with
ADHD. The C8 study found no association for PFOA, but elevated risks for PFOS and PFHxS with
odds ratios (ORs) of 1.3 and 1.6, respectively (Stein 2011).
The C8 study found a slight increase in risk (OR=1.2) for higher exposures to PFHxS and learning
problems but no associations for PFOS or PFOA (Stein 2011). A recent study that evaluated measures of
executive function of “clinical relevance” found elevated risks especially for PFOS and PFHxS (Vuong
2016). However, in general, the effects observed have not been large for neurobehavioral outcomes.
Evaluating the evidence for associations between PFAS exposures and IQ, ADHD, and other
neurobehavioral outcomes is hampered by different methods for ascertaining ADHD, different methods
for testing IQ, lack of consistency in the other neurobehavioral outcomes evaluated, and the small
number of studies that have been conducted. Therefore additional research is necessary to determine
whether PFAS exposures are associated with adverse neurobehavioral outcomes in children such as IQ,
depression, deficits in executive function, ADHD, and developmental delay,

8. Immune function
Few studies have been conducted to evaluate immune function and PFAS exposure. These studies are
presented in Table A8. Three studies have evaluated whether PFAS exposures suppress the antibody
response to specific vaccines, but only two of these studies evaluated the same vaccine, i.e., rubella.
Both of these studies found deficits in serum rubella antibody response (Granum 2013, Stein 2016). The
studies in the Faroes have evaluated tetanus and diphtheria longitudinally and found deficits in antibody
to these vaccines (Grandjean 2012, 2016).
Asthma was evaluated in three studies, with strong risks found in a Taiwan case-control study (Dong
2013) but considerably weaker risks found in two NHANES studies (Humblet 2014, Stein 2016). Other
outcomes such as atopic dermatitis and infectious diseases such as gastroenteritis and the common cold
were not evaluated in more than one study.
A systematic review of the evidence for immunotoxicity associated with exposures to PFOA and PFOS
conducted by NTP concluded that these exposures alter immune function in humans but that the
epidemiological evidence was too limited to conduct meta-analyses. Issues include the heterogeneity of
the studies and the small number of studies that evaluated the same outcome. NTP concluded that more
research is needed to evaluate the same vaccines and hypersensitivity-related outcomes in children
across different populations using similar research methods.
9. Hypertension and adiposity
Studies of hypertension and adiposity are presented in Table A8. One study has evaluated hypertension
in children using NHANES data and found no elevation in risk (Geiger 2014b). Three studies have
evaluated adiposity in children, adolescents and/or young adults. In one study, an association between
PFAS and measures of adiposity was found only in girls (Mora 2016). A second study (Karlsen 2016)
found slightly elevated risks for being overweight except for PFOA among children aged 5 years where
a stronger risk was observed (OR = 1.88, 95% CI: 1.05, 3.35). In a third study, PFOS was found to be
associated with measures of adiposity (Domazet 2016). Additional research is needed to determine
whether PFAS exposures increase the risk of hypertension or adiposity in children.
91

Summary of the Literature Review for childhood diseases
For most adverse outcomes in children evaluated in this assessment, the information was inadequate to
determine whether children exposed to PFAS were at increased risk. In particular, very few studies have
been conducted of PFHxS exposures, the PFAS that was considerably elevated in the serum of the
children tested at Pease. Additional research is needed for PFAS exposures and adverse birth outcomes;
thyroid, liver, kidney and immune function; uric acid; sex hormones; delays in reaching puberty; ADHD
and other neurobehavioral outcomes; hypertension; and adiposity.

92

Description of sample size calculations
Sample size calculations were conducted using OpenEpi Version 3.03. (Dean AG, Sullivan KM, Soe
MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health, www.OpenEpi.com, updated
2014/09/22). For some health-related endpoints, calculations could not be conducted because of a lack
of information in the studies on the parameters needed to make the calculations.
Sample size calculation for mean difference:
N1 = (variance of group 1 + variance of group 2 / (N2/N1)) (Z1-α/2 + Z1-β)2
(Mean difference)2
Where N2/N1 is the ratio of the two sample sizes. Then N2 is simply this ratio multiplied by N1. For a
type 1 error (or α error) of .05, the Z1-α/2 value is 1.96. This calculation is for a two-tailed hypothesis test
and equivalent to using a 95% confidence interval to determine statistical significance. For a one-tail test
with α =.05, the Z1-α/2 in the above equation is replaced by Z1-α and its value is 1.65, equivalent to using
a 90% confidence interval to determine statistical significance. The Z1-β in the above equation is the Z
value for the selected power. For 80% power, Z1-β = 0.84, for 90% power, Z1-β = 1.28, and for 95%
power, Z1-β = 1.65. (See Rosner B. Fundamentals of Biostatistics, 7th Edition, equation 8.27, p. 302).
The sample size calculations for odds ratios, risk ratios, etc. are as follows:
The sample size formula without the correction factor by Fleiss is:

For the Fleiss method with the correction factor, take the sample size from the uncorrected sample size
formula and place into the following formula:

When the input is provided as an odds ratio (OR) rather than the proportion of exposed with disease,
the proportion of exposed with disease is calculated as:

When the input is provided as a risk (or prevalence) ratio (RR) rather than the proportion of exposed
with disease, the proportion of exposed with disease is calculated as:

Fleiss JL. Statistical Methods for Rates and Proportions. John Wiley & Sons, 1981.
93

Sample size calculations for the mean difference in an effect biomarker between the exposed and
unexposed groups used the following formula:

Bernard Rosner. Fundamentals of Biostatistics (7th edition). Brooks/Cole, Boston 2011 (equation 8.27, p.
302)

Note: In some studies, the standard deviation is not presented but instead, the interquartile range (IQR)
is given. Assuming a normal distribution for the outcome under evaluation (e.g., thyroid function
measures), the standard deviation can be calculated by dividing the IQR range by 1.35. However if the
outcome is not normally distributed, this formula could underestimate the standard deviation. In
particular, if the outcome under evaluation has been log-transformed presumably to achieve a normal
distribution, the untransformed outcome is unlikely to have a normal distribution. Therefore, using this
formula when the outcome does not have a normal distribution may underestimate the SD by as much as
20% according to simulations conducted in Wan X. 2014. A higher SD would increase the sample size
requirement.

94

Children’s Study
The following provides information on the parameters (e.g., standard deviation, disease prevalence) used
in the sample size calculations provided in Tables 6a-c for the children study.

Lipids
In the C8 study (Frisbee 2010), the mean total cholesterol level in the study population was 160.7 mg/dL
and the standard deviation (SD) was 29.3. The sample size calculations assumed the same SD in the
Pease children and the unexposed group. For hypercholesterolemia (total cholesterol ≥ 170 mg/dL), the
prevalence in the C8 study was 34.2%.
Uric Acid
In the NHANES study (Geiger 2013), the mean uric acid level in the study population was 5.07 mg/dL
with a SD of 1.19. The sample size calculations assumed the same SD in the Pease children and the
unexposed group. The prevalence of hyperuricemia (uric acid ≥ 6 mg/dL) in the NHANES study was
16%.
Kidney Function
The mean estimated glomerular filtration rate (eGFR) in the C8 study of children and adolescents
(Watkins 2013) was 133 mL/min/1.73 m2 with a SD of 23.9. The sample size calculations assumed the
same SD in the Pease children and the unexposed group.

Attention Deficit/Hyperactivity Disorder (ADHD)
In the C8 study (Stein 2011), the prevalence of participant-reported ADHD was 12.4% and the
prevalence for participant-reported + used medications for ADHD was 5.1%. Sample size calculations
used the 12.4% prevalence. (Using the 5.1% prevalence would require much larger sample sizes.)

Hypersensitivity-related Outcomes
From an NHANES study (Stein 2016), the prevalences of current asthma and rhinitis among those aged
12-19 were 10.9% and 25.6%, respectively. For atopic dermatitis, the prevalence for children and
adolescents (ages 5-17) is about 12% based on data from the National Health Interview Survey.

The following sample size calculations were conducted using the minimum detectable effect levels seen
in the C8 and other studies that corresponded to similar serum levels of PFAS as observed among the
Pease children and adults.

95

Sex hormones and Insulin-like growth factor – 1 (IGF-1)
C8 study of children (Lopez-Espinosa 2016)
a. Estradiol
For PFOS, there was a -4% difference in the natural log estradiol among boys (per interquartile range of
the natural log of PFOS). Among boys, the median estradiol level was 10 pg/mL, with an interquartile
range (IQR) of 9.71 ng/ml) compared to the reference level of PFOA exposure. The
standard error for the reference group was 0.26 with N=32 in this group; and the standard error for the
90th percentile group was 0.33 with N=6. The standard deviations for the reference and 90th percentile
groups were therefore 1.47 and 0.81, respectively.
Assuming a 95% CI and 80% power, the sample size = 89/group; for a ratio of 2, the sample sizes = 158
and 79.
Assuming a 95% CI and 95% power, the sample size = 147/group; for a ratio of 2, the sample sizes =
260 and 130.
b. For females aged 12-19, there was a mean difference in the log TSH of -.35 mIU/L for PFOA
levels in the 90th percentile (>9.71 ng/ml) compared to the reference level of PFOA exposure. The
standard error for the reference group was 0.18 with N=71 and the standard error for the 90th percentile
group was 0.24 with N=14. The standard deviations for the reference and 90th percentile groups were
therefore 1.52 and 0.90, respectively.
Assuming a 95% CI and 80% power, the sample size = 200/group; for a ratio of 2, the sample sizes =
348 and 174.
Assuming a 95% CI and 95% power, the sample size = 331/group; for a ratio of 2, the sample sizes =
578 and 289.

Sample sizes for the categorical outcomes in Tables 6a-c were based on the following prevalences in
children:
Hypercholesterolemia: 34.2%
Hyperuricemia: 16%
Thyroid disease: 0.6%
ADHD 12.4% reported only; 5.1% reported with additional reporting on medications used for ADHD
Asthma: 11%
98

Rhinitis: 25.6%
Atopic dermatitis: 10.7%
Hypertension: 23.4%
Obesity: 17%

In addition to the sample size calculations presented in Tables 6 a-c, sample size calculations were done
assuming that a national PFAS study of children may be conducted in the future. Sample size
calculations were conducted with type 1 (“α error”) set at .05 and type 2 error (“β error) set at .20.
Sample sizes per stratum and sample sizes assuming a 2:1 ratio of exposed group to reference group
were calculated. It was considered important that a national study have a total sample size so that
exposures could be categorized into tertiles (i.e., reference level, medium level, and high level) or
preferably into quartiles (i.e., reference level, low, medium and high).
Studies were selected that were considered the most representative of U.S. populations exposed via
drinking water to PFOA, PFOS and/or PFHxS as a result of the migration of these PFAS chemicals into
ground water or surface water sources from the use of aqueous film forming foam (AFFF). The PFAS
serum results from the Pease International Tradeport testing program were used as representative PFAS
serum levels.
Studies conducted using NHANES data had PFOA and PFHxS serum levels similar to or lower than
those observed at Pease. In some of the more recent NHANES studies, the PFOS serum levels were only
moderately higher than at Pease. Therefore the PFOS, PFOA and PFHxS results in the NHANES studies
were used in many of the following sample size calculations for a possible national study. One major
drawback of the NHANES studies was that children under the age of 12 were not included because
PFAS was not measured in their serum. Recently, NHANES serum PFAS data for children under the
age of 12 was published, so it is likely that future NHANES studies will include this age group.
For those outcomes not included in NHANES studies, the C8 studies were used. The C8 results were
considered more representative of U.S. populations (e.g., in background disease rates and prevalence of
non-PFAS risk factors) than studies conducted in other countries, although the PFOS, and especially the
PFOA, serum levels in the C8 studies were higher than at Pease.
Table 8 provides the sample size calculations for several health outcomes. For some health outcomes
such as IQ, antibody response to vaccines, and delayed puberty, the information was insufficient to
estimate sample sizes. For the biomarkers, lipids, uric acid, testosterone, insulin-like growth factor – 1, a
total sample size of between 2,000 and 2,500 children should be sufficient. NHANES studies of ADHD,
rhinitis and antibody response to the MMR vaccine observed statistically significant findings with total
sample sizes considerably smaller than 2,000 children although the age range for the ADHD study was
limited to 12 – 15 years. Studies of executive function, attention, IQ, adiposity and obesity conducted in
the U.S. and other countries also had total sample sizes less than 2,000. An NHANES study of estimated
glomerular filtration rate observed statistically significant findings with a total sample size of just under
2,000 children.
For thyroid function, estradiol, asthma, and delayed puberty, total sample sizes exceeding 3,000 children
may be necessary. Outcomes such as specific childhood cancers and autism spectrum disorders would
also require total sample sizes much larger than 3,000.
99

In summary, a total sample size of 2,000 to 2,500 would be sufficient to evaluate a wide range of
biomarkers and outcomes including lipids (and hypercholesterolemia), uric acid (and hyperuricemia),
estimated glomerular filtration rate, testosterone, IGF-1, neurobehavioral measures (executive function,
attention, IQ) and ADHD, rhinitis, antibody response to MMR (and possibly also DPT) adiposity and
obesity.

100

Table 8. Children’s Study (ages 4-17 years)

Health-related
Endpoint

Relevant Study

Observed Effect Size

Assumptions

Total Cholesterol
(mg/dL)

Frisbee 2010, C8
Study
1,971 boys <12 yrs
2,773 boys 12-18 yrs
1,886 girls <12 yrs
2,520 girls 12-18 yrs
Lopez-Espinosa
2012, C8
1,078 1-5 yrs
3,132 6-10 yrs
6,447 >10 – 17 yrs

PFOS: 5th vs 1st quintile
Age: <12 yrs
12-18
Boys: +6.2
+9.3
Girls: +4.6
+9.4

PFOA: OR=1.44
(PFOS: OR < 1.0)
PFOS: 4th vs 1st quartile =
+0.19 mg/dL

Mean PFOS serum
levels were about 20
µg/L.
SD for total
cholesterol=29.3 mg/dL
Prevalence=34.2%
Mean PFOS serum
levels were about 20
µg/L. SD for TT4 as
estimated at 1.4. Percent
change in TT4 was
converted to mean
difference assuming the
median TT4 was ref.
level.
Prevalence=0.6%
(used PFOA results)
Mean PFOS serum level
= 12.8 µg/L. SD = 1.19.

PFOS: 4th vs 1st quartile,
OR=1.65

PFOA mean serum level
=3.5 µg/L. mean
difference= -6.6

Thyroid function
TT4

Thyroid disease
Uric Acid

Hyperuricemia

eGFR

Kataria 2015,
NHANES
1,960 12-18 yrs
Geiger 2013,
NHANES
1,772 12-18 years
Kataria 2015

OR = 1.6
PFOS, 4th vs 1 quartile:
2.3% change (mean
difference = 0.17 µg/dL)

Sample Size/Stratum
α error = .05
β error = .20
+4.6: 637/stratum
+6.2: 351/stratum
+9.3: 156/stratum

Sample Size:
ratio of 2:1 for
exposed vs ref.
956/478
526/263
234/117

300/stratum
1,080/stratum

446/223
1,620/810

>16,000/stratum
556/stratum

834/417

Mean PFOS serum level
=16.6. Prevalence=16%

400/stratum

572/286

Standard deviation=27.6

275/stratum

412/206

101

Health-related
Endpoint

Relevant Study

Observed Effect Size

Assumptions

Testosterone

Lopez-Espinosa
2016, C8
1,169 boys, 6-9 yrs
1,123 girls, 6-9 yrs

PFOS (IQR):
-5.8% boys (diff=1.9)
-6.6% girls (diff=2.45)

Estradiol

Lopez-Espinosa
2016, C8

PFOS (IQR):
Boys: -4% (diff=0.88)
Girls: -0.3%

Percent change was
converted to mean
difference assuming
median testosterone
level was ref. level. SD
estimated at 11.85 for
girls and 9.63 for boys.
Percent change was
converted to mean
difference assuming
median estradiol in boys
as ref. level. SD
estimated at 8.5.
Percent change was
converted to mean
difference assuming
median IGF-1 in boys as
ref. level. SD estimated
as 52.6
OR for delayed puberty
and the number of days
delayed puberty had
narrow CIs

IGF-1 (Insulin-like Lopez-Espinosa
growth factor – 1) 2016, C8

PFHxS (IQR):
Boys: -2.5% (diff=17.3
Girls: -2.1%

Delayed Puberty

Lopez-Espinosa
2011. C8
3,072 boys, 8-18 yrs
2,903 girls, 8-18 yrs

PFOS: mean serum level
was about 19 µg/L.

ADHD

Stein 2011, C8
10,546 aged 5-18 yrs.

PFHxS mean serum level
was 5.2 µg/L. 4th vs 1st
quartile,
OR=1.5
OR=1.6

Prevalence:
ADHD Dx: 12.4%
ADHD Dx/meds: 5.1%
102

Sample Size/Stratum
α error = .05
β error = .20
Boys: 404/stratum
Girls: 368/stratum

Sample Size:
ratio of 2:1 for
exposed vs ref.
606/303
552/276

Sample size
calculation for boys
only:
1,465/stratum

Sample size
calculated for
boys only:
2,198/1,099

146/stratum

218/109

Insufficient
information to
calculate sample
size, but sample
sizes in this study
were more than
enough for sufficient
precision

764/stratum
1,204/stratum

1,092/546
1,684/842

Health-related
Endpoint

Relevant Study

Observed Effect Size

Assumptions

Asthma

Stein 2016,
NHANES
640 12-19 yrs
Stein 2016,
NHANES
640 12-19 yrs
Wang 2011 (Taiwan)

PFOA mean serum level =
3.6 µg/L.
OR=1.2
PFOA mean serum level =
3.6 µg/L.
OR=1.35
PFOS mean serum
level=5.5 µg/L., 4th quartile
OR=2.19
PFOA mean serum
level=2.22 µg/L. OR=1.88

Rhinitis

Atopic dermatitis

Obesity

Karlsen 2016
(Faroes)

Prevalence = 11%

Sample Size/Stratum
α error = .05
β error = .20
2,400/stratum

Sample Size:
ratio of 2:1 for
exposed vs ref.
3,488/1,744

Prevalence=25.6%

858/stratum

1,260/630

Prevalence=10.7%

220/stratum

320/160

Prevalence=17%

250/stratum

368/184

ref.: referent group
Note: Observed effect sizes focused on the results for serum levels of PFOS and/or PFHxS unless the serum levels of PFOA were similar to
those observed at the Pease International Tradeport.

103

Adult Study
The following provides information on the parameters (e.g., standard deviation, disease prevalence) used
in the sample size calculations provided in Tables 7a-b for the adult study.

Liver Function – Adults
In the C8 study (Darrow 2016), the mean alanine aminotransferase (ALT) level was 26 IU/L and the
standard deviation was 19. The linear regression coefficient for the natural log ALT in the fifth quintile
level of cumulative natural log PFOA was 0.058. Assuming that the reference group had an ALT level
equal to the mean, the natural log of the mean ALT would be 3.26. Therefore the natural log of ALT for
the fifth quintile cumulative log PFOA would be 3.32. Exponentiating 3.32 equals 27.6. The mean
difference in the untransformed ALT is then 1.6.
Assuming a 95% CI and 80% power, the sample size = 2,214/group.
Assuming a 95% CI and 95% power, the sample size = 3,665/group.
In the C8 study (Gallo 2012), the linear regression on the natural log of ALT resulted in a regression
coefficient for the natural log PFOS of 0.029. The top quintile of PFOS level in the Pease adult
population was about 15 ng/mL. The natural log of 15 is 2.71; multiplying by 0.029 results in a natural
log ALT increase of 0.08. From the graph in the article, the reference level of ALT is about 21.3 IU/L.
The natural log of 21.3 is 3.06. Adding 0.08 to 3.06 equals 3.14, and exponentiating 3.14 equals 23.1.
Therefore the mean difference is 23.1 – 21.3 which equals 1.8.
The ALT standard deviation for the entire population was 20.1, and it was assumed that this was the
standard deviation for each quintile PFOS.
Assuming a 95% CI and 80% power, the sample size = 1,958/group.
Assuming a 95% CI and 95% power, the sample size = 3,241/group.

Thyroid Function – Adults
In a study done by Shrestha 2015, the sample size was 87 adults aged 55-74. Mean and SD for TSH was
2.58 µIU/mL and 1.47, respectively. The linear regression of the natural log TSH resulted in a
coefficient for the natural log PFOS of 0.129. Using a PFOS level of 15 ng/mL, the natural log of 15 is
2.71; multiplied by 0.129 equals 0.35. The reference level TSH was assumed to be the median TSH of
2.15 µIU/mL. The natural log of 2.15 is 0.77; adding 0.35 equals 1.12. Exponentiating 1.12 equals 3.06.
The mean difference is then 3.06 – 2.15 = 0.91. The standard deviation of 1.47 was used for each group.
Assuming a 95% CI and 80% power, the sample size = 41/group.
Assuming a 95% CI and 95% power, the sample size = 68/group.
104

a. TSH
In Ji 2012, the sample size was 633, ≥12 years of age and the median TSH level was 1.37 µIU/mL with
an IQR of 0.90, 2.01. The standard deviation was estimated as the IQR range divided by 1.35: (2.01 .90)/1.35 = 0.82. This standard deviation was assumed for each group. For TSH, the linear regression
coefficients for PFOS and PFHxS were 0.062 and 0.013, respectively. Using a PFOS level of 15 ng/mL
and a PFHxS level of 9 ng/mL, the mean difference for PFOS and PFHxS are 0.93 and 0.12,
respectively.
Assuming a 95% CI and 80% power, the sample size = 13/group for PFOS
Assuming a 95% CI and 95% power, the sample size = 21/group for PFOS
Assuming a 95% CI and 80% power, the sample size = 733/group for PFHxS
Assuming a 95% CI and 95% power, the sample size = 1,214/group for PFHxS
b. TT4 (total thyroxine)
In Ji 2012, the sample size was 633, ≥12 years of age and the median TT4 level was 7.4 µg/dL and the
IQR was 6.7, 8.1. The standard deviation was estimated: (8.1 – 6.7)/1.35 = 1.04. This standard deviation
was assumed for each group. For TT4, the linear regression coefficients for PFOS and PFHxS were 0.021 and -0.007, respectively. Using a PFOS level of 15 ng/mL and a PFHxS level of 9 ng/mL, the
mean difference for PFOS and PFHxS are -0.32 and -0.06, respectively.
Assuming a 95% CI and 80% power, the sample size = 166/group for PFOS
Assuming a 95% CI and 95% power, the sample size = 275/group for PFOS
Assuming a 95% CI and 80% power, the sample size = 4,716/group for PFHxS
Assuming a 95% CI and 95% power, the sample size = 7,809/group for PFHxS

Sample sizes for the categorical outcomes in Tables 7a-b were based on the following prevalences in
adults:
Hypercholesterolemia: 15%
Hyperuricemia: 24%
Thyroid disease: 6.5% (reported and confirmed by medical records); 11.5% (reported only)
Chronic kidney disease: 1.4%
Elevated ALT: 11.2%
Elevated GGT: 14%
Elevated bilirubin: 1.1%
Liver disease: 2%
Cardiovascular disease: 13%
Hypertension: 37%
Ulcerative colitis: 0.5%
Rheumatoid arthritis: 1.2%
Lupus: 0.2%
Multiple sclerosis: 0.32%
Osteoporosis: 5%
105

Osteoarthritis: 7.6%
Endometriosis: 7%
Pregnancy-induced hypertension: 8.5%
Kidney cancer: 0.3%

In addition to the sample size calculations presented in Tables 7 a-b, sample size calculations were done
assuming that a national PFAS study of adults may be conducted in the future. Sample size calculations
were conducted with type 1 (“α error”) set at .05 and type 2 error (“β error) set at .20. Sample sizes per
stratum and sample sizes assuming a 1:1 ratio of exposed group to reference group were calculated. It
was considered important that a national study have a total sample size so that exposures could be
categorized into tertiles (i.e., reference level, medium level, and high level) or preferably into quartiles
(i.e., reference level, low, medium and high).
Studies were selected that were considered the most representative of U.S. populations exposed via
drinking water to PFOA, PFOS and/or PFHxS as a result of the migration of these PFAS chemicals into
ground water or surface water sources from the use of aqueous film forming foam (AFFF). The PFAS
serum results from the Pease International Tradeport testing program were used as representative PFAS
serum levels.
Studies conducted using NHANES data had PFOA and PFHxS serum levels similar to or lower than
those observed at Pease. In some of the more recent NHANES studies, the PFOS serum levels were only
moderately higher than at Pease. Therefore the PFOS, PFOA and PFHxS results in the NHANES studies
were used in many of the sample size calculations. For those outcomes not included in NHANES
studies, the C8 studies were used. The C8 results were considered more representative of U.S.
populations (e.g., in background disease rates and prevalence of non-PFAS risk factors) than studies
conducted in other countries, although the PFOS, and especially the PFOA, serum levels in the C8
studies were higher than at Pease.
Table 9 provides the sample size calculations for several health outcomes. For lipids, uric acid,
cardiovascular disease, and osteoarthritis, a total sample of 6,000 should be sufficient. For thyroid
disease and thyroid function, a total sample size of 7,000 – 8,000 might be needed. However, NHANES
studies of thyroid function and thyroid disease obtained statistically significant findings with total
sample sizes considerably less than 6,000. NHANES studies of liver function also obtained statistically
significant findings with total sample sizes considerably less than 6,000. For ulcerative colitis, a sample
size of 6,000 might be sufficient if the effect size in the C8 study (i.e., OR=3.05) was likely for PFOA
serum levels considerably lower than those in the C8 study. For more likely effect sizes (e.g., ORs <
2.75), a total sample size of 6,000 would not be adequate to evaluate ulcerative colitis effectively. For
outcomes such as MS, lupus, rheumatoid arthritis, liver disease, kidney disease and kidney cancer, a
total sample size of 6,000 would likely not be adequate. For biomarkers of immune function (e.g.,
immunoglobin, C-reactive protein and cytokines) and fatty liver disease, there was insufficient
information to calculate sample sizes. However, a total sample size of 6,000 should be sufficient to
effectively evaluate these biomarkers.
In summary, a total sample size of 6,000 – 7,000 should be sufficient to evaluate a broad range of
biomarkers and outcomes such as lipids (and hypercholesterolemia), uric acid (and hyperuricemia),
cardiovascular disease, osteoarthritis, immune biomarkers and biomarkers for fatty liver disease. It also
may be sufficient to evaluate thyroid disease, thyroid function and liver function.
106

Table 9. Adult study (ages ≥18 years)
Health-related
Endpoint

Relevant Study

Observed Effect Size

Assumptions

Total Cholesterol
(mg/dL)

Steenland 2009, C8
46,294 aged ≥18 yrs

SD=41.9

Total Cholesterol
(mg/dL)
Thyroid disease

Fisher 2013, Canada

PFOS, mean serum level = 19.6 µg/L,
10th vs 1st decile:+11 mg/dL
4th vs 1st quartile, OR=1.51
PFHxS, mean serum level = 2.2 µg/L,
4th vs 1st quartile, OR=1.57
PFOA, mean serum level=3.5 µg/L,
4th vs 1st quartile:
Thyroid disease ever:
Women, OR=1.64
Men, OR=1.58
Thyroid disease with current meds
Women, OR=1.86
Men, OR=1.89
PFHxS mean serum level averaged
about 2 µg/L. Unit increase in
Ln(PFHxS):
Women, OR=3.10
Men, OR=1.57
PFOA mean serum level = 3.5 µg/L,
4th vs 1st quartile:
+0.44 mg/dL

Sample Size/Stratum
α error = .05
β error = .20
228/stratum

Prevalence=15%
Prevalence=44%

660/stratum
290/stratum

Prevalences:
16.18%
3.06%

410/stratum
2,035/stratum

Hyperuricemia, 4th vs 1st quartile:
OR=1.97

Prevalence:
19.2%

Subclinical
hypothyroidism

Uric Acid

Melzer 2010, NHANES
1,900 men, aged ≥20 yrs
2,066 women, aged ≥20 yrs

Wen 2013, NHANES
672 males aged ≥20 yrs
509 females aged ≥20 yrs
Shankar 2011, NHANES
3,883 aged ≥20 yrs

PFOS mean serum level = 17.9 µg/L
Hyperuricemia, 4th vs 1st quartile:
OR=1.5

107

9.89%
1.88%

Prevalences:
1.6%
2.2%
SD = 2.5

365/stratum
1,575/stratum

475/stratum
2,918/stratum
507/stratum

200/stratum

550/stratum

Health-related
Endpoint

Relevant Study

Observed Effect Size

Assumptions

Uric Acid

Steenland 2010, C8
53,458 aged ≥20 yrs

PFOS mean serum level = 20.2 µg/L,
10th vs 1st decile: +0.22 mg/dL
Hyperuricemia, 5th vs 1st quintile:
OR=1.26
PFOA and PFOS mean serum levels
were 28 µg/L and 20.3 µg/L,
respectively.
PFOA: OR=1.54
PFOS: OR=1.25
The top quintile of serum PFOS in
the Pease population was 15 µg/L.
This would approximately
correspond to a mean difference in
ALT of +1.8 µIU/mL
PFHxS mean serum level = 1.8 µg/L.
4th vs 1st quartile: OR=1.37
OR=3.05

SD=1.55

Liver function
Elevated ALT

Gallo 2012, C8
46,452 aged ≥18 yrs

Liver function
ALT (µIU/mL)

Gallo 2012, C8
46,452 aged ≥18 yrs

Liver function
Elevated ALT
Ulcerative colitis

Gleason 2015, NHANES
4,333 aged ≥12 yrs
Steenland 2013, C8
28,541 community and
3,713 worker cohorts
Steenland 2013, C8
28,541 community and
3,713 worker cohorts
Innes 2011, C8
49,432 aged >20 yrs
Uhl 2013, NHANES
4,102 aged 20-84

Rheumatoid
arthritis
Osteoarthritis
Osteoarthritis

Cardiovascular
disease

Shankar 2012, NHANES
1,216 aged ≥40 years

Prevalence:
24%
Prevalence = 11.2%

Sample Size/Stratum
α error = .05
β error = .20
780/stratum

1,525/stratum

SD=1.47

725/stratum
2,917/stratum
1,958/stratum

Assumed similar prevalence as
in the C8 study
Prevalence=0.5%

1,570/stratum
1,480/stratum

OR=1.35

Prevalence=1.2%

12,750/stratum

OR=1.42

Prevalence=7.6%

1,580/stratum

PFOA mean serum level = 5.4 µg/L ,
4th vs 1st quartile: OR=1.55

Assumed similar prevalence as
in the C8 study

978/stratum

PFOS mean serum level = 24.6 µg/L,
4th vs 1st quartile: OR=1.77
PFOA mean serum level = 4.2 µg/L,
4th vs 1st quartile: OR=2.01
108

550/stratum
Prevalence = 13%

250/stratum

Other sites with PFAS-contaminated drinking water from the UCMR-3
Table A1 shows the maximum combined levels of PFHxS and PFOS in any sample taken from
each utility. Only utilities with detectable levels of either PFHxS or PFOS are listed. The data are from
the UCMR-3 database as of July 2016 (US EPA 2016b). The ten utilities with the highest PFOS/PFHxS
levels in a sample are: the Commonwealth Utilities Corporation serving the Mariana Islands, the
Artesian Water Company serving portions of the state of Delaware, the Security Water and Sanitation
Districts serving Colorado Springs, the Horsham Water & Sewer (PA), the Warminster Municipal
Authority (PA), the Oatman Water Company (AZ), the Issaquah Water System (WA), the Hyannis
Water System (MA), the Suffolk County Water Authority (NY) and the Warrington Township Water &
Sewer (PA). Three of the top 10 utilities are located near each other in the vicinity of Philadelphia, PA:
Horsham, Warminster, and Warrington. ATSDR is currently considering whether it is feasible to include
children and adults from these towns in studies that would also evaluate the Pease populations.
Willow Grove Naval Air Station/Air Reserve Station (a.k.a. Naval Air Station Joint Reserve Base
and Air Force Reserve Station), Montgomery County, Pennsylvania
The Naval Air Station Joint Reserve Base (NASJRB) and Air Reserve Station (ARS) at Willow
Grove (“Willow Grove”) are two separate, but co-located military facilities in Montgomery County,
Pennsylvania. The Navy acquired site in 1942 and began jet training there in 1949; the air force base
began operations in 1958. In 2001, the Willow Grove bases employed 1,571 active-duty individuals, 993
members of the National Guard, 3,500 members of the Reserves, and 778 civilians with approximately
1,700 staff on-station daily. About 230 people resided on the bases year-round: less than 30 people
resided in single family dwellings and less than 200 resided in barracks. Additionally, there were five
officer family units, 200 enlisted family units, and 250 unaccompanied enlisted units as well as a
daycare center on base for 96 children. The Willow Grove Branch Medical Clinic was also located there
and provided primary care, medical support, preventive medicine, and occupational health services to
20,000 active duty, reserve, retired personnel, and their family members (ATSDR 2002a). Willow Gove
became an Air National Guard Base in September 2011. The surplus land with the runways was turned
over to Horsham Township for redevelopment.
AFFF used on the Willow Grove bases resulted in PFAS contamination of two nearby water
systems – the Warrington Township Water and Sewer Department (WTWSD) which served the eastern
portion of Warrington and the Horsham Water and Sewer Authority (HWSA).
In late October 2014, three of nine wells in the southern portion of the WTWSD were above the
EPA Provisional Health Advisory Level (PHAL) for PFOS and were taken out of service. PFOS levels
were the following: Well 1 (0.21 μg/L), Well 2 (1.6 μg/L), and Well 6 (1.3 μg/L). Although the wells
pump directly into the distribution system, wells 1, 2, and 6 are blended together at a tank and enter the
distribution system at one point. These wells constituted about 30% of the WTWSD supply. Well 3, in
the northeast area of the eastern section, and well 9, which is centrally located in the eastern section, had
very low levels of contamination.
Using currently available water distribution system information, ATSDR determined that for
“present-day” conditions, the northern part of the eastern section of the WTWDS system generally
received water that did not contain PFOA and PFOS. If any customers in the northern part of the system
received water containing PFOA and PFOS, it was at levels below the EPA Lifetime Health Advisory
(LTHA). The central part of the eastern section of the system may have received water containing PFOA
and PFOS concentrations above the EPA LTHA. The southeastern part of the eastern section of the
system received water containing PFOA and PFOS concentrations up to 10 times the EPA LTHA. More
109

detailed analyses of the water-distribution system need to be conducted to estimate historical PFAS
concentrations at specific housing areas. These analyses would involve looking at the water-distribution
system operating conditions, historical monthly well pumping records, and customer consumption
information in more detail.
The western section of the Warrington system is supplied by water purchased from North Wales
Water Authority (NWWA) and is not contaminated with PFAS. However, there is another
interconnection between the eastern sections of the system and NWWA which is used when there is a
need in the eastern section.
Warrington Township Water and Sewer Department (WTWSD) UCMR 2014-2015 data*
Well
PFOS (µg/L)
PFHxS (µg/L)
PFOA (µg/L)
PFNA (µg/L)
Wells 1, 2, 6
0.67
0.24
0.12
Well 3
0.06
0.04
0.02
Well 9
0.09
0.06
0.03
*Wells 1, 2, 3, and 6 were sample 11/11/2014; Well 9 was sampled 5/11/2015
The HWSA is served by 15 wells as well as interconnections with other nearby water utilities.
The water system is separated into two pressure zones, “high” and “low,” with the wells in each zone
pumping to fill storage tanks. The high zone has two storage tanks supplied by three wells and two
separate interconnections with North Wales Water Authority. The low zone has three storage tanks
served by 11 wells and an interconnection with Aqua Pennsylvania Southeastern Division. (Note: the
Aqua system had 0.009 µg/L of PFOS and .005 µg/L PFOA during UCMR-3 sampling in 4/16. Samples
taken between August 2016 and August 2017 at the Aqua Interconnection where it enters the HWSA
system measured an average of 0.0116 µg/L for PFOS and 0.0079 µg/L for PFOA.). June 2014 UCMR3
drinking water source sample results indicated that PFAS contamination was solely in the low pressure
zone which serves the majority of the service area. Prior to 1996 the system did not have pressure zones
which means customers located in the current high pressure zone may have received water from wells in
the low pressure zone. Generally, demand in each zone is met using water from the storage tanks within
that zone. The wells in each zone begin to pump simultaneously to meet customer demand and refill the
tanks when the water level in the tanks in the respective zone drop to a predetermined level Varying well
production rates and the active or inactive state of the wells and interconnections affect the quantity of
water from a specific source supplying the storage tanks, thus the PFAS concentrations in the tanks will
vary. While the wells are pumping, water is being supplied throughout the interconnected system both
from the tanks and from the various sources of water entering the system. When the tanks refill, the
wells turn off and water is supplied from the storage tanks. Water in the tanks is diminished and
restored based on demand and the wells cycle accordingly between active and inactive pumping states.
As a result, sources of supply are blended within the storage tanks and throughout the distribution
system. However it is possible that a property in close proximity to a well which has a demand at the
same time the well is pumping will have a higher percentage of water from the nearby well than from
other sources of supply.
Through the controlled operation of valves at two separate locations in the system, water from
the high zone can be moved by gravity into the low zone. A single booster location allows the
movement of water from the low zone into the high zone through a controlled pump and valve
operation.
In June 2014, HWSA wells were tested for PFAS as part of the UCMR3. PFAS were not
detected at nine supply wells. Three wells, Well 10, Well 17, and Well 21, had PFAS detected but
below the EPA PHALs. Two wells, well #26 and well #40, had levels of PFOA below the EPA PHAL
of 0.4 μg/l but levels of PFOS greater than the EPA PHAL of 0.2 μg/l.. Both wells #26 and #40 were
110

removed from service in July 2014 upon receipt of the sample results. The two contaminated wells
generally supplied about 25% of the water for the system; however, there were times that the two
contaminated wells supplied as much as 35% of the water for the system. A 15th well, Well 6 has been
in a reserve status since 2009 and was not sampled. PFAS levels from the UCMR-3 for the Horsham
supply wells are shown in the table below.
Following receipt of the first set of UCMR3 results, HWSA began actively monitoring its supply
wells for PFAS. Analyses performed at minimum reporting limits (MRL) lower than the UCMR3
MRLs ultimately revealed detections at all 14 wells. In May 2016 subsequent to the EPA announcement
of its Lifetime Health Advisory (LTHA) for PFOA/PFOS, wells 10, 17, and 21 were immediately taken
out of service. The other nine wells have tested below the LTHA.
ATSDR used currently available water-distribution system information to determine that for
“present-day” conditions, some areas in the southern and southeastern part of the low pressure zone
received water containing PFOA and PFOS concentrations up to 9 times the EPA LTHA. The
northeastern part of the low pressure zone received water containing PFOA and PFOS concentrations
less than the EPA LTHA. More detailed analyses of the system need to be conducted to estimate
historical PFAS concentrations at specific housing areas. These analyses would involve looking at the
water-distribution system operating conditions, historical monthly well pumping records, and customer
consumption information in more detail.
In addition to the five HWSA public wells exceeding the LTHA, the Navy and National Guard
Bureau (NGB) have identified over 90 additional private wells in Horsham that are at or above the
LTHA of 70 parts per trillion (ppt). The Navy or NGB is providing bottled water to these private well
owners.
Horsham Water and Sewer Authority (HWSA) UCMR 2014 data*
Well
PFOS (µg/L)
PFHxS (µg/L)
PFOA (µg/L)
PFNA (µg/L)
Well 10
0.05
0.04
0.03
-ND
Well 17
0.10
0.05
0.03
-ND
Well 21
0.14
0.08
ND-ND
Well 26
0.70
0.39
0.29
-ND
Well 40
1.00
0.59
0.06
-ND
*Wells 10 and 17 were sampled 12/9/2014; Wells 21, 26, and 40 were sampled 6/24/2014
Other drinking water contaminants
Supply wells on base contained volatile organic compounds (VOC) and metals. Maximum
detected levels in supply wells from sampling conducted in 1979-1984 were 91 ppb for PCE and 300
ppb for TCE. After contamination was detected, the well with the highest levels of contamination was
used mainly for fire protection. Additionally, the Navy installed an air stripper to treat groundwater prior
to distribution, and monitoring of treated water between 1996 and 1998 found no contaminants above
EPA’s Maximum Contaminant Levels (MCLs) (ATSDR 2002a). According to the EPA, over 800
employees at the two facilities may have drank or come into contact with treated water from the Navy
supply wells (https://cumulis.epa.gov/supercpad/cursites/csitinfo.cfm?id=0303820). VOC contamination
in off-site wells has not been attributed to the base, and the local water authorities (HWSA and
WTWSD) treat the water for VOCs before distribution (ATSDR 2002a).

111

Naval Air Warfare Center (a/k/a Naval Air Development Center), Warminster Township, Bucks
County, Pennsylvania
The former Naval Air Warfare Center (NAWC) is located in Warminster Township. The base
operated from 1944 until its closure in September 1996. In 1994, approximately 1,850 civilians and
1,000 military personnel were stationed or employed on base. At its peak, the base employed 2,800
civilians, 200 military personnel, and up to 300 daily contractors (ATSDR 2002b).
Approximately 800 to 1,000 military personnel and their families stationed at nearby Willow
Grove Naval Air Station lived in two on-base housing areas at NAWC while as many as six families
may have resided in officer housing. Between 450-550 enlisted personnel and their families lived at the
Shenandoah Woods housing complex. Site 5, a former landfill, was located in Shenandoah Woods.
Quarters A and B, located within Area C, provided housing for the base’s commanding officer and
second-in-command (ATSDR 2002b).
Four out of eighteen of the Warminster Municipal Authority (WMA) public water supply wells
are in close proximity to the former NAWC site. The WMA provides water to approximately 40,000
people. The water supplied to the customers is from water supply wells in the WMA system and may be
purchased from the North Wales Water Authority (NWWA) as well as the Upper Southampton
Municipal Authority on an emergency basis. WMA's water supply wells are connected individually to
the distribution network and are subsequently blended within the distribution system in tanks and
standpipes. Therefore, customers located geographically closest to a given water supply well will likely
receive more water from that well than users located further away (ATSDR 2016).
AFFF was used for decades at the base for firefighting training activities. PFAS were first tested
for in groundwater as emerging contaminants in preparation for the CERCLA 2012 Five Year Review
for this site. In summer 2013, PFOS levels above the EPA PHAL were first discovered in groundwater
on the former Navy property. As part of the EPA's UCMR-3, sampling for six PFAS in the WMA first
occurred in November 2013. UCMR-3 monitoring for PFAS is required at the entry point to the
distribution system for each well and at any interconnection in operation. Accordingly, WMA conducted
sampling in November 2013 and May 2014 for all wells and conducted sampling in November 2013 and
February, May, and August 2014 for the interconnection with NWWA (ATSDR 2016).
Samples taken in the WMA system detected levels of PFOS, PFOA, PFHxS and/or PFHpA. The
source of the contamination was the use of AFFF at NAWC. In November 2013, three WMA public
water wells had levels at or above EPA’s PHAL for PFOS. In this sampling event, 17 samples covering
17 wells in the WMA and one sample of the NWWA interconnection were taken and analyzed for
PFAS. One of the 17 WMA samples represents the combined water extracted from WMA Wells 43 and
44. Water from these two wells is combined for treatment and samples are taken after treatment at the
entry point to the distribution system. PFOS was detected in 6 public wells and PFOA was detected in 8
public wells. PFOS was detected in Well 26 at 0.791 µg/L, more than three times the 0.2 µg/L PFOS
PHAL value. Wells 10 and 13 had PFOS concentrations of 0.193 and 0.16 µg/L that can be rounded to
0.2 µg/L. None of the PFOA detections exceeded the PFOA PHAL in the WMA wells. Well 26 had the
highest detections for PFOA and PFOS. In summer 2014, PFOS was detected in four public wells. The
highest concentrations were in Well 26 at 1.09 µg/L, more than five times the 0.2 µg/L PFOS PHAL
value, and in Well 10 at 0.176 µg/L. PFOA was detected in four wells, including Well 26 at 0.349 µg/L,
close to the 0.4 µg/L PHAL for PFOA. Wells 13 and 26 were shut down in June 2014. Well 10 was shut
down in September 2014. On May 19, 2016, wells 2, 14 and 15 were removed from service due to the
EPA new lifetime health advisory level for PFOA/FPOS (ATSDR 2016).
PFOS levels above the PHAL were also detected in private drinking water samples. As of
September 2015, 100 private wells (94 residential and 6 non-residential) were identified and sampled
within an approximate 1-3 mile radius of the site. At least one PFAS was detected in the majority (93
112

out of 100) of these private water wells. Of the 94 residential private water wells, five were non-detect
for PFOA and PFOS, 18 had detections of PFOA only, and 71 had both PFOA and PFOS. Eleven
exceeded the PFOS PHAL, ranging from 0.152 µg/L to 0.729 µg/L. The PFOS PHAL exceedances are
in two general locations: one location is south of the Jacksonville Road and East Bristol Road
intersection and the other location is in the area of York Road and W Street. Six residential wells with
PFOS levels that range from 0.102 to 0.109 µg/L (50% of the PHAL) are located at the
Jacksonville/East Bristol Roads intersection (ATSDR 2016).
The Navy and EPA provided a limited number of residents whose private well water was at or
above EPA's PHAL (with rounding up to one significant digit) with bottled water to use for drinking and
cooking water, and is currently working to connect these locations to public water (ATSDR 2016).
Using currently available water-distribution system information, ATSDR determined that for
“present day” conditions, the southwestern part of the Warminster system typically received water that
did not contain PFOA and PFOS concentrations. If any customers in this part of the system received
water containing PFOA and PFOS concentrations, it was at levels below the EPA LTHA. The
northwestern part of the Warminster system typically received water containing PFOA and PFOS
concentrations at or below the EPA LTHA. Some areas in the eastern parts of the Warminster system
received water containing PFOA and PFOS concentrations at levels up to three times the EPA LTHA,
and areas in the central part received water containing concentrations at level up to 15 times the EPA
LTHA. More detailed analyses of the system need to be conducted to estimate historical PFAS
concentrations at specific housing areas. These analyses would involve looking at the water-distribution
system operating conditions, historical monthly well pumping records, and customer consumption
information in more detail.
Although some WMA customers received the majority of their water from one of the
contaminated wells, the majority of water customers likely received water that either did not contain
PFAS or had levels less than the PHALs (but levels may be higher than the EPA LTHA for
PFOS/PFOA). If one assumes that all the wells supply a similar amount of water to the system (each
well typically supplied 5-10% of the water to the system), then the number of customers potentially
exposed to elevated PFAS in their drinking water could be approximately 7,000.
Warminster Municipal Authority (WMA) UCMR 2013-2014 data*
Well
PFOS (µg/L)
PFHxS (µg/L)
PFOA (µg/L)
PFNA (µg/L)
Well 2
0.06
0.03
0.03
Well 10
0.19
0.10
0.09
Well 13
0.16
0.09
0.12
Well 14
0.06
0.03
0.02
Well 15
0.06
0.04
0.02
Well 26
1.09
0.39
0.35
*Wells 2, 10, 13, 14, and 15 were sampled 11/19/2013; Well 26 was sampled 6/9/2014
Other drinking water contaminants
Samples taken in 1979 showed maximum levels of contamination in on-site supply wells of 36
ppb for PCE and 293 ppb for TCE. These wells were closed in 1979. Contamination levels in samples
taken from off-site municipal supply wells found 17 ppb for PCE and 67.8 ppb for TCE; past off-base
residents may have been exposed to these VOCs between 1974, when the well first began supplying
water, until it was closed in 1979. Sampling of VOCs in off-site private wells detected PCE at 31 ppb; as
a result, affected homes were connected to municipal water supplies or groundwater treatment systems
were installed (ATSDR 2002b).
113

Because the TCE- and PCE-contaminated wells were shut down in 1979, military service
personnel and DOD civilian workers who began service/employment at NAWC after 1979 might be
eligible for a PFAS study. More information is needed to determine when the water supply may have
been contaminated with PFAS.
More detailed analyses will help determine which specific housing areas received water
containing PFOA and PFOS from the NASJRB and ARS at Willow Grove and the NAWC in
Warminster. To conduct more detailed analyses, including modeling, additional information and specific
data pertinent to each water system’s operations needs to be obtained from site visits to the water
utilities.

114

Appendix tables

115

Table A1. Maximum levels (parts per billion) of combined PFHxS and PFOS from the US EPA’s Third Unregulated Contaminant
Monitoring Rule (UCMR-3)
Water Utility Name
Commonwealth Utilities Corp. (Saipan)
Artesian Water Company
Security WSD
Horsham Water & Sewer Authority
Warminster Municipal Authority
Oatman Water Company
Warrington Township Water & Sewer Department
Issaquah Water System
Hyannis Water System
Suffolk County Water Authority
United Water PA
Emerald Coast Utilities Authority
GU Waterworks Authority - Northern System
Widefield WSD
Oakdale
City of Tucson
City of Cleveland Heights
Sanford Water District
Wright-Patterson AFB Area A/C
Liberty Water LPSCO
Westfield Water Department
City of Zephyrhills
Bemidji
City of Fountain
City of Stuart Water Plant

State
MP
DE
CO
PA
PA
AZ
PA
WA
MA
NY
PA
FL
GU
CO
MN
AZ
OH
ME
OH
AZ
MA
FL
MN
CO
FL
116

Size
L
L
L
L
L
S
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L

PFHxS & PFOS sum
8.60
2.48
1.89
1.59
1.479
1.03
0.91047
0.841
0.7
0.67
0.572
0.56
0.55
0.54
0.4913
0.476
0.4
0.4
0.36
0.33
0.33
0.32
0.32
0.29
0.259

Water Utility Name
City of Tempe
CA American Water Co. - Suburban
City of Newburgh
CA Water Service - Visalia
Eastern Municipal Water District
New Windsor Consolidated Water District
VAW Water System, Inc.
Freeport
La Crosse Waterworks
Salt River Public Works
City of Martinsburg
Dyer Water Department
Atlantic City MUA
West Morgan - East Lawrence Water Authority
City of Greensboro
Rome
Dover Water Department
CA Water Service - Chico
Moore County Public Utilities - Pinehurst
Rhinelander Water & Wastewater
Bayleaf Master
City of Ocala
NJ American Water Co. - Raritan
Mahwah Water Department
City of Abilene
West Lawrence Water Co-op
Hampton Bays Water District
Fort Drum

State
AZ
CA
NY
CA
CA
NY
AL
IL
WI
09*
WV
IN
NJ
AL
NC
GA
NH
CA
NC
WI
NC
FL
NJ
NJ
TX
AL
NY
NY
117

Size
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
S
L
L
L
L
L
L
L
L

PFHxS & PFOS sum
0.245
0.241
0.24
0.212
0.202
0.1936
0.18
0.18
0.172
0.166
0.157
0.1437
0.142
0.13
0.124
0.12
0.12
0.118
0.118
0.1173
0.11
0.104
0.103
0.098
0.09781
0.09
0.082
0.08

Water Utility Name
City of Lathrop
Northeast Alabama Water System
City of Anaheim
Fair Lawn Water Department
City of Orange
Montebello Land & Water Company
Vienna
Chatsworth
Bethany
City of Pico Rivera Water Department
Camp Pendleton (South)
Montgomery County Water Services #2
Rainbow City Utilities Board
Florence Water-Wastewater Department
Plainfield Township
Pendleton County Water District #1/South
City of Miami Beach
Ridgewood Water
Woodbury
Montgomery County Water Services #1
CA Water Service - East Los Angeles
Town of Nashville
Metropolitan DWID
City of Downey Water Department
Pierre
Park Water Company - Bellflower/Norwalk
Washington Township MUA

State
CA
AL
CA
NJ
CA
CA
WV
GA
OK
CA
CA
OH
AL
AL
MI
KY
FL
NJ
MN
OH
CA
NC
AZ
CA
SD
CA
NJ

118

Size
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
S
L
L
L
L
L
S
L
L
L
L
L

PFHxS & PFOS sum
0.076
0.07
0.07
0.06603
0.0659
0.065
0.0641
0.06303
0.063
0.062
0.062
0.061
0.06
0.06
0.06
0.05853
0.058
0.058
0.0577
0.0542
0.054
0.05312
0.053
0.053
0.053
0.051
0.0503

Water Utility Name
Colbert County Rural Water System
Gadsden Waterworks & Sewer Board
Southside Waterworks
City of North Miami
Kennebunk, Kennebunkport & Wells WD
Bell Arthur Water Corp.
City of Garden Grove
City of Lauderhill
FKAA
Yorba Linda Water District
City of Miramar
Miami International Airport
City of Corona
Orchard Dale Water District
Lima City Water
Pico Water District
Golden State Water Co. - Norwalk
MDWASA - Main System
Ann Arbor
City of Fullerton
Cliffdale West
Central ASG
City of DeFuniak Springs Water System
Cottage Grove
City of Great Bend
City of Pleasanton
Sacramento Suburban Water District

State
AL
AL
AL
FL
ME
NC
CA
FL
FL
CA
FL
FL
CA
CA
OH
CA
CA
FL
MI
CA
NC
AS
FL
MN
KS
CA
CA

119

Size
L
L
L
L
L
S
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L

PFHxS & PFOS sum
0.05
0.05
0.05
0.05
0.05
0.05
0.0496
0.049
0.049
0.0474
0.047
0.047
0.046
0.045
0.045
0.044
0.043
0.043
0.043
0.0412
0.041
0.04
0.04
0.0381
0.037
0.036
0.036

Water Utility Name
Mashpee Water District
Belvidere

State
MA
IL

L=large system (serves >10,000); S=small system (serves <10,000)
*

Tribal nation located in Arizona

120

Size
L
L

PFHxS & PFOS sum
0.033
0.03167

Table A2. PFAS studies on cancers and other chronic diseases in adults.
Reference

Exposure*

Outcome

RR (SIR, SMR, OR, HR) &
95% CI

Alexander 2003
Decatur, AL
POSF plant
Geometric mean
serum level:
PFOS=0.9 ppm
PFOA=1.13 ppm
Olsen 2004
Decatur, AL
POSF plant
Grice 2007
Decatur, AL
POSF plant
Alexander 2007
Decatur, AL
POSF plant

POSF/PFOS
High exposure job
≥1 yr

Bladder Cancer (and other
urinary organs) mortality

3 cases
SMR=12.77 (2.63, 37.35)
SMR=16.12 (3.32, 47.14)

Eriksen 2009
Danish Pop.
Median serum
level:
PFOS=35 ng/ml
PFOA=6.9 ng/ml

Hardell 2014
Swedish Pop.
Median serum
level: (ng/ml)
PFOS=9.0
PFOA=2.0
PFHxS=0.91

Duration/Intensity/cumulative exp.

Urinary Cancers mortality
High exposure job
≥1 yr
POSF/PFOS

SMR=4.02 (0.83, 11.75) (3 cases)
SMR=5.11 (1.05, 14.93)
Colon Cancer
Rectal Cancer
Prostate Cancer

POSF/PFOS

RR=5.4 (0.5, >100) 4 exposed
RR=1.8 (0.3, 12.4) 4 exposed
RR=7.7 (0.9, >100) 5 exposed
Odds Ratios:

Colon Cancer
Prostate Cancer
Bladder Cancer (11 cases)

POSF/PFOS
Ever low
Ever low or high
Low or high (≥1 yr)
Ever high
High (≥1 yr)
PFOS
Prostate Cancer
PFOA
(PFOS & PFOA: no association
for cancers of bladder or liver)

SIR=2.26 (0.91, 4.67)
SIR=1.70 (0.77, 3.22)
SIR=1.31 (0.48, 2.85)
SIR=1.74 (0.64, 3.79)
SIR=1.12 (0.23, 3.27)
Trend: RR=1.05 (0.97, 1.14)/10 ppb
Trend: RR=1.03 (0.99, 1.07)/1 ppb
(plasma concentration)

Pancreatic Cancer
PFOS
PFOA
PFAS (the median
level of each PFAS
among the controls
was used as the cut
point for the
calculation of the
OR.)

No association with PFOS
Trend: RR=1.03 (0.98, 1.10)/1 ppb
Prostate Cancer
(case-control)

PFHxS: OR=1.3 (0.8, 1.9)
PFOS: OR=1.0 (0.6, 1.5)
PFOA: OR=1.1 (0.7, 1.7)
PFNA: OR=1.2 (0.8, 1.8)
PFDA: OR=1.4 (0.9, 2.1)
PFUnDA: OR=1.2 (0.8, 1.9)

121

Long term: RR=12 (0.8, >100)
Long term: RR=11 (0.8, >100)
Long term: RR=8.2 (0.8, >100)
Low/high L/H (>1 yr) High (>1 yr)
1.2 (0.5, 2.9) 1.4 (0.6, 3.3) 1.7 (0.7, 4.2)
1.3 (0.6, 2.9) 1.4 (0.6, 3.0) 1.1 (0.4, 2.7)
Cumulative exposure:
Low RR=0.83 (0.15, 4.65)
Medium RR=1.92 (0.30, 12.06)
High RR=1.52 (0.21, 10.99)

RRs PFOS
Low: 1.35 (0.97, 1.87)
Med: 1.31 (0.94, 1.82)
High: 1.38 (0.99, 1.93)

PFOA
1.09 (0.78, 1.53)
0.94 (0.67, 1.32)
1.18 (0.84, 1.65)

Low:
0.88 (0.49, 1.57)
Med:
1.33 (0.74, 2.38)
High:
1.55 (0.85, 2.80)
ORs: PSA ≥11 Heredity, PFC>median
1.5 (0.9, 2.6) 4.4 (1.7, 12)
0.8 (0.4, 1.3) 2.7 (1.04, 6.8)
1.3 (0.8, 2.1) 2.6 (1.2, 6.0)
1.2 (0.7, 2.1) 2.1 (0.9, 4.8)
1.2 (0.7, 2.0) 2.6 (1.1, 6.1)
1.0 (0.6, 1.6) 2.6 (1.1, 5.9)

Reference

Exposure*

Outcome

RR (SIR, SMR, OR, HR) &
95% CI

Duration/Intensity/cumulative exp.

Raleigh 2014
APFO production
workers, Cottage
Grove, 3M

PFOA

Cancer:

Cumulative exposure, Mortality
Q1: HR=0.34 (0.25, 1.60)
Q2: HR=1.12 (0.53, 2.37)
Q3: HR=0.36 (0.11, 1.17)
Q4: HR=1.32 (0.61, 2.84)

Cumulative exposure, Incidence
Q1: HR=0.80 (0.57, 1.11)
Q2: HR=0.85 (0.61, 1.19)
Q3: HR=0.89 (0.66, 1.21)
Q4: HR=1.11 (0.82, 1.49)

Kidney

Q1-Q2: HR=0.38 (0.11, 1.23)
Q3-Q4: HR=0.39 (0.11, 1.32)

Q1-Q2: HR=1.07 (0.46, 2.46)
Q3-Q4: HR=0.85 (0.36, 2.06)

Pancreas

Q1: HR=0.32 (0.08, 1.35)
Q2: HR=0.89 (0.34, 2.31)
Q3: HR=0.82 (0.32, 2.12)
Q4: HR=1.23 (0.50, 3.00)

Q1-Q2: HR=0.13 (0.02, 1.03) (1 case)
Q3-Q4: HR=1.36 (0.59, 3.11)

Bladder

Q1-Q2: HR=1.03 (0.27, 3.96)
Q3-Q4: HR=1.96 (0.63, 6.15)

Q1: HR=0.81 (0.36, 1.81)
Q2: HR=0.78 (0.33, 1.85)
Q3: HR=1.50 (0.80, 2.81)
Q4: HR=1.66 (0.86, 3.18)

Breast

Q1-Q2: HR=0.61 (0.25, 1.48)
Q3-Q4: HR=0.54 (0.15, 1.94)

Q1-Q2: HR=0.46 (0.25, 0.87)
Q3-Q4: HR=1.27 (0.70, 2.31)

Geometric mean
serum level:
PFOA=815 ng/ml
PFOA-related
manufacturing,
PFOA=2,538
ng/ml

Prostate

Chronic Diseases:
Ischemic Heart Disease

Q1: HR=0.93 (0.73, 1.18)
Q2: HR=0.87 (0.66, 1.13)
Q3: HR=0.88 (0.68, 1.13)
Q4: HR=0.89 (0.66, 1.21)

Cerebrovascular Disease

Q1: HR=0.57 (0.32, 1.02)
Q2: HR=0.70 (0.39, 1.24)
Q3: HR=0.93 (0.57, 1.53)
Q4: HR=0.98 (0.53, 1.81)

Diabetes

Q1: HR=0.27 (0.10, 0.76)
Q2: HR=0.42 (0.17, 1.04)
Q3: HR=0.80 (0.42, 1.51)
Q4: HR=0.72 (0.34, 1.52)

122

Reference

Exposure*

Outcome

RR (SMR, HR) & 95% CI

Duration/Intensity/cumulative exp.

Lundin 2009
APFO production
workers, Cottage
Grove, 3M€

PFOA

Mortality

Job classification

Cumulative Exposure-Years

Prostate Cancer

Low (ref.)
Med HR=3.0 (0.9, 9.7)
High HR=6.6 (1.1, 37.7)

<1
≥1

(ref.)
HR=2.0 (0.7, 5.3)

Pancreatic Cancer

Low (ref.)
Med/hi HR=1.6 (0.5,4.8)

<1
≥1

(ref.)
HR=1.8 (0.6, 5.6)

Bladder Cancer

Low (ref.)
Med/hi HR=0.7 (0.2, 3.4)

<1
≥1

(ref.)
HR=1.7 (0.4, 7.8)

Cerebrovascular Disease

Low (ref.)
Med HR=1.8 (0.9, 3.7)
High HR=4.6 (1.3, 17.0)

<1 (ref.)
1-4.9 HR=0.6 (0.2, 2.2)
≥5 HR=2.1 (1.0, 4.6)

Diabetes

Low (ref.)
Med/hi HR=3.4 (1.3, 9.3)

<1 (ref.)
≥1 HR=1.3 (0.6, 3.1)
cumulative exposure-years (SMR¥≠)
Q1 Q2 Q3
Q4
2.39 0 2.01 0.32
1.18 1.02 1.09 0.92
0.58 0.63 1.09 0.75
1.49 0
0.87 0
1.07 0.82 0.65 0.57
1.07 1.37 0
2.66
1.24 2.49 0.39 0.36
0
0
1.73 6.27
1.54 0.99 0.85 0.96
0.28 2.34 0.57 1.03
1.85 1.47 2.30 1.90
1.07 1.02 0.87 0.93
0.63 0.78 1.34 0.69
0.93 1.00 1.30 0.93
1.32 2.10 0.37 0.72
0
3.79 1.83 8.60

Steenland and
Woskie 2012
Dupont plant,
Parkersburg WV
(Washington
Works Plant)
Median serum
level: (1979-2004)
PFOA=580 ng/ml
Nonexposed
workers:
PFOA=160 ng/ml
Directly exposed
workers:
PFOA=2,880
ng/ml

PFOA

Mortality
Liver Cancer
Pancreatic Cancer
Lung Cancer
Breast Cancer
Prostate Cancer
Kidney Cancer
Bladder Cancer
Mesothelioma
NHL
Leukemias
Diabetes
Ischemic Heart Disease
Stroke
COPD
Chronic Liver Disease
Chronic Renal Disease

SMR=1.07 (0.51, 1.96)
SMR=1.04 (0.62, 1.64)
SMR=0.78 (0.62, 1.64)
SMR=0.65 (0.13, 1.90)
SMR=0.76 (0.47, 1.16)
SMR=1.28 (0.66, 2.24)
SMR=1.08 (0.52, 1.99)
SMR=2.85 (1.05, 6.20)
SMR=1.05 (0.57, 1.76)
SMR=1.05 (0.57, 1.76)
SMR=1.90 (1.35, 2.61)
SMR=0.97 (0.86, 1.09)
SMR=0.86 (0.64, 1.14)
SMR=1.05 (0.75, 1.42)
SMR=1.09 (0.54, 1.95)
SMR=3.11 (1.66, 5.32)

123

Reference

Exposure*

Outcome

Steenland 2015
Dupont plant,
Parkersburg WV

PFOA

Ulcerative colitis
Rheumatoid arthritis
Bladder cancer
Colorectal cancer
Prostate cancer
Melanoma
Liver disease (non-hepatitis)
Thyroid disease (males)
Thyroid disease (females)
Coronary heart disease
Hypertension
High cholesterol
Osteoarthritis
Stroke
COPD
Asthma
Kidney disease
Diabetes

Median serum
level: (2005/2006)
PFOA=113 ng/ml

cumulative exposure-years (RR & 95% CI), 10 year exposure lag
Q1 (ref.) Q2
Q3
Q4
3.00 (0.82, 11.0)
1.74 (0.45, 6.77)
0.55 (0.12, 2.61)
0.31 (0.90, 1.11)
1.92 (0.56, 6.58)
0.85 (0.27, 2.71)
1.46 (0.42, 5.04)
1.23 (0.57, 2.66)
0.79 (0.42, 1.50)
1.20 (0.82, 1.75)
0.95 (0.81, 0.98)
0.93 (0.79, 1.10)
0.74 (0.49, 1.10)
1.48 (0.56, 3.69)
0.75 (0.38, 1.48)
0.48 (0.10, 2.29)
1.32 (0.32, 5.43)
1.06 (0.75, 1.49)

124

3.26 (0.70, 15.1)
2.12 (0.40, 11.1)
0.47 (0.10, 2.21)
0.99 (0.33, 2.94)
1.89 (0.57, 6.34)
1.10 (0.34, 3.58)
2.13 (0.59, 7.71)
1.70 (0.74, 3.91)
0.87 (0.37, 2.02)
1.06 (0.71, 1.58)
0.91 (0.75, 1.16)
1.01 (0.84, 1.22)
0.56 (0.34, 0.93)
1.53 (0.60, 3.89)
1.16 (0.60, 2.26)
0.57 (0.11, 2.93)
0.50 (0.11, 2.34)
1.10 (0.76, 1.61)

6.57 (1.47, 29.4)
2.62 (0.47, 14.7)
0.31 (0.06, 1.54)
1.06 (0.34, 3.31)
2.15 (0.64, 7.26)
0.75 (0.21, 2.67)
2.02 (0.50, 8.10)
1.71 (0.68, 4.25)
0.23 (0.05, 1.01)
0.93 (0.61, 1.41)
0.95 (0.77, 1.16)
0.96 (0.78, 1.18)
0.67 (0.39, 1.14)
1.33 (0.51, 3.43)
0.77 (0.38, 1.57)
0.52 (0.09, 2.83)
0.67 (0.15, 3.05)
1.12 (0.76, 1.66)

Reference
BonefeldJorgensen 2011
Greenlandic Inuit
Median serum
level:
PFOS=45.6 ng/ml
PFOA=2.5 ng/ml
BonefeldJorgensen 2014
Case-control
study,
premenopausal
mothers nested in
the Danish
National Birth
Cohort
BonefeldJorgensen 2014
(cont.)
Mean serum
level:
PFOS=30.6 ng/ml
PFOA=5.2 ng/ml
PFHxS=1.2 ng/ml
PFNA=0.5 ng/ml
PFOSA=3.5 ng/ml

Exposure
PFCs
PFOS
PFOA
sumPFSA£
sumPFCA±

Outcome
Breast Cancer (case-control)

Odds Ratio & 95% CI
(per ng/ml, serum)
OR=1.03 (1.00, 1.07)
OR=1.20 (0.77, 1.88)
OR=1.03 (1.00, 1.05)
OR=1.07 (0.96, 1.18)

Odds Ratios & 95% CI

Breast Cancer
PFOS, Q2
Q3
Q4
Q5

Quartile 1 (ref.)
OR=1.51 (0.81, 2.71)
OR=1.51 (0.82, 2.84)
OR=1.13 (0.59, 2.04)
OR=0.90 (0.47, 1.70)

Age, Dx ≤40
OR=1.2 (0.5, 2.9)
OR=1.4 (0.6, 3.3)
OR=0.8 (0.3, 1.9)
OR=1.0 (0.4, 2.5)

Age, Dx > 40
OR=2.3 (0.9, 5.6)
OR=1.9 (0.7, 5.0)
OR=2.2 (0.9, 5.7)
OR=0.9 (0.3, 2.4)

PFOA, Q2
Q3
Q4
Q5

OR=0.97 (0.53, 1.75)
OR=1.02 (0.56, 1.89)
OR=1.14 (0.62, 2.12)
OR=0.94 (0.51, 1.76)

OR=0.7 (0.3, 1.6)
OR=1.3 (0.5, 3.2)
OR=0.8 (0.4, 2.0)
OR=0.8 (0.3, 1.9)

OR=1.8 (0.7, 4.3)
OR=0.9 (0.4, 2.3)
OR=1.9 (0.8, 4.8)
OR=1.2 (0.5, 2.9)

PFNA, Q2
Q3
Q4
Q5

Quartile 1 (ref.)
OR=1.10 (0.60, 2.02)
OR=0.75 (0.41, 1.40)
OR=1.08 (0.58, 1.99)
OR=0.80 (0.43, 1.47)

Age, Dx ≤40
OR=1.1 (0.4, 2.5)
OR=0.5 (0.2, 1.3)
OR=0.8 (0.4, 1.9)
OR=0.6 (0.2, 1.4)

Age, Dx > 40
OR=1.2 (0.5, 3.0)
OR=1.3 (0.5, 3.3)
OR=1.9 (0.7, 4.9)
OR=1.1 (0.5, 2.8)

PFHxS, Q2
Q3
Q4
Q5

OR=0.64 (0.34, 1.18)
OR=0.70 (0.38, 1.29)
OR=0.38 (0.20, 0.70)
OR=0.61 (0.33, 1.12)

OR=0.4 (0.2, 0.9)
OR=0.6 (0.2, 1.4)
OR=0.3 (0.1, 0.7)
OR=0.4 (0.2, 1.0)

OR=1.2 (0.4, 3.4)
OR=1.0 (0.4, 2.7)
OR=0.5 (0.2, 1.4)
OR=1.0 (0.4, 2.5)

PFOSA, Q2
Q3
Q4
Q5

OR=1.38 (0.75, 2.52)
OR=0.91 (0.49, 1.66)
OR=1.11 (0.60, 2.05)
OR=1.89 (1.01, 3.54)

OR=1.5 (0.7, 3.3)
OR=1.0 (0.5, 2.4)
OR=1.1 (0.5, 2.6)
OR=2.5 (1.0, 6.0)

OR=1.3 (0.5, 3.6)
OR=1.0 (0.4, 2.5)
OR=1.4 (0.5, 3.6)
OR=1.6 (0.6, 4.3)

Breast Cancer

125

Reference
Barry et al 2013
C8 Health Project
Entire
cohort=32,254
Community
Cohort = 28,541
Worker Cohort =
3,713
Median serum
level:
Community,
PFOA=24.2 ng/ml
Worker,
PFOA=112.7
ng/ml

Exposure
PFOA∆

Outcome
Bladder Cancer
Brain Cancer
Breast Cancer
Cervical Cancer
Colorectal Cancer
Esophageal Cancer
Leukemia
Liver Cancer
Lung Cancer
Lymphoma
Melanoma
Oral Cancer
Ovarian Cancer
Pancreatic Cancer
Prostate Cancer
Soft Tissue Cancer
Stomach Cancer
Uterine Cancer
Kidney Cancer
Q2
Q3
Q4

Hazard Ratio (95% CI) 10 yr. lag
HR=0.98 (0.88, 1.10)
HR=1.06 (0.79, 1.41)
HR=0.93 (0.88, 0.99)
HR=0.98 (0.69, 1.38)
HR=0.99 (0.92, 1.07)
HR=0.97 (0.72, 1.31)
HR=1.02 (0.88, 1.18)
HR=0.74 (0.43, 1.26)
HR=0.92 (0.81, 1.04)
HR=0.98 (0.88, 1.10)
HR=1.04 (0.96, 1.13)
HR=0.66 (0.43, 1.02)
HR=0.90 (0.69, 1.16)
HR=0.96 (0.75, 1.22)
HR=0.99 (0.94, 1.05)
HR=0.72 (0.48, 1.09)
HR=0.77, 0.49, 1.22)
HR=0.99 (0.86, 1.15)
HR=1.09 (0.97, 1.21)
HR=0.99 (0.53, 1.85)
HR=1.69 (0.93, 3.07)
HR=1.43 (0.76, 2.69)

Community
Occupational
0.90 (0.75, 1.09) 0.73 (0.55, 0.98)
1.02 (0.68, 1.52) 0.73 (0.32, 1.67)
0.95 (0.89, 1.01) 1.03 (0.59, 1.79)
1.02 (0.72, 1.43) one case
0.98 (0.89, 1.09) 1.08 (0.84, 1.39)
1.01 (0.67, 1.52) 1.17 (0.19, 7.36)
0.92 (0.75, 1.13) 1.30 (0.78, 2.18)
0.53 (0.21, 1.34) one case
0.89 (0.76, 1.05) 1.04 (0.68, 1.58)
1.02 (0.89, 1.17) 1.10 (0.73, 1.65)
1.02 (0.92, 1.14) 0.93 (0.73, 1.18)
0.77 (0.47, 1.27) one case
0.94 (0.73, 1.22) no cases
0.98 (0.72, 1.34) 1.14 (0.33, 3.89)
0.98 (0.90, 1.06) 0.98 (0.83, 1.16)
0.64 (0.36, 1.13) 0.91 (0.25, 3.33)
0.74 (0.41, 1.31) one case
0.99 (0.84, 1.16) 0.96 (0.42, 2.18)
1.11 (0.96, 1.29) 0.99 (0.67, 1.46)
0.94 (0.45, 1.99) 1.22 (0.28, 5.30)
1.08 (0.52, 2.25) 3.27 (0.76, 14.1)
1.50 (0.72, 3.13) 0.99 (0.21, 4.68)

Testicular Cancer
Q2
Q3
Q4

HR=1.28 (0.95, 1.73)
HR=0.87 (0.15, 4.88)
HR=1.08 (0.20, 5.90)
HR=2.36 (0.41, 13.7)

1.53 (1.09, 2.15)
0.98 (0.13, 7.14)
1.54 (0.19, 12.2)
4.66 (0.52, 41.6)

1.61 (0.21, 12.2)
two cases total
in the cohort

Thyroid Cancer
Q2
Q3
Q4

HR=1.04 (0.89, 1.20)
HR=2.06 (0.93, 4.56)
HR=2.02 (0.90, 4.52)
HR=1.51 (0.67, 3.39)

1.00 (0.84 (1.20)
2.09 (0.91, 4.82)
1.92 (0.82, 4.50)
1.42 (0.60, 3.37)

1.12 (0.61, 2.05)
1.65 (0.09, 31.5)
4.52 (0.10, 198)
5.85 (0.13, 257)

126

Reference
Vieira 2013
C8 Health Project

Exposure
PFOA

Outcome
Cancers©:
Breast Cancer
Kidney Cancer
NHL
Ovarian Cancer
Prostate Cancer
Testicular Cancer

Odds Ratio & 95% CI
OR for the Little Hocking system
OR=1.2 (0.8, 2.0)
OR=1.7 (0.9, 3.3)
OR=1.6 (0.9, 2.8)
OR=1.8 (0.7, 4.4)
OR=1.4 (0.9, 2.3)
OR=5.1 (1.6, 15.6)

Odds Ratios & 95% CI
OR for Very High serum level category
OR=1.4 (0.9, 2.3) (non-monotonic trend)
OR=2.0 (1.0, 3.9) (non-monotonic trend)
OR=1.8 (1.0, 3.4) (non-monotonic trend)
OR=2.1 (0.8, 5.5) (non-monotonic trend)
OR=1.5 (0.9, 2.5) (non-monotonic trend)
OR=2.8 (0.8, 9.2) (non-monotonic trend)

Reference

Exposure*

Outcome

SMR & 95% CI

Cumulative exposure (SMR)

Consonni 2013
TFE synthesis &
polymerization
workers (including
the WV Dupont
plant workers and
6 other production
sites in NJ and
Europe)

APFO (PFOA)

Median serum
level:
PFOA=28.2 ng/ml
Estimate PFOA
median serum
level in 1995 at
Little Hocking
WD =125 ng/ml

APFO (PFOA)

Mortality:
Esophageal cancer
Liver cancer
Pancreatic cancer
Lung cancer
Kidney/other urinary cancers
Leukemias
Stomach cancer
Colon cancer
Rectal cancer
Laryngeal cancer
Prostate cancer
Testicular cancer
Bladder cancer
Brain cancer
NHL

Multiple myeloma
Diabetes mellitus
Circulatory diseases
Respiratory diseases
Liver cirrhosis
Nephritis, nephrosis

1.44 (0.72, 2.57)
1.43 (0.57, 2.94)
1.05 (0.51, 1.94)
0.73 (0.54, 0.97)
1.69 (0.81, 3.11)
1.61 (0.80, 2.88)
0.52 (0.17, 2.21)
0.48 (0.19, 0.99)
1.03 (0.38, 2.25)
0.76 (0.09, 2.75) (2 cases)
0.24 (0.05, 0.70) (3 cases)
1.35 (0.03, 7.49) (1 case)
0.55 (0.11, 1.60) (3 cases)
0.64 (0.17, 1.63)
0.79 (0.26, 1.84)

0.66 (0.08, 2.39) (2 cases)
0.57 (0.23, 1.17)
0.88 (0.77, 1.00)
0.63 (0.42, 0.89)
1.00 (0.60, 1.56)
0.92 (0.25, 2.37)

127

No Exp.
0
0.72
1.66
0.75
0
0.79

Low Med
1.62 1.54
0.70 1.25
0 1.30
0.91 0.75
1.57 1.50
1.64 1.35

High
1.16
2.14
1.84
0.54
2.00
1.85

No Exp. Low Med

High

1.19
0

0.52
0.67

1.49
0.67

0.92
1.49

* Exposures are occupational unless otherwise noted.
€

The cohort evaluated in the Lundin 2009 study is the same cohort included in the Raleigh 2014 study. The Lundin 2009 study is included in the table
because it provides additional information that can be used to interpret the more recent Raleigh 2014 study.
¥
Reference rates are from other Dupont workers.
≠
Exposure was not lagged.
£
sumPFSA: sum of PFOS, PFHxS and PFOSA
±
sumPFCA: sum of PFHpA, PFOA, PFNA, PFDA, PFUnA, PFDoA and PFTrA
∆
The hazard ratio is per unit of log estimated cumulative PFOA serum concentration (ng/ml). For cancers of the kidney, testes and thyroid, hazard ratios
are also provided for quartiles of cumulative PFOA serum concentration with the first quartile as referent.
©
Except for kidney cancer, these are cancers with ORs at Little Hocking ≥ the ORs for the other water systems and also with elevated ORs in the very
high PFOA serum exposure category. For kidney cancer, the OR for the Tuppers Plains system was 2.0 compared to the OR for Little Hocking of 1.7.
Other cancers not listed were not elevated in the Little Hocking system and/or were not elevated in the very high serum category.

Mortality: Alexander 2003, Grice 2007, Alexander 2007, Lundin 2009, Steenland & Woskie 2012, Consonni 2013, Raleigh 2014
Incidence: Grice 2007, Alexander 2007, Eriksen 2009, Barry 2013, Vieira 2013, Hardell 2014, Raleigh 2014, Steenland 2015

128

Table A3. Other Adult Diseases
Reference,
Location

Study
population

Dhingra 2016a
C8

Dhingra 2016b
C8

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome

Difference detected (95% CI)

28,240 ≥20 years of
age

28.2

Chronic kidney disease
397 cases (retrospective)
212 cases (prospective)

29,641 ≥20 years of
age

28.2

Estimated glomerular
filtration rate (eGFR)
(mL/min/1.73 m2);
Earlier menopause
(reported as of
2005/2006); and
Reverse causation

Hazard ratios for quintiles of cumulative exposure
Retrospective
Prospective
1.36 (0.89, 2.09)
2nd: 1.26 (0.90, 1.75)
0.94 (0.62, 1.45)
3rd: 1.12 (0.80, 1.55)
1.08 (0.70, 1.66)
4th: 1.12 (0.81, 1.56)
5th: 1.24 (0.88, 1.75)
1.12 (0.72, 1.75)
Measured serum PFOA quintiles
eGFR (β±S.E.)
Menopause (OR, 95% CI)
1.68 (1.21, 2.35)
2nd: -0.64 ± 0.268
1.45 (1.04, 2.02)
3rd: -1.03 ± 0.269
1.39 (1.00, 1.93)
4th: -0.84 ± 0.271
1.58 (1.14, 2.19)
5th: -0.98 ± 0.274

6,342 women, aged
30-65, who had not
had a hysterectomy

Steenland 2010
C8

53,458 aged ≥20
years

20.2

27.9

Uric Acid mg/dL

Hyperuricemia

Shankar 2011
NHANES, 19992000; 2003-2006

3,883 aged ≥20 years

17.9

3.5

Uric Acid mg/dL

hyperuricemia

129

Modeled serum PFOA quintiles
eGFR (β±S.E.)
Menopause (OR, 95% CI)
0.98 (0.70, 1.37)
2nd: -0.08 ± 0.268
1.05 (0.75, 1.45)
3rd: 0.37 ± 0.268
0.78 (0.56, 1.08)
4th: 0.21 ± 0.269
5th: 0.23 ± 0.271
0.92 (0.65, 1.30)
Highest decile serum PFOA: 0.28 ± 0.02
Highest decile serum PFOS: 0.22 ± 0.02
Monotonic exposure-response for PFOA and PFOS
Highest quintile serum PFOA: OR=1.47 (1.37, 1.58)
Highest quintile serum PFOS: OR=1.26 (1.17, 1.35)
Monotonic exposure-response for PFOA and PFOS
4th quartile serum PFOA: 0.44 (0.32, 0.56)
4th quartile serum PFOS: 0.27 (0.13, 0.41)
Monotonic exposure-response for PFOA and PFOS
4th quartile serum PFOA: OR=1.97 (1.44, 2.70)
4th quartile serum PFOS: OR=1.48 (0.99, 2.22)
Monotonic exposure-response for PFOA

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Gleason 2015
NHANES, 20072010

4,333 aged ≥12 years

11.3

1.8

3.7

Gallo 2012
C8

46,452 ≥18 years

20.3

Outcome

Difference detected (95% CI)

Hyperuricemia

Serum PFAS quartiles, odds ratios
PFOA
PFOS
PFHxS
2nd 1.46 (1.16, 1.85) 1.18 (0.88, 1.56) 0.83 (0.64, 1.09)
3rd 1.74 (1.35, 2.25) 1.09 (0.81, 1.47 1.14 (0.88, 1.48)
4th 1.88 (1.37, 2.58) 1.20 (0.88, 1.63) 1.13 (0.89, 1.43)

Elevated ALT

2nd 1.42 (1.11, 1.83) 1.29 (0.93, 1.78) 1.35 (1.01, 1.80)
3rd 1.55 (1.14, 2.10) 1.26 (0.88, 1.81) 1.37 (1.06, 1.77)
4th 1.51 (1.18, 1.94) 1.23 (0.87, 1.74) 1.18 (0.94, 1.49)

Elevated GGT

2nd 1.09 (0.79, 1.52) 1.16 (0.87, 1.55) 0.97 (0.80, 1.81)
3rd 1.11 (0.80, 1.52) 1.14 (0.87, 1.51) 1.02 (0.81, 1.30)
4th 1.34 (1.00, 1.80) 1.10 (0.82, 1.47) 0.91 (0.75, 1.11)

Elevated AST

2nd 1.31 (1.03, 1.65) 1.03 (0.78, 1.36) 1.29 (0.97, 1.71)
3rd 1.26 (0.97, 1.64) 1.14 (0.87, 1.50) 1.29 (0.96, 1.73)
4th 1.39 (1.06, 1.80) 0.91 (0.69, 1.21) 1.30 (0.94, 1.78)

Total Bilirubin

2nd 1.31 (0.98, 1.75) 1.44 (1.12, 1.84) 1.10 (0.86, 1.40)
3rd 1.66 (1.19, 2.32) 1.65 (1.25, 2.18) 1.40 (1.04, 1.88)
4th 1.68 (1.31, 2.14) 1.51 (1.06, 2.15) 1.32 (0.97, 1.81)

28.0
Ln-ALT (IU/L)
Ln-GGT (IU/L)
Ln-Direct bilirubin
(mg/dL)
Elevated ALT
Elevated GGT
Elevated Direct bilirubin

130

Ln-PFOA
Ln-PFOS
β = 0.022 (0.018, 0.025) β = 0.020 (0.014, 0.026)
β = 0.015 (0.010, 0.019) β = 0.008 (-0.000, 0.016)
β = 0.001 (-0.002, 0.004) β = 0.029 (0.024, 0.034)
tenth decile odds ratios;
1.54 ( 1.33, 1.78)
1.25 (1.08, 1.44)
1.06 (0.92, 1.20)
0.94 (0.83, 1.07)
1.01 (0.66, 1.53)
1.23 (0.82, 1.83)
No monotonic exposure-response relationships

Reference,
Location

Study
population

Darrow 2016
C8

30,723 aged ≥20
years

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome

Difference detected (95% CI)

Ln-ALT (IU/L)
Ln-GGT (IU/L)
Ln-Direct bilirubin
(mg/dL)

Fifth quintile, estimated cumulative serum PFOA
β = 0.058 (0.040, 0.076)
β = 0.020 (-0.004, 0.044)
β = -0.017 (-0.032, -0.001)
monotonic exposure-response for ALT

28.2

Elevated ALT
Elevated GGT
Elevated Direct bilirubin

Ever reported:
Arthritis
Asthma
COPD
Diabetes
Heart Disease
Liver disease (current)

5th quintile cumulative exposure:
OR=0.95 (0.70, 1.27)
4th quartile serum concentrations, ORs:
PFOA
PFOS
1.28 (0.97, 1.68)
0.74 (0.53, 1.04)
0.93 (0.64, 1.36)
0.79 (0.50, 1.26)
0.85 (0.54, 1.34)
0.58 (0.43, 0.76)
0.69 (0.41, 1.16)
0.87 (0.57, 1.31)
1.08 (0.70, 1.69)
0.91 (0.50, 1.64)
0.61 (0.21, 1.78)
0.95 (0.39, 2.29)

Women:
Thyroid disease ever
Thyroid disease with
current medication

1.64 (1.09, 2.46)
1.15 (0.70, 1.91)
1.86 (1.12, 3.09)
1.31 (0.72, 2.36)
Exposure-response trends were non-monotonic

Men:
Thyroid disease ever
Thyroid disease with
current medication

1.58 (0.74, 3.39)
1.58 (0.72, 3.47)
1.89 (0.60, 5.90)
1.89 (0.72, 4.93)
Exposure-response trends were non-monotonic

ALT (IU/L)
Ln-GGT (IU/L)
Total bilirubin (µM)

Regression coefficient (s.e.) per unit increase in log serum PFAS
PFOA
PFOS
PFHxS
1.86 (0.62)
1.01 (0.53)
0.19 (0.48)
0.08 (0.03)
0.01 (0.03)
-0.00 (0.02)
-0.09 (0.20)
-0.30 (0.24)
0.38 (0.20)

Any liver disease
Melzer 2010
NHANES, 19992000, 2003-2006

Lin 2010
NHANES, 19992000, 2003-2004

3,974 aged ≥20 years

17.9

Fifth quintile, estimated cumulative serum PFOA
OR=1.16 (1.02, 1.33)
OR=0.96 (0.85, 1.09)
OR=0.95 (0.66, 1.37)
No monotonic exposure-response relationships

3.5

2,216 aged ≥18 years

131

Reference,
Location

Study
population

Winquist 2014b
C8

32,254 aged ≥20
years

Mattsson 2015
Sweden: male
cohort of farmers
and rural residents

231 cases of coronary
heart disease and 231
controls

Shankar 2012
NHANES, 19992000, 2003-2004

1,216 aged ≥40 years

Steenland 2009
C8

46,294 aged ≥18
years

PFOS
serum
level

22.4

19.6

PFHxS
serum
level

1.6

PFOA
serum
level

Outcome

Difference detected (95% CI)

26.1

Hypertension
Hypercholesterolemia
Men 40-59
Coronary artery disease

4.1

Coronary heart disease

4.2

Cardiovascular disease
Peripheral arterial
disease
Either CVD or PAD

Fifth quintile, cumulative serum PFOA, HR=0.98 (0.91, 1.06)
Fifth quintile, cumulative serum PFOA, HR=1.19 (1.11, 1.28)
Fifth quintile, cumulative serum PFOA, HR=1.44 (1.28, 1.62)
Fifth quintile, cumulative serum PFOA, HR=1.07 (0.93, 1.23)
None of the analyses had a monotonic trend.
Quartile of serum PFAS, Odds ratios (95% CI):
PFOS
PFOA
PFHxS
2nd 0.82 (0.46, 1.45) 0.79 (0.44, 1.43) 0.91 (0.51, 1.63)
3rd 1.30 (0.74, 2.26) 1.18 (0.67, 2.06) 1.00 (0.56, 1.77)
4th 1.07 (0.60, 1.92) 0.88 (0.50, 1.55) 0.95 (0.54, 1.67)
Serum PFOA level, Odds ratios (95% CI)
CVD
PAD
Either CVD or PAD
2nd 1.58 (0.80, 3.12) 0.75 (0.37, 1.52) 1.41 (0.81, 2.45)
3rd 1.77 (1.04, 3.02) 1.18 (0.47, 2.96) 1.72 (1.13, 2.64)
4th 2.01 (1.12, 3.60) 1.78 (1.03, 3.08) 2.28 ( 1.40, 3.71)

26.6

Coronary heart disease
Stroke
Log total cholesterol

Log total cholesterol
Log HDL
Log LDL
Log triglycerides
Log total
cholesterol/HDL

High total cholesterol
(≥240 mg/dL)

132

Coronary heart disease
Stroke
4.39 (1.44, 13.4)
2nd 0.90 (0.37, 2.23)
3.94 (1.48, 10.5)
3rd 1.90 (0.89, 4.08)
4th 2.24 (1.02, 4.94)
4.26 (1.84, 9.89)
10th decile, serum PFAS: change in the log total cholesterol, (SE)
PFOA: 0.05 (0.004)
PFOS: 0.06 (0.004)
(equivalent to an increase of 11-12 mg/dL in total cholesterol)
Linear regression coefficient (SD)
Log serum PFOA
Log serum PFOS
0.01112 (0.00076)
0.02660 (0.00140)
0.00276 (0.00094)
0.00355 (0.00173)
0.01499 (0.00121)
0.04176 (0.00221)
0.00169 (0.00219)
0.01998 (0.00402)
0.00831 (0.00110)
0.02290 (0.00202)
PFOA (OR, 95% CI) PFOS
1.14 (1.05, 1.23)
2nd 1.21 (1.12, 1.31)
1.28 (1.19, 1.39)
3rd 1.33 (1.23, 1.43)
4th 1.38 (1.28, 1.50)
1.51 (1.40, 1.64)

Reference,
Location

Study
population

PFOS
serum
level

Fitz-Simon 2013
C8 (longitudinal
study)

560 aged >20 years

8.2

Nelson 2010
NHANES, 20032004

860 aged ≥20 years

PFHxS
serum
level

PFOA
serum
level

Outcome

30.8
Total cholesterol
LDL
HDL
Triglycerides

21.0

1.8

3.9
Total Cholesterol

Fisher 2013
Canadian Health
Measures Survey

2,700 aged ≥18 years
2,345

8.4

Fu 2014
China

133 aged 0 to 88
years

1.47

2.18

Difference detected (95% CI)

Percentage decrease in lipid per halving of serum PFAS
PFOA: 1.65 (0.32, 2.97) PFOS: 3.20 (1.63, 4.76)
PFOA: 3.58 (1.47, 5.66) PFOS: 4.99 (2.46, 7.44)
PFOA: 1.33 (-0.21, 2.85) PFOS: 1.28 (-0.59, 3.12)
PFOA: -0.78 (-5.34, 3.58) PFOS: 2.49 (-2.88, 7.57)
Mean difference in lipid
PFOS
PFOA
PFHxS
2nd 6.12 (-4.45, 16.7) 5.40 (-2.11, 12.9) -3.22 (-11.8, 5.30)
3rd 5.07 (-4.24, 14.4) 7.50 (-3.71, 18.7) -2.27 (-8.95, 4.41)
4th 13.42 (3.83, 23.0) 9.76 (-0.23, 19.7) -7.01 (-13.2, -0.79)

LDL cholesterol

2nd 2.78 (-12.7, 18.3) 6.07 (-8.65, 20.8) -4.49 (-12.4, 3.40)
3rd 0.38 (-9.64, 10.4) 0.71 (-12.6, 14.0) -4.07 (-13.4, 5.28)
4th 8.50 (-7.10, 24.1) 2.94 (-10.8, 16.7) -9.67 (-20.1, 0.71)

Non-HDL cholesterol

2nd 4.64 (-6.87, 16.2) 7.41 (-0.97, 15.8) -3.94 (-12.2, 4.37)
3rd 4.97 (-5.30, 15.2) 9.11 (-2.44, 20.7) -2.02 (-9.32, 5.28)
4th 12.55 (1.62, 23.5) 11.03 (1.20, 20.9) -9.32 (-15.9, -2.77)
Odds ratios (95% CI)
PFOA
PFOS
PFHxS
2nd 1.61 (1.02, 2.53) 0.97 (0.58, 1.62) 1.05 (0.69, 1.61)
3rd 1.26 (0.76, 2.07) 0.94 (0.58, 1.54) 1.43 (0.85, 1.40)
4th 1.50 (0.86, 2.62) 1.36 (0.87, 2.12) 1.57 (0.93, 2.64)
Odds ratios PFOA
PFOS
2nd
0.82 (0.14, 4.81)
0.57 (0.12, 2.81)
2.60 (0.56, 12.1)
0.82 (0.17, 3.91)
3rd
0.55 (0.09, 3.31)
2.27 (0.47, 10.9)
4th

2.46
High cholesterol

1.43
High total cholesterol

High LDL cholesterol

2nd
3rd
4th

0.55 (0.11, 2.82)
1.70 (0.40, 3.49)
0.71 (0.14, 3.49)

1.06 (0.25, 4.53)
1.11 (0.25, 4.82)
2.27 (0.50, 10.4)

High Triglycerides

2nd
3rd
4th

1.73 (0.57, 5.21)
1.03 (0.33, 3.20)
1.97 (0.59, 6.55)

0.52 (0.17, 1.56)
0.93 (0.31, 2.80)
1.26 (0.41, 3.90)

133

Reference,
Location

Study
population

Winquist 2014a
C8

32,254 aged ≥20
years (community &
worker cohort)

28,541 (community
cohort only)

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome

Difference detected (95% CI)

Functional thyroid
disease

Quintiles of serum PFOA, cumulative exposure, hazard ratios
All
Females
Males
1.12 (0.69, 1.79)
2nd 1.21 (1.01, 1.45) 1.24 (1.02, 1.51)
0.83 (0.51, 1.37)
3rd 1.17 (0.97, 1.41) 1.27 (1.04, 1.55)
4th 1.27 (1.06, 1.52) 1.36 (1.12, 1.66) 1.01 (0.63, 1.62)
5th 1.28 (1.06, 1.53) 1.37 (1.11, 1.68) 1.05 (0.66, 1.66)

26.1

Hyperthyroidism

2nd
3rd
4th
5th

0.94 (0.62, 1.42)
1.12 (0.75, 1.68)
1.22 (0.82, 1.82)
1.20 (0.80, 1.81)

1.04 (0.65, 1.67)
1.33 (0.84, 2.11)
1.45 (0.92, 2.28)
1.39 (0.86, 2.26)

Hypothyroidism

2nd
3rd
4th
5th

1.31 (1.06, 1.63)
1.27 (1.01, 1.58)
1.30 (1.04, 1.62)
1.40 (1.12, 1.75)

1.32 (1.04, 1.67) 1.43 (0.77, 2.66)
1.33 (1.05, 1.69) 1.12 (0.59, 2.14)
1.34 (1.06, 1.70) 1.32 (0.71, 2.45)
1.47 (1.15, 1.88) 1.36 (0.74, 2.48)

Functional thyroid
disease

2nd
3rd
4th
5th

1.24 (1.03, 1.49) 1.24 (1.02, 1.52)
1.21 (1.00, 1.46) 1.27 (1.04, 1.56)
1.32 (1.09, 1.59) 1.36 (1.11, 1.66)
1.36 (1.11, 1.66) 1.36 (1.10, 1.69)

1.06 (0.62, 1.80)
0.80 (0.46, 1.40)
1.02 (0.59, 1.75)
1.21 (0.69, 2.12)

Hyperthyroidism

2nd
3rd
4th
5th

0.96 (0.62, 1.47)
1.13 (0.74, 1.73)
1.23 (0.81, 1.88)
1.28 (0.81, 2.01)

1.07 (0.66, 1.73)
1.36 (0.85, 2.18)
1.45 (0.91, 2.33)
1.45 (0.88, 2.40)

0.60 (0.22, 1.64)
0.47 (0.17, 1.32)
0.57 (0.21, 1.58)
0.70 (0.24, 2.06)

Hypothyroidism

2nd
3rd
4th
5th

1.34 (1.07, 1.68)
1.31 (1.04, 1.64)
1.35 (1.07, 1.70)
1.49 (1.17, 1.89)

1.31 (1.03, 1.66) 1.53 (0.76, 3.11)
1.32 (1.04, 1.68) 1.26 (0.60, 2.61)
1.33 (1.04, 1.69) 1.55 (0.75, 3.18)
1.43 (1.10, 1.84) 1.89 (0.90, 3.97)

community cohort only:
Functional thyroid
disease
Hyperthyroidism
Hypothyroidism

Prospective Analyses: cumulative exposure, quintile 5 results:

0.71 (0.29, 1.74)
0.57 (0.23, 1.43)
0.70 (0.30, 1.66)
0.74 (0.33, 1.65)

24.2

134

1.13 (0.83, 1.54) 0.97 (0.68, 1.37) 1.88 (0.95, 3.75)
1.81 (0.77, 4.25) 1.71 (0.68, 4.30) 2.26 (0.23, 22.1)
1.09 (0.75, 1.59) 0.90 (0.59, 1.39) 2.08 (0.90, 4.80)
Non-monotonic trends in the prospective analyses.

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Starling 2014
Norway

891 pregnant women,
aged 19-44

13.03

0.60

2.25

Outcome

Total cholesterol

per ln-ng/ml PFAS
LDL cholesterol

per ln-ng/ml PFAS
Ln-triglycerides

Knox 2011
C8

50,113 aged ≥20
years

Median or geometric mean
values for PFOA and PFOS
were not provided

per ln-ng/ml PFAS
Thyroxine (total T4)
T3 uptake
Thyroid stimulating
hormone (TSH)
Serum albumin

Difference detected (95% CI)

Linear regression coefficients for lipids (mg/dL)
PFOA
PFOS
PFHxS
2nd 1.49 (-6.49, 9.48) -3.35 (-10.3, 3.64) 0.65 (-6.87, 8.17)
3rd 3.54 (-4.51, 11.6) 3.06 (-4.93, 11.1) 1.62 (-6.08, 9.32)
4th 3.90 (-5.00, 12.8) 7.59 (-0.42, 15.6) 4.25 (-3.88, 12.4)
2.58 (-4.32, 9.47) 8.96 (1.70, 16.2) 3.00 (-1.75, 7.76)
2nd 0.94 (-6.08, 7.96)
3rd 4.16 (-3.19, 11.5)
4th 3.35 (-4.35, 11.1)
2.25 (-3.97, 8.48)

-3.23 (-9.28, 2.83) 0.44 (-6.19, 7.08)
2.60 (-4.49, 9.70) 0.50 (-6.15, 7.16)
5.51 (-1.62, 12.6) 1.48 (-5.89, 8.85)
6.48 (-0.07, 13.0) 1.92 (-2.50, 6.33)

2nd 0.03 (-0.04, 0.11) 0.00 (-0.06, 0.07) -0.04 (-0.11, 0.02)
3rd 0.01 (-0.08, 0.09) -0.03 (-0.10, 0.05) -0.02 (-0.10, 0.05)
4th -0.04 (-0.12, 0.04) 0.00 (-0.07, 0.07) -0.02 (-0.09, 0.05)
0.00 (-0.07, 0.06) -0.02 (-0.09, 0.04) -0.01 (-0.05, 0.03)
No confidence intervals were provided. (Many p-values were presented
as less than a value, so CIs could not be estimated.)
PFOA results:
PFOA was associated with increased thyroxine and decreased T3 uptake
in women of all ages and in men who were >50 years of age. PFOA was
associated with a slight increase in albumin among all participants. T3
uptake was lower in women than in men; TSH was lower in men than in
women; T4 was higher in women than in men; and albumin was higher
in men.
PFOS results:
PFOS was associated with increased thyroxine, decreased T3 uptake, and
increased albumin in all participants. Thyroxine was higher in women
than in men; T3 update was lower in women than in men; and albumin
was lower in women than in men.

135

Reference,
Location

Study
population

Wen 2013
NHANES, 20072010

1,181 aged ≥20 years
672 males
509 females

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

17.14
11.09

2.60
1.41

4.70
3.53

Outcome

Total T4 (µg/mL)
Ln-free T4 (ng/dL)
Total T3 (ng/dL)
Ln-free T3 (pg/mL)
Ln TSH (mIU/L)Ln
Thyroglobulin (ng/mL)
Total T4 (µg/mL)
Ln-free T4 (ng/dL)
Total T3 (ng/dL)
Ln-free T3 (pg/mL)
Ln TSH (mIU/L)
Ln Thyroglobulin
(ng/mL)
Subclinical
hypothyroidism: Men:
Women:
Subclinical
hyperthyroidism: Men:
Women:

Webster 2016
NHANES, 20072008

1,525 aged ≥18 years

14.2

2.0

4.2

Free T3
Free T4
Free T3/Free T4
Thyroid stimulating
hormone (TSH)
Total T3
Total T4
Shrestha 2015
Hudson River area,
NY

87 aged 55-74

29.8

9.3
Thyroid stimulating
hormone (TSH)
Free T4
Total T4
Total T3

136

Difference detected (95% CI)

Regression coefficient per unit increase in log PFAS, Men:
PFOA
PFOS
PFHxS
0.000 (-0.280, 0.280) -0.020 (-0.223, 0.183) -0.032 (-0.175, 0.111)
-0.010 (-0.041, 0.022) -0.009 (-0.034, 0.017) -0.016 (-0.029, -0.003)
0.775 (-3.048, 4.598) -1.111 (-3.856, 1.634) -0.081 (-1.698, 1.536)
0.013 (-0.004, 0.031) 0.002 (-0.008, 0.012) 0.005 (-0.003, 0.012)
0.004 (-0.081, 0.090) 0.003 (-0.070, 0.076) 0.019 (-0.057, 0.524)
-0.096 (-0.258, 0.066) -0.047 (-0.149, 0.055) -0.049 (-0.185, 0.087)
Women:
0.082 (-0.369, 0.532) 0.087 (-0.143, 0.318) 0.260 ( 0.108, 0.413)
-0.004 (-0.047, 0.039) 0.009 (-0.019, 0.036) 0.003 (-0.024, 0.030)
6.628 (0.545, 12.712) 1.453 (-1.987, 4.891) 4.074 ( 2.232, 5.916)
0.016 (-0.018, 0.051) -0.007 (-0.024, 0.010) 0.003 (-0.021, 0.026)
-0.030 (-0.215, 0.154) -0.048 (-0.156, 0.060) -0.019 (-0.128, 0.090)
0.095 (-0.111, 0.302) 0.135 (-0.007, 0.277) -0.018 (-0.122, 0.087)
Odds ratio
1.29 (0.40, 4.10)
7.42 (1.14, 48.1)

1.98 (1.19, 3.28)
3.03 (1.14, 8.07)

1.57 (0.76, 3.25)
3.10 (1.22, 7.86)

0.38 (0.16, 0.95)
0.92 (0.19, 4.46)
0.56 (0.24, 1.20)
0.99 (0.13, 7.59)
1.90 (0.53, 6.80)
2.27 (1.07, 4.80)
Low iodine & high thyroid peroxidase antibody (TPOAb):
percent difference per interquartile ratio increase in log serum PFAS
PFOS
PFOA
PFHxS
4.7 (3.9, 5.5)
4.8 (3.7, 5.8)
3.9 (2.3, 5.5)
-4.4 (-7.6, -1.1)
-2.7 (-6.1, 0.8)
-8.3 (-15.8, -0.2)
9.5 (5.8, 13.2)
7.7 (3.6, 12.0)
13.3 (4.4, 22.9)
17.1 (6.6, 17.7)
16.2 (5.1, 28.5)
27.3 (0.7, 60.9)
12.0 (6.7, 17.7)
12.4 (7.0, 18.1)
13.8 (6.0, 22.1)
2.5 (-1.3, 6.5)
3.9 (-0.3, 8.3)
1.8 (-3.9, 7.8)
Regression coefficient, log serum PFAS
PFOS
PFOA
0.129 (-0.023, 0.281)
0.102 (-0.047, 0.250)
0.054 (0.002, 0.106)
0.016 (-0.036, 0.069)
0.766 (0.327, 1.205)
0.380 (-0.070, 0.830)
2.631 (-2.248, 7.510)
3.032 (-1.725, 7.789)

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Ji 2012
Korea

633 aged >12 years

7.96

1.51

2.74

Wang 2014
Taiwan

285 pregnant women
(average age=29)

12.73

Webster 2014
Canada

152 pregnant women
without thyroid
disease, aged ≥25
years

4.8

375 pregnant women
aged 18-43

8.03

Berg 2015
Norway

0.81

1.0

0.44

Outcome

Difference detected (95% CI)

Thyroid stimulating
hormone (TSH)
Total T4

Regression coefficients, serum PFAS
PFOS
PFOA
PFHxS
0.062 (-0.069, 0.192) -0.066 (-0.220, 0.089) 0.013 (-0.094, 0.120)
-0.021 (-0.048, 0.005) -0.020 (-0.051, 0.012) -0.007 (-0.029, 0.015)

Thyroid stimulating
hormone (TSH)
Free T4
Total T4
Total T3

Regression coefficients, serum PFAS
PFOS
PFOA
-0.005 (-0.024, 0.013) 0.011 (-0.057, 0.078)
0.001 (-0.002, 0.003) -0.003 (-0.012, 0.005)
0.019 (-0.016, 0.053) 0.011 (-0.108, 0.130)
0.000 (-0.002, 0.001) -0.000 (-0.002, 0.009)

Thyroid stimulating
hormone (TSH)
Free thyroxine (fT4)

Percent change (compared to the median thyroid function in the study
population) per interquartile increase in PFAS among those with high
TPOAb: PFOS
PFOA
PFHxS
69% (15%, 123%) 54% (8%, 100&) 2% (-45%, 48%)
-7% (-18%, 3%)
-4% (-14%, 5%)
-5% (-15%, 4%)

2.39

1.7

1.53

Thyroid stimulating
hormone (TSH) mIU/L
Subclinical
hypothyroidism (%)

Berg 2016
Norway

370 pregnant women
aged 18-43

8.03

Kato 2016
Japan

392 pregnant women

5.2

0.44

1.53

Thyroid stimulating
hormone (TSH) mIU/L

1.2
Ln-Thyroid stimulating
hormone (TSH)
Ln-Free Thyroxine (T4)

137

PFHxS
0.105 (0.002, 0.207)
-0.010 (-0.023, 0.003)
-0.130 (-0.316, 0.057)
-0.002 (-0.005, 0.001)

Mean difference in TSH, and proportion with subclinical hypothyroidism
per quartile of PFOS
1st 12.8% (n=12)
17.8% (n=16)
2nd 0.18 (0.06, 0.31)
25.3% (n=24)
3rd 0.26 (0.13, 0.40)
4th 0.35 (0.21, 0.50)
31.3% (n=30)
Percent change in TSH per quartile of PFOS
2nd 4 (-3.1, 11.4)
3rd 8 (0.6, 15.4)
4th 10 (1.6, 16.9)
Regression coefficient per Ln-PFAS (confidence intervals are estimated)
PFOS
PFOA
-0.214 (p-value presented as <0.001) 0.039 (-0.067, 0.144)
0.061 (-0.039, 0.161)
0.004 (-0.102, 0.110)

Reference,
Location

Study
population

Steenland 2013
C8

28,541 community
cohort
3,713 worker cohort
Total=32, 254 aged
≥20 years

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome

Difference detected (95% CI)

Ulcerative colitis

10 year lagged cumulative PFOA quartiles, RR:
Retrospective:
Prospective:
1.21 (0.43, 3.34)
2nd 1.71 (0.89, 3.27)
2.16 (0.80, 5.81)
3rd 2.05 (1.07, 3.91)
1.51 (0.43, 4.30)
4th 3.05 (1.56, 5.96)

24
113
26
Rheumatoid arthritis

Crohn’s disease
Type 1 diabetes
Lupus
Multiple sclerosis

0.31 (0.14, 0.71)
2nd 1.53 (0.61, 2.58)
0.90 (0.41, 0.72)
3rd 1.73 (1.10, 2.71)
0.32 (0.14, 0.72)
4th 1.35 (0.87, 2.11)
10 year lagged cumulative PFOA quartiles, retrospective analyses, RR:
3rd Q
4th Q
2nd Q
0.80 (0.32, 1.99)
0.97 (0.36, 2.60)
0.69 (0.26, 1.82)
0.50 (0.05, 4.90)
1.32 (0.14, 12.4)
0.71 (0.07, 7.14)
0.79 (0.27, 2.34)
1.26 (0.40, 4.03)
0.61 (0.19, 1.91)
1.16 (0.54, 2.47)
1.62 (0.75, 3.52)
1.32 (0.61, 2.84)

Innes 2011
C8

49,432 aged ≥21
years

20.3

28.2

Osteoarthritis

PFOA
odds ratio
2nd 1.16 (1.03, 1.31)
3rd 1.21 (1.07, 1.36)
4th 1.42 (1.26, 1.59)
1.07 (1.04, 1.1) per 1-unit increment in ln-PFOA

Uhl 2013
NHANES, 20032008

4,102 aged 20-84

24.6
(mean)

5.4
(mean)

Osteoarthritis

PFOA All
2nd 1.32 (0,78, 2.23)
3rd 1.20 (0.72, 2.00)
4th 1.55 (0.99, 2.43)
Per ln-PFOA: 1.20 (0.96, 1.49)
PFOS
2nd 1.04 (0.58, 1.85)
3rd 1.99 (1.14, 3.49)
4th 1.77 (1.05, 2.96)
Per ln-PFOS: 1.15 (0.94, 1.40)

138

PFOS
0.91 (0.81, 1.01)
0.94 (0.84, 1.06)
0.76 (0.68, 0.85)

Female
1.44 (0.80, 2.62)
1.18 (0.67, 2.08)
1.98 (1.24, 3.19)
1.35 (1.02, 1.79)

Male
0.97 (0.42, 2.27)
0.98 (0.46, 2.08)
0.82 (0.40, 1.70)
0.89 (0.67, 1.19)

0.88 (0.46, 1.70)
1.92 (0.98, 3.75)
1.73 (0.97, 3.10)
1.22 (0.94, 1.58)

1.32 (0.41, 4.25)
1.86 (0.55, 6.25)
1.56 (0.54, 4.53)
0.95 (0.73, 1.23)

Reference,
Location

Study
population

PFOS
serum
level

Lin 2014
NHANES, 20052008

2,339 aged ≥20 years
1,192 men
1,147 women
842 not in
menopause
305 in menopause

15.32

PFHxS
serum
level

PFOA
serum
level

Outcome

3.96
Total Lumbar spine bone
mineral density (BMD)
(g/cm2)

Total hip BMD (g/cm2)

All fractures

Hip fracture

Wrist fracture

Spine fracture

139

Difference detected (95% CI)

Regression coefficient for change in BMD per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
0.006 (-0.014, 0.026) 0.001 (-0.020, 0.022)
0.018 (-0.014, 0.049)
Regression coefficient for change in BMD per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
0.000 (-0.013, 0.013) -0.022 (-0.038, -0.007)
-0.004 (-0.026, 0.034)
Regression coefficient for change in BMD per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
-0.002 (-0.023, 0.019) 0.008 (-0.010, 0.027)
0.022 (-0.011, 0.055)
Regression coefficient for change in BMD per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
-0.003 (-0.016, 0.010) 0.000 (-0.017, 0.017)
0.017 (-0.012, 0.047)
OR per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
0.84 (0.67, 1.07) 0.98 (0.75, 1.28)
1.53 (0.63, 3.74)
OR per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
0.92 (0.73, 1.16) 0.97 (0.75, 1.24)
1.59 (0.88, 2.86)
OR per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
0.64 (0.39, 1.06)
1.59 (0.57, 4.46)
0.48 (0.06, 4.16)
OR per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
1.07 (0.76, 1.52) 1.12 (0.62, 2.03)
0.83 (0.23, 3.00)
OR per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
1.12 (0.75, 1.70) 1.07 (0.65, 1.77)
1.21 (0.46, 3.13)
OR per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
1.09 (0.72, 1.66) 1.04 (0.63, 1.72)
1.22 (0.61, 2.45)
OR per unit-increase in ln-PFOA
Men
Women not in menopause Women in menopause
1.54 (0.85, 2.79) 1.83 (0.59, 5.61)
0.84 (0.46, 1.53)
OR per unit-increase in ln-PFOS
Men
Women not in menopause Women in menopause
1.27 (0.67, 2.42) 0.52 (0.15, 1.86)
1.12 (0.26, 4.78)

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome

Khalil 2016
NHANES, 20092010

1,914 aged ≥12 years

9.70
Means:
12.70
15.10
10.30

1.70
Means:
2.50
3.10
1.90

5.20
Means:
3.70
4.10
3.30

Total femur
PFOA Men
Women not in menopause Women in menopause
bone mineral
2nd -0.010 (-0.034, 0.055) -0.026 (-0.051, -0.001)
-0.011 (-0.059, 0.037)
-0.002 (-0.049, 0.045)
density (BMD) 3rd -0.012 (-0.056, 0.033) 0.006 (-0.041, 0.052)
-0.024 (-0.072, 0.024)
4th -0.001 (-0.042, 0.041) -0.029 (-0.068, 0.010)
Per Ln-PFOA -0.007 (-0.028, 0.014) -0.017 (-0.038, 0.004)
-0.012 (-0.043, 0.019)
PFOS Men
Women not in menopause Women in menopause
2nd -0.029 (-0.074, 0.016) -0.013 (-0.050, 0.023) -0.001 (-0.072, 0.069)
3rd -0.029 (-0.063, 0.006) -0.017 (-0.048, 0.014) 0.002 (-0.065, 0.070)
4th -0.032 (-0.072, 0.008) -0.013 (-0.046, 0.021) -0.059 (-0.115, -0.002)
Per Ln-PFOS -0.010 (-0.027, 0.006) -0.004 (-0.020, 0.012) -0.033 (-0.049, -0.015)
PFHxS Men
Women not in menopause Women in menopause
-0.013 (-0.050, 0.023)
-0.001 (-0.072, 0.069)
2nd -0.004 (-0.046, 0.038)
-0.017 (-0.048, 0.014)
0.002 (-0.065, 0.070)
3rd -0.029 (-0.063, 0.006)
-0.013 (-0.046, 0.021)
-0.059 (-0.115, -0.002)
4th -0.032 (-0.072, 0.008)
Per Ln-PFHxS -0.010 (-0.025, 0.004) -0.009 (-0.026, 0.007)
-0.009 (-0.029, 0.011)

956 men
959 women
590 premenopausal
368 postmenopausal

Difference detected (95% CI)

Total femur
neck mineral
density

PFOA Men
Women not in menopause
Women in menopause
2nd
0.011 (-0.021, 0.043) -0.028 (-0.060, 0.003)
-0.022 (-0.077, 0.033)
3rd -0.013 (-0.053, 0.028)
0.014 (-0.031, 0.060)
-0.024 (-0.083, 0.035)
0.004 (-0.035, 0.043) -0.019 (-0.056, 0.018)
-0.041 (-0.098, 0.016)
4th
Per Ln-PFOA 0.001 (-0.025, 0.022) -0.012 (-0.030, 0.007)
-0.020 (-0.049, 0.010)
PFOS Men
Women not in menopause Women in menopause
-0.005 (-0.087, 0.077)
2nd -0.036 (-0.077, 0.006) -0.005 (-0.028, 0.018)
-0.001 (-0.082, 0.080)
3rd -0.027 (-0.063, 0.009) -0.005 (-0.028, 0.017)
-0.062 (-0.134, 0.009
4th -0.046 (-0.078, -0.015) -0.001 (-0.029, 0.029)
Per Ln-PFOS -0.013 (-0.024, 0.002) -0.001 (-0.015, 0.015)
-0.033 (-0.049, -0.017)
PFHxS Men
Women not in menopause Women in menopause
-0.025 (-0.100, 0.050)
2nd -0.002 (-0.042, 0.038) -0.004 (-0.042, 0.033)
-0.017 (-0.092, 0.058)
3rd -0.004 (-0.031, 0.023) -0.010 (-0.018, 0.038)
-0.026 (-0.104, 0.051
4th -0.013 (-0.052, 0.025) -0.010 (-0.039, 0.016)
Per Ln-PFHxS -0.009 (-0.024, 0.006) -0.001 (-0.015, 0.013)
-0.005 (-0.024, 0.013)

Lumbar spine PFOA Men
Women not in menopause Women in menopause
bone mineral
2nd 0.013 (-0.042, 0.068) -0.008 (-0.041, 0.025) -0.001 (-0.089, 0.088)
density (BMD) 3rd -0.023 (-0.083, 0.037) 0.020 (-0.020, 0.060) 0.011 (-0.090, 0.113)
4th -0.005 (-0.058, 0.049) -0.010 (-0.042, 0.021) -0.017 (-0.111, 0.077)
Per Ln-PFOA -0.011 (-0.039, 0.017) 0.001 (-0.020, 0.021) -0.017 (-0.058, 0.024)

140

Lumbar spine PFOS Men
BMD
2nd -0.023 (-0.064, 0.018)
3rd -0.026 (-0.066, 0.014)
4th -0.023 (-0.064, 0.017)
Per Ln-PFOS -0.011 (-0.028, 0.006)
PFHxS Men
2nd 0.015 (-0.021, 0.050)
3rd 0.021 (-0.015, 0.057)
4th 0.005 (-0.022, 0.033)
Per Ln-PFHxS 0.001 (-0.011, 0.012)

Women not in menopause
0.001 (-0.044, 0.045)
0.009 (-0.026, 0.045)
0.015 (-0.022, 0.052)
0.010 (-0.008, 0.027)
Women not in menopause
0.026 (-0.017, 0.069)
0.028 (-0.017, 0.073)
-0.014 (-0.043, 0.015)
0.003 (-0.013, 0.019)

Odds ratio for osteoporosis in women
PFOA
PFOS
0.42 (0.13, 1.32)
2nd 1.25 (0.38, 4.06)
0.83 (0.45, 1.51)
3rd 1.23 (0.37, 4.05)
1.07 (0.36, 3.19)
4th 2.59 (1.01, 6.67)
Per Ln: 1.84 (1.17, 2.90)
1.14 (0.68, 1.94)

Reference,
Location

Study
population

PFOS
serum
level

Looker 2014
C8

403 aged >18 years

8.32

PFHxS
serum
level

PFOA
serum
level

Outcome

PFHxS
9.29 (1.81, 47.6)
8.06 (1.84, 35.3)
13.20 (2.72, 64.2)
1.64 (1.14, 2.38)

Difference detected (95% CI)

33.74
Influenza Type B
Geometric mean
antibody titer (GMT)
Log10-titer rise
Log10-titer ratio:
Postvaccine / prevaccine

141

Women in menopause
-0.040 (-0.165, 0.085)
-0.023 (-0.144, 0.097)
0.058 (-0.192, 0.075)
-0.019 (-0.047, 0.009)
Women in menopause
-0.017 (-0.103, 0.069)
0.035 (-0.067, 0.137)
0.001 (-0.089, 0.091)
-0.001 (-0.021, 0.020)

PFOA quartiles
regression coefficient, log10-PFOA
GMT
log10-titer rise
log10-titer ratio
1st 49.5 (38.1, 64.1)
.05 (-.09, .19)
2nd 46.0 (35.3, 60.0) -.03 (-.19, .13)
.07 (-.07, .22)
3rd 43.6 (33.1, 57.3) -.02 (-.19, .15)
-.03 (-.17, .12)
4th 20.9 (16.6, 28.2) -.07 (-.24, .10)
Regression coefficient per unit increase in log10-PFOA:
-.02 (-.13, .09)
-.02 (-.11, .08)
PFOA Odds ratio: Seroconversion
Seroprotection
1.43 (0.76, 2.70)
0.76 (0.40, 1.45)
2nd
1.39 (0.73, 2.66)
1.13 (0.57, 2.23)
3rd
0.71 (0.38, 1.36)
0.77 (0.39, 1.50)
4th
Odds ratio per unit-increase in log10-PFOA:
0.80 (0.53, 1.21)
1.04 (0.68, 1.60)
PFOS quartiles
regression coefficient, log10-PFOS
GMT
log10-titer rise
log10-titer ratio
1st 42.3 (33.4, 53.4)
.02 (-.13, .18)
.004 (-.14, .14)
2nd 41.5 (30.7, 56.0)
-.02 (-.16, .12)
3rd 41.1 (31.7, 53.4) -.03 (-.19, .14)
.04 (-.14, .21)
.03 (-.12, .18)
4th 52.8 (38.9, 71.7)
Regression coefficient per unit increase in log10-PFOS:
.05 (-.11, .21)
.05 (-.09, .18)

Influenza Type A H1N1
Geometric mean
antibody titer (GMT)
Log10-titer rise
Log10-titer ratio:
Postvaccine / prevaccine

Influenza Type A H3N2
Geometric mean
antibody titer (GMT)
Log10-titer rise
Log10-titer ratio:

142

PFOS Odds ratio: Seroconversion
Seroprotection
2nd
0.72 (0.39, 1.33)
0.67 (0.35, 1.25)
0.81 (0.42, 1.53)
0.82 (0.42, 1.59)
3rd
0.87 (0.43, 1.74)
0.73 (0.36, 1.47)
4th
Odds ratio per unit-increase in log10-PFOS:
1.17 (0.63, 2.17)
0.85 (0.44, 1.64)
PFOA quartiles
regression coefficient, log10-PFOA
GMT
log10-titer rise
log10-titer ratio
1st 476.2 (360.8, 628.7)
-.08 (-.29, .12)
2nd 352.2 (255.3, 485.9) -.09 (-.27, .08)
-.04 (-.25, .18)
3rd 306.3 (232.6, 403.2) -.10 (-.28, .09)
.07 (-.14, .29)
4th 274.8 (202.9, 372.2) -.12 (-.30, .06)
Regression coefficient per unit increase in log10-PFOA:
-.03 (-.14, .09)
.07 (-.06, .21)
PFOA Odds ratio: Seroconversion
Seroprotection
0.74 (0.34, 1.59)
0.74 (0.17, 3.28)
2nd
1.11 (0.49, 2.50)
1.59 (0.33, 7.70)
3rd
2.23 (0.90, 5.53)
6.47 (0.91, 45.9)
4th
Odds ratio per unit-increase in log10-PFOA:
1.51 (0.89, 2.56)
2.34 (0.91, 6.07)
PFOS quartiles
regression coefficient, log10-PFOS
GMT
log10-titer rise
log10-titer ratio
1st 342.3 (256.0, 457.7)
-.07 (-.28, .13)
2nd 280.4 (197.6, 397.9) -.04 (-.21, .14)
.13 (-.04, .31)
.03 (-.18, .24)
3rd 417.7 (319.0, 547.1)
.10 (-.09, .29)
.03 (-.19, .26)
4th 341.8 (258.0, 452.8)
Regression coefficient per unit increase in log10-PFOS:
.15 (-.02, .32)
.10 (-.11, .30)
PFOS Odds ratio: Seroconversion
Seroprotection
0.97 (0.44, 2.14)
0.55 (0.13, 2.37)
2nd
0.78 (0.35, 1.75)
1.81 (0.32, 10.2)
3rd
0.94 (0.38, 2.31)
1.26 (0.24, 6.61)
4th
Odds ratio per unit-increase in log10-PFOS:
1.10 (0.51, 2.37)
0.93 (0.23, 3.71)
PFOA quartiles
regression coefficient, log10-PFOA
GMT
log10-titer rise
log10-titer ratio
1st 228.9 (161.5, 324.3)
-.10 (-.30, .10)
2nd 125.4 ( 86.0, 182.7) -.28 (-.51, -.06)
-.37 (-.60, -.13)
-.07 (-.28, .14)
3rd 104.1 ( 72.5, 149.6)
-.12 (-.36, .13)
-.22 (-.43, -.01)
4th 183.7 (127.3, 265.2)
Regression coefficient per unit increase in log10-PFOA:
log10-titer rise
log10-titer ratio

Postvaccine / prevaccine

-.01 (-.17, .14)
-.12 (-.25, .02)
PFOA Odds ratio: Seroconversion
Seroprotection
0.90 (0.48, 1.68)
0.34 (0.14, 0.83)
2nd
1.13 (0.59, 2.17)
0.28 (0.11, 0.70)
3rd
0.62 (0.33, 1.16)
0.39 (0.15, 0.99)
4th
Odds ratio per unit-increase in log10-PFOA:
0.76 (0.51, 1.15)
0.66 (0.39, 1.12)
PFOS quartiles
regression coefficient, log10-PFOS
GMT
log10-titer rise
log10-titer ratio
1st 137.7 ( 98.7, 192.1)
.03 (-.19, .26)
-.06 (-.26, .14)
2nd 147.3 ( 99.3, 218.5)
.18 (-.06, .41)
.02 (-.18, .23)
3rd 211.0 (141.2, 315.3)
-.04 (-.28, .21)
-.03 (-.24, .19)
4th 126.7 ( 88.9, 180.7)
Regression coefficient per unit increase in log10-PFOS:
.09 (-.13, .32)
-.005 (-.20, .19)
PFOS Odds ratio: Seroconversion
Seroprotection
1.08 (0.59, 1.97)
0.85 (0.38, 1.88)
2nd
1.10 (0.59, 2.06)
1.09 (0.47, 2.56)
3rd
1.41 (0.72, 2.78)
0.56 (0.24, 1.28)
4th
Odds ratio per unit-increase in log10-PFOS:
1.17 (0.63, 2.15)
0.63 (0.26, 1.49)

755 aged >18 years

Self-reported cold or flu
in last 12 months

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Kielsen 2016
Denmark

12 adults

9.52

0.37

1.69

PFOA
odds ratio
PFOS
1.66 (1.00, 2.75)
2nd 1.21 (0.73, 2.00)
1.19 (0.74, 1.94)
3rd 1.10 (0.67, 1.81)
1.15 (0.69, 1.91)
4th 0.84 (0.52, 1.36)
OR per unit increase in log10-PFAS:
0.85 (0.62, 1.16)
0.90 (0.55, 1.48)

Outcome

Difference detected (95% CI)

Diphtheria
Tetanus

% change in antibody per doubling of PFAS
PFOA
PFOS
PFHxS
-8.22 (6.44, -20.85) -11.90 (-0.33, -21.92) -13.31 (0.29, -25.07)
0.23 (12.1, -10.40) --3.59 (5.51, -11.91)
-4.35 (6.04, -13.72)

143

Reference,
Location

Study
population

PFOS
serum
level

PFHxS
serum
level

PFOA
serum
level

Outcome
Difference detected
(95% CI)

Reference,
Location

Stein 2016b
New York

75 aged 21-49 years

5.22

1.1

2.28

FluMist Seroconversion
measured by immunohistochemistry

RR

Serum Immune Markers:
Interferon-α2

PFOS
2nd 2.6 (0.9, 7.4)
3rd 2.4 (0.9, 6.6)

PFOA
0.6 (0.2, 2.0)
1.8 (0.7, 4.3)

PFHxS
1.1 (0.4, 2.9)
1.7 (0.6, 4.8)

Mean change between baseline and FluMist response:
-21 (-51, 8.61)
2nd 8.69 (-23, 40.5) 2.54 (-28, 32.9)
-29 (-64, 5.20)
3rd -5.6 (-37, 25.7) 10.9 (-19, 40.4)

Interferon-γ
2nd 8.00 (-34, 49.6) -23 (-61, 14.1)
3rd 9.95 (-28, 47.9) -23 (-61, 16.1)

-40 (-76, -3.7)
-40 (-84, 2.69)

Tumor necrosis factor-α
2nd -0.06 (-4.6, 4.45) 1.28 (-2.8, 5.39)
3rd 0.59 (-3.6, 4.81) -0.96 (-5.2, 3.26)
Interferon-γ-inducible
protein 10 (IP-10)
Monocyte chemoattractant protein-1
Macrophage inflammatory protein-1a
Granulocyte colonystimulating factor
Nasal Secretion Immune
Markers:
IP-10
Monocyte chemoattractant protein-1
Mucosal immuneglobulin A

144

2nd 12.0 (-44, 67.6) -3.4 (-55, 48.7)
3rd -42 (-94, 10.4) -28 (-82, 25.0)

-5.3 (-9.2, -1.3)
-4.8 (-9.4, -0.1)
-32 (-84, 20.1)
-15 (-76, 46.9)

2nd 4.75 (-60, 69.7) -9.9 (-70, 49.8)
3rd 19.0 (-42, 79.8) -2.7 (-64, 58.6)

-39 (-97, 19.5)
20.6 (-48, 89.5)

2nd -1.6 (-22, 18.8) 4.25 (-15, 23.7)
3rd -0.51 (-20, 19.2) 5.09 (-15, 24.8)

-6.6 (-25, 12.1)
-5.0 (-28, 18.4)

2nd -8.7 (-60, 42.8) -1.4 (-49, 45.7)
3rd -0.36 (-49, 47.9) -13 (-62, 35.2)

-10 (-56, 35.9)
36.6 (-18, 91.3)

2nd 429 (-1309, 2166) -30 (-1623, 1563)
3rd -215 (-1841, 1412) -564 (-2200, 1072)

-691 (-2279, 896)
-713 (-2596, 1170)

2nd 2.34 (-9.2, 13.8) 0.72 (-9.9, 11.3)
3rd -6.7 (-17, 4.04) -6.6 (-18, 4.24)

-0.7 (-11, 10.6)
-3.6 (-16, 9.08)

2nd 73.9 (-216, 364) -17 (-285, 251)
3rd -88 (-359, 183)
28.7 (-246, 304)

294 (35.2, 553)
238 (-69, 545)

Table A4. PFAS studies on infertility/subfertility.
Reference
Fei 2009
Danish study

Exposure
PFOA and PFOS in blood samples at
gestational week 17

Outcome
Subfecundity
(n= 1240 women)

OR/FR/β & 95% CI
Infertility OR = 1.77 (1.06, 2.95) and 2.54 (1.47, 4.39) for
highest vs lowest quartile of PFOS and PFOA, respectively
Fecundity OR (FOR) = 0.74 (0.58, 0.93) and 0.60 (0.47, 0.76)
for highest vs lowest quartile of PFOS and PFOA, respectively

Whitworth 2012a
Norway study

PFOA and PFOS in blood samples at
gestational week 17

Subfecundity
(n= 910)

Buck Louis 2013
LIFE study

PFOS, PFOA, PFNA and 4 other PFCs in
serum

Velez 2015
MIREC study

PFOS
PFOA
PFHxS
Measured in the 1st trimester

Fecundability ORs (TTP)
(n=501 couples followed for 12
months)
Female fecundity odds ratio (FOR)
as measured by TTP
Infertility

Jorgensen 2014

PFOA, PFOS, PFHxS and PFNA

(n= 1743 women recruited before
14 weeks of gestation)
Fecundability ratios (FRs)
infertility
(n= 938 women; 448 were from
Greenland, 203 from Poland, and
287 from Ukraine)

FORs <1 indicate decreased fecundity and a longer TTP
ORs = 0.7 0.4–1.3) for highest PFOS quartile and 0.5 (0.2–1.2)
for highest PFOA quartile among primiparous women and there
was a monotonic exposure-response relationship for PFOS
FORs about 1 for PFOS, PFOA, and PFNA

FOR = 0.89 (0.83–0.94), 0.91 (95% CI 0.86–0.97), and 0.96
(0.91–1.02) per one SD increase in log-transformed serum
concentrations of PFOA, PFHxS, and PFOS, respectively
ORs for infertility = 1.31 (1.11–1.53) for PFOA
1.14 (0.98–1.34) for PFOS, and 1.27 (1.09–1.48) for PFHxS
log-scale FR = 0.80 (0.69-0.94) and 0.90 (0.76, 1.07) for PFNA
and PFOS, respectively, in pooled sample
FRs in Greenland ranged from 0.71-0.90 for categorical
analyses and there were monotonic exposure-response
relationships for PFOA, PFOS, and PFNA
FRs in Poland were 0.90 and 0.94 for categorical analyses of
PFOS and PFHxS, respectively
FRs in Ukraine were 0.93 and 0.88 for categorical analyses of
PFOS and PFNA, respectively
log-scale OR = 1.53 (1.08-2.15), 1.11 (0.74, 1.66), and 1.39
(0.93, 2.07) for PFNA, PFOA, and PFOS, respectively, and
infertility in pooled sample
ORs in Greenland ranged from 1.22-2.15 for categorical
analyses and there was a monotonic exposure-response
relationships for PFOA

145

ORs in Poland were 1.41 and 1.92 for categorical analyses of
PFOA and PFOS, respectively, and there were monotonic
exposure-response relationships for PFOA
ORs in Ukraine ranged from 1.17-1.22 for categorical analyses
of PFOA, PFOS, and PFNA

Reference
Bach 2015a
Danish study

Exposure
plasma PFOS and PFOA

Outcome
Time to pregnancy (TTP)
(n=440 for sample 1 and n= 1161
for sample 2)

associations were weaker in a sensitivity analysis of primiparous
women
OR/FR/β & 95% CI
PFOA:
FRs ranging from 0.74-0.86 comparing the highest and lowest
quartiles
FRs was 0.74 and 0.82 comparing the highest and lowest
quartiles in nulliparous women in sample 2 and the pooled
analysis, respectively, and the log FR were 0.67 and 0.84 and
there was an exposure-response relationship for sample 2
FRs ranged from 0.74-0.76 comparing the highest and lowest
quartiles in parous women and the log FRs ranged from 0.660.72 and there was an exposure-response relationship in the
pooled sample
PFOS:
FRs ranged from 0.69-0.97 comparing the highest and lowest
quartiles in nulliparous women and the log FR was 0.62 in
sample 2 and 0.78 in the pooled analysis

Vested 2013
Danish study

In utero exposure to PFOA and PFOS
measured in maternal blood samples from
pregnancy week 30

Adult male semen quality,
testicular volume, and reproductive
hormone levels (n=169)

Maternal and adult son questionnaires on
dietary/health and lifestyle habits
Mothers’ median plasma concentrations of
PFOA and PFOS were 3.8 ng/mL (2.8–4.7

FRs was 0.93 and 0.97 comparing the highest and lowest
quartiles in parous women in sample 2 and the pooled analysis,
respectively, and the log FR ranged from 0.85-0.91
PFOA was associated with a lower percentage of adjusted sperm
concentration (–34%, –58, 5), total sperm count (–34%, –62,
12), and morphologically normal spermatozoa (–19%, –42, 13)
and with higher adjusted levels of luteinizing hormone (6 IU/L,
–11, 27) and follicle-stimulating hormone (15 IU/L, –8, 44).
Monotonic exposure-response relationships for sperm
concentration and LH and FSH levels
PFOS was associated with a lower percentage of adjusted total
sperm count (–23%, –56, 38) and morphologically normal

146

ng/mL) and 21.2 ng/mL (17.4–26.5
ng/mL), respectively

Bach 2016a systematic review

PFOS and PFOA measured in blood

Fertility measured by:

Men: Average exposure
levels in the non-occupational studies
ranged from 4.6-44.7 ng/mL for PFOS and
1.3-9.2 ng/mL for
PFOA

Reproductive hormones and TTP
in men and women
Semen characteristics

Women: Average exposure levels ranged
from 3.8-36.3 ng/mL for PFOS and 1.5-5.6
ng/mL for PFOA

Women: TTP n’s varied from 2221743; hormone n’s varied from
178-825

Men: n’s varied from 56-857

147

spermatozoa (–14%, –39, 20) and with higher adjusted levels of
follicle-stimulating hormone (20 IU/L, –5, 51). Monotonic
exposure-response relationships for morphologically normal
spermatozoa and FSH levels
Men: Inconsistent results for semen volume, sperm
concentration, total sperm count, motility, and morphology;
levels of testosterone, free androgen index/free testosterone,
estradiol, SHBG, LH, FSH, and inhibin B; and TTP
Women: Inconsistent results except for mostly positive
associations for infertility and fecundability in parous women

Table A5. PFAS studies on pregnancy-induced hypertension/pre-eclampsia.
Reference
Stein 2009
C8 project

Savitz 2012a
C8 project

Starling 2014
Norwegian Mother and Child
Cohort Study
Avanasi 2015
C8 project

Exposure
PFOA
PFOS

Serum PFOA levels at the time of
pregnancy from drinking water
contaminated by chemical plant releases
analysis uses modeled serum PFOA
estimates
PFOA, PFOS, and PFHxS in
maternal plasma extracted midpregnancy

Outcome
Pre-eclampsia
(n=1,845 pregnancies for PFOA
and 5,262 pregnancies for PFOS)
Preeclampsia

OR/HR & 95% CI
For exposures >90th percentile:
Pre-eclampsia: OR = 1.6 (1.2, 2.3) for PFOS; OR <1 for PFOA

OR = 1.2 (1.0–1.6) for highest vs lowest quartile

(n= 11,737 pregnancies)

Preeclampsia

PFOS: HR = 1.09 (0.75, 1.58) for highest vs lowest quartile and
HR = 1.13 (0.84, 1.52) for per ln-unit

(n= 466 cases, 510 noncases)
Estimated and simulated PFOA
concentrations

Savitz 2012b
C8 project

Historical estimates of serum PFOA from
a fate and transport model using address at
delivery (birth records) and a survey with
residential history data

Darrow 2013
C8 project

PFOA and PFOS measurements

Preeclampsia

PFHxS: HRs <1
OR in 12 simulations ranged between 1.10 and 1.12

(n= 10,149 participants for each
of the 12 Monte Carlo simulations
[500 iterations per simulation])
Pregnancy-induced hypertension
(PIH) (n=224 cases and n=3616
controls)

OR = 1.11 (0.99, 1.24) from original exposure assignments

Pregnancy-induced hypertension
(n=106 and n=1630 total births)

ORs = 3.16 (1.35, 7.38) and 1.56 (0.72, 3.38) for the highest vs
lowest quintile of PFOA and PFOS, respectively

148

OR = 1.2 (0.8, 1.7) using survey data and comparing the highest
vs lowest quintile of PFOA (Bayesian calibration)

Table A6. PFAS studies on adverse birth outcomes.
Reference
Apelberg 2007
Baltimore THREE study

Exposure
PFOA and PFOS in cord serum
samples

Outcome
Gestational age, birth weight,
and birth size
(n = 293)

OR/β & 95% CI

birth weight per ln-unit: β = –69 g ( –149 to 10) for PFOS
and –104 g ( –213 to 5) for PFOA
ponderal index per ln-unit: β = –0.074 g/cm3 × 100 ( –
0.123 to –0.025) for PFOS and = –0.070 g/cm3 × 100 ( –
0.138 to –0.001) for PFOA
head circumference per ln-unit: β = –0.32 cm ( –0.56 to –
0.07) for PFOS and –0.41 cm ( –0.76 to –0.07) for PFOA

Fei 2007
Danish study

PFOS and PFOA

Preterm birth, low birth
weight, SGA
(n=1400)

Stein 2009
C8 project

PFOA
PFOS

Preterm birth
Low birth weight
(n=1,845 pregnancies for
PFOA and 5,262 pregnancies
for PFOS)

Fei 2008
Danish study

PFOS and PFOA in maternal blood
samples taken early in pregnancy

Placental weight
Birth length
Head and abdominal
circumferences
Ponderal index
(n=1400)

149

length per ln-unit: β = -0.10 cm ( -0.64 to 0.44) for PFOA
β = –10.63 g ( –20.79, –0.47) for PFOA and birth weight
OR = 4.82 (0.56–41.16) and 2.44 (0.27–22.25) for LBW
and highest vs lowest quartiles of PFOS and PFOA,
respectively
OR = 1.43 (0.50–4.11) and 1.71 (0.55–5.28) for preterm
birth and highest vs lowest quartiles of PFOS and PFOA,
respectively
For exposures >90th percentile:
Pre-term birth: OR = 1.4 (1.1, 1.7 ) for PFOS and there
was a monotonic exposure-response relationship and OR
<1 for PFOA
Low birth weight OR = 1.8 (1.2, 2.8) for PFOS and there
was a monotonic exposure-response relationship and OR
<1 for PFOA
Placental weight β = -21.3 g (-46.1, 3.4) and -10.8 g
(-33.4, 11.8) for highest vs lowest quartiles of PFOA and
PFOS, respectively; monotonic exposure-response
relationship for PFOA
Birth length β = -0.49 cm (-0.81, -0.16) for highest vs
lowest quartiles of PFOA

Head circumference β = -0.14 cm (-0.39, 0.12) for highest
vs lowest quartiles of PFOA
Abdominal circumference β = -0.29 cm (-0.63, 0.06) for
highest vs lowest quartiles of PFOA; monotonic exposureresponse relationship for PFOA
Reference
Nolan 2009
Little Hocking
communities

Exposure
PFOA

Washino 2009
Japan

PFOS and PFOA in maternal serum

Andersen 2010
Danish study

maternal plasma levels of PFOS and
PFOA and cord blood samples

Hamm 2010
Canada

PFOS
PFOA
PFHxS

Outcome
Mean birthweight, mean
gestational age, low
birthweight, and preterm birth
(n=1555)
Birth weight
Birth length
Chest circumference
Head circumference
(n=428)
Weight, length, and body mass
index development during 1st
year of life
(n=1400 born in 1996-2002)
Birth weight
Fetal growth
SGA
Preterm birth
(n=252)

OR/β & 95% CI

β = −8.81 g (−86.1–68.5) for mean birth weight
comparing LHWA only to no LHWA water service

β = –269.4 g (–465.7 to –73.0 g) and –76.7 (–234.7 to
81.3) for PFOS and PFOA per log10 unit and birth weight
only in female infants

Weight: aβ = -0.8 g (-4.2, 2.6) at 5 months and -5.8 g
(-10.4,-1.2) at 12 months for PFOS; -9.4 g (-28.6, 9.9) at 5
months and -19.0 g (-44.9, 6.8)
at 12 months for PFOA
adjusted changes in birth weight per natural log (ng/ml) of
PFOA were -37.4 g (-86.0 to 11.2 g)
Difference of -0.086 (- 0.62 to 0.45) gestation week for
highest vs lowest tertile of PFOA
RR= 2.35 (0.63–8.72) for highest vs lowest tertile of
PFHxS and SGA
RR = 1.31 (0.38-4.45) and 1.11 (0.36–3.38) for preterm
birth comparing highest vs lowest tertile of PFOA and
PFOS, respectively; monotonic exposure-response
relationship for PFOS

150

Reference
Chen 2012
Taiwan

Exposure
Cord blood for PFOA, PFOS, PFNA,
and PFUA

Outcome
Gestational age
Birth weight
Birth length
Head circumference Ponderal
index (indicator of
disproportionate or
asymmetric growth restriction)
Preterm birth
Low birth weight
SGA
(n=429)

OR/β & 95% CI

per ln unit:
β = -0.37 (-0.60, -0.13) weeks for gestational age and 0.17 cm
(-0.42, 0.09) for birth length and PFOS
βs ranged from -19.2 to -110.2 g for birth weight and
-0.05 to -0.25 cm for head circumference and PFOA,
PFOS, and PFUA; there was a monotonic exposureresponse relationship for head circumference and PFOS
when exposures were categorized into quartiles
βs ranged from -0.01 to -0.02 for Ponderal index and
PFOA, PFOS, PFNA, and PFUA
ORs of preterm birth and low birth weight = 2.45 (1.47,
4.08) and 2.61 (0.85, 8.03), respectively, for PFOS
ORs = 2.27 (1.25, 4.15) and 1.24 (0.75, 2.05) for SGA and
PFOS and PFOA, respectively

Savitz 2012a
C8 project

Maisonet 2012
Avon Longitudinal Study
(Britain)

Serum PFOA levels at the time of
pregnancy from drinking water
contaminated by chemical plant
releases
analysis uses modeled serum PFOA
estimates
PFOS
PFOA
PFHxS

Preterm birth, term low
birthweight

An exposure response relationship was observed for PFOS
and head circumference and preterm birth
OR ≤ 1 for preterm birth and term low birthweight

(n= 11,737 pregnancies)

Fetal and postnatal growth in
girls
(n=447)

Birth weight: β ranged from -107.93 to -14.01 g for the
highest vs lowest tertiles of PFOS, PFOA, and PFHxS,
respectively; exposure-response relationships were
observed for all 3 chemicals
Birth length: β ranged from -0.44 to -0.82 cm for the
highest vs lowest tertiles of PFOS, PFOA, and PFHxS;

151

exposure-response relationships were observed for PFOA
and PFHxS
Gestational age: β ranged from -0.15 to -0.24 weeks for
the highest vs lowest tertiles of PFOS, PFOA, and PFHxS,
respectively; exposure-response relationships were
observed for all 3 chemicals
At 20 months, girls usually weighed more for the highest
vs lowest tertiles of the chemicals
Reference
Whitworth 2012b
Norway study

Bach 2015b meta-analysis

Exposure
Maternal plasma samples of PFOS and
PFOA obtained around 17 weeks of
gestation

PFOA or PFOS in maternal blood
during pregnancy or umbilical cord

Wu 2012
China

PFOA

Darrow 2013
C8 project

PFOA and PFOS measurements in
2005-2006

Outcome
Birth weight z scores, preterm
birth, SGA
and large for gestational age
(LGA)
(n=901)
Birth weight

Gestational age
Birth weight
Birth length
Apgar scores
(n=167)
Preterm birth (n=158), low
birth weight (n=88), and birth
weight among full-term infants

OR/β & 95% CI

β = -0.18 (-0.41, 0.05) and -0.21 (-0.45, 0.04) for adjusted
birth weight z scores and highest vs lowest quartiles of
PFOS and PFOA, respectively
OR = 1.3 (0.5, 3.4 0.51) for SGA and highest vs lowest
quartiles of PFOS
PFOA exposure was associated with decreased measures
of continuous birth weight in all 14 studies
PFOS exposure and birth weight were associated in some
studies
Adjusted results for 1 lg-unit change in PFOA:
Gestational age: −15.99 days (−27.72 to −4.25)
Birth weight: −267.30 g (−573.27 to −37.18)
Birth length: −1.91 cm (−3.31 to −0.52)
5-minute Apgar score: −1.37 (−2.42 to −0.32)
β = -54 g (–124, 17) for birth weight in full-term infants
and the highest vs lowest quintile of PFOS
(β = -105 g [–196, –13] when restricted to births
conceived after the blood sample collection

(n=1630)
OR = 1.32 (0.53, 3.32) for preterm birth and highest vs
lowest quintile of PFOA restricted to births conceived
after the blood sample collection
OR = 1.33 (0.60, 2.96) for LBW and the highest vs lowest
quintile for PFOS

152

Reference
Kishi 2015
Hokkaido Study
Bach 2015b
Danish study

Exposure
prenatal PFOS and PFOA levels were
measured in maternal serum samples

Outcome
Birthweight

(n= 306 mother-child pairs)
serum levels of PFHxS, PFHpS, PFOS, Birth weight
PFOA,
Birth length
Head circumference
PFNA, PFDA, and PFUnA
Gestational age at birth
measured between 9-20 completed
Preterm birth
gestational weeks
(n=1507 mother-child pairs)

OR/β & 95% CI

β = - 186.6 g (–363.4, –9.8) for females comparing the 4th
and 1st quartiles of PFOS
βs for birthweight ranged from -8 to -23 g for the highest
vs lowest quartiles of PFHxS, PFHpS, PFOS, and PFUnA
βs for birthweight for girls ranged from -39 to -76 g for
the highest vs lowest quartiles of PFHxS, PFHpS, PFOS,
PFNA, PFDA, and PFUnA
Association between PFAA exposures and birth length,
head circumference, and gestational age were all close to
zero

Lenters 2015
Greenland, Poland and
Ukraine
Verner 2015
Meta-analysis

PFHpA, PFHxS, PFOS, PFOA, PFNA,
PFDA, PFUnDA, and PFDoDA

Term birth weight (n=1250)

PFOA and PFOS

Birth weight

Lee 2016
South Korea

PFBS
PFHxS
PFHpS
PFOS
PFOA
PFNA
PFDA
PFUnA
PFDoA

Birthweight in (n=85 births)

153

OR = 1.18 (0.65, 2.16) for preterm birth for the highest vs
lowest quartile of PFNA
ln‒PFOA β = -63.77 g (-122.83, -4.71); represents change
per a 2-SD increase in ln‒transformed exposure biomarker
or untransformed continuous covariate levels
summary β coefficients for g birth weight per ng/ml
increase in PFOA and PFOS levels were -14.7 g (-21.7, 7.8) and -5.0 g (-8.9, -1.1), respectively
lnPFOS β = −0.14 (95% CI: −0.33, 0.03)
lnPFOA β = −0.03 (95% CI: −0.25, 0.18)
lnPFNA β = −0.14 (95% CI: −0.39, 0.10)
lnPFDA β = −0.12 (95% CI: −0.39, 0.14)
lnPFDoA β = −0.03 (95% CI: −0.36, 0.30)

Reference
Callan 2016
Western Australia

Lauritzen 2017

Exposure
PFOS
PFOA
PFHxS
And 11 other PFAS measured in whole
blood

Outcome
Birth weight
Birth length
Head circumference

OR/β & 95% CI

(n=98 pregnant women)

Median (in μg/L):
PFOS 1.99
PFHxS 0.33
PFOA 0.86

Proportion of optimal birth
weight (POBW), proportion of
optimal birth length (POBL)
and proportion of optimal head
circumference
(POHC)

β= - 69 g (−231, 94), -48 g (−203, 108) and -103 g (−221,
15) for birthweight and ln-unit increase in PFOS, PFOA,
and PFHxS

PFOS and PFOA
Median serum levels (ng/ml):
PFOA: 2.33 in Sweden and 1.62 In
Norway
PFOS: 16.4 in Sweden and 9.74 in
Norway

(n=82-89 infants)
Birth weight, birth length,
head circumference,
gestational age, SGA
424 mother-child pairs,
excluding 1st time mothers:
143 SGA births and 281 nonSGA controls

OR = 3.5 (1.1–11.5) for being 95% of their calculated
optimal birth weight comparing the highest to lowest
tertile of PFHxS

Sweden:
PFOA:
Birthweight β= -359 g (-596, -122)
Birth length β= -1.3 (-2.3, -0.3)
Head circumference β= -0.4 (-1.0, 0.1)
Gestational age β= -0.3 (-0.9, 0.3)
SGA OR=5.25 (1.68-16.4)
Differences more pronounced in boys
PFOS:
Birthweight β= -292 (-500, -84)
Birth length β= -1.2 (-2.1, -0.3)
Head circumference β= -0.4 (-0.9, 0.04)
Gestational age β= -0.4 (-0.9, 0.2) 0.201
SGA OR=2.51 (0.93-6.77)
In Norway, SGA ORs < 1 and βs > 1 or very close to 0

154

Table A7. PFAS studies on congenital malformations.
Reference
Nolan 2010
Little Hocking communities

Savitz 2012a
C8 project

Stein 2014a
C8 project

Vesterholm Jensen 2014

Toft 2016
Danish study

Kim 2016
South Korea

Exposure
PFOA

Serum PFOA levels at the time of
pregnancy from drinking water
contaminated by chemical plant
releases
analysis uses modeled serum PFOA
estimates
Modeled PFOA

Cord blood PFAS levels

Amniotic fluid PFOS level

16 PFAS in infant sera
Mean concentrations were: PFOS
4.05ng/mL), PFOA
(2.12ng/mL), PFHxS 1.17ng/mL)

Outcome
Congenital anomaly
(n=168 served by LHWA only
and 1171 no LHWA)
Birth defects

OR/β/RR & 95% CI
Serviced entirely by contaminated LHWA water:
OR = 7.0 (0.4-113) for both heart and circulatory defect
OR = 21 (0.9-517) for club foot
OR ≤ 1 for birth defects

(n= 11,737 pregnancies)

Maternally reported birth
defects (n = 325) among 10,262
births

Cryptorchidism
(n=29 Danish cases and 30
matched controls and 78
Finnish cases and 78 matched
controls)
Cryptorchidism
Hypospadias
(n= 270 cryptorchidism cases,
75 hypospadias cases, and 300
controls)
Congenital hypothyroidism
(CH) measured by serum
thyroid stimulating hormone
(TSH),free

155

Brain defect OR = 2.6 (1.3-5.1) for IQR increase from
25th to 75th percentile and OR = 16.1 (0.8, 325) for
highest vs lowest tertile
Limb defect ORs = 1.2 (0.7, 2.0) for IQR increase and
1.5 (0.2, 9.7) for highest vs lowest tertile
Eye defect OR = 1.3 (0.2-8.4) for highest vs lowest tertile
Heart defect ORs = 1.2 (0.8, 1.7) for IQR increase and
1.4 (0.4, 5.1) for highest vs lowest tertile
OR = 1.14 (0.19–6.95) and 1.30 (0.27–6.39) for ln PFOA
and PFOS in Danish cases, respectively
OR = 2.34 (0.16–34.67) for the highest vs lowest tertile
of PFOS in Denmark
ORs for cryptorchidism and hypospadias were < 1

large difference in PFOA concentrations between cases
and controls (2.12 ng/mL in controls and 5.40 ng/mL in
cases)

thyroxine (FT4), total T3,
thyroglobulin antibody (TGAb),
relevant microsomal antibodies
(microAb), and thyroid
stimulating immuno-globulin
(TSI)

Reference
Liew 2014
Danish study

Exposure
PFASs in maternal plasma collected
in early or midpregnancy: PFOS,
PFOA, PFHxS, PFNA, PFHpS,
PFDA

(n=27 infants with CH and 13
healthy infants)
Outcome
Cerebral palsy
(n=156 cases and 550 controls)

Median (ng/ml)
PFOS: 27.40
PFOA: 4.00
PFHxS: 0.92

mean concentrations of PFOA, PFNA, PFDA, PFUnDA,
and total PFAS in cases (0.525–16.8ng/mL) were
“significantly” higher than in controls (0.298–
10.0ng/mL) (data presented in figure only)
in CH infants, correlations were -0.482 and -0.642 for
TSI and PFOA and PFHxS, respectively (results not
shown for PFOS)
OR/β/RR & 95% CI
per 1-unit (natural-log ng/mL) increase in boys:
RR = 1.7 (1.0, 2.8) for PFOS
RR = 2.1 (1.2, 3.6) for PFOA
RR = 1.2 (0.9, 1.7) for PFHxS
RR = 1.2 (0.6, 2.5) for PFNA
RR = 1.5 (1.0, 2.2) for PFHpS
RR = 1.1 (0.7, 1.7) for PFDA
and there was an exposure response relationship for
PFHxS, PFNA, and PFHpS when exposure was
categorized
RRs generally increased for boys born at term

156

Table A8. PFAS studies on adverse health outcomes in children ages ≥2 years.
Reference,
Location

Study population

Pease Tradeport

children <12 years, N=366
children <18 years, N=396
1,971 boys <12 years
2,773 boys 12 - <18

PFOS
serum
level
8.3
8.1
19.9
20.3

1,886 girls <12 years
2,520 girls 12 - <18

21.7
18.2

n/a

30.7
22.9

Zeng 2015
Taiwan

102 boys, age 12-15
123 girls, age 12-15
225 total age 12-15

29.9
28.8
28.9

1.4
1.2
1.3

0.5
0.5
0.5

Total cholesterol (all
children)

Maisonet 2015a
Avon, UK

Maternal serum
(N=199 girls aged 7 and 15)

20.0

3.6

Total cholesterol

Geiger 2014a
NHANES

815 children, aged 12-18
1999-2008

17.7
(mean)

4.2
(mean)

Total cholesterol

Frisbee 2010
C8 study

PFHxS
serum
level
4.2
4.0
n/a

PFOA
serum
level
3.6
3.4
35.1
30.1

Outcome

Total cholesterol
(mg/dL)
Total cholesterol
(mg/dL)
≥170 mg/dL

High cholesterol
Nelson 2010
NHANES

322 boys, 12-19 (2003-2004)
263 girls, 12-19 (2003-2004)

19.9

2.4

4.0
Total cholesterol

Non-HDL cholesterol

Lopez-Espinosa 2012
C8 study

1,078 children, ages 1–5 years
3,132 children ages 6–10 years
6,447 ages >10–17 years

16.3
21.8
19.6

33.8
32.2
26.9

thyroid stimulating
hormone, TT4
Thyroid disease

157

Difference detected (95% CI)

< 12 years
12 - < 18 yrs
PFOA: +6.3¶ β=1.6 (0.4)
+4.8¶ β=1.1 (0.4)
PFOS: +6.2¶ β=1.2 (0.5)
+9.3¶ β=2.1 (0.4)
¶
PFOA: +5.8 β=1.1 (0.4)
+3.9¶ β=1.0 (0.4)
¶
PFOS: +4.6 β=1.3 (0.5)
+9.4¶ β=1.9 (0.4)
OR=1.6 (1.4, 1.9), PFOS (5th quintile)
OR=1.2 (1.1, 1.4) PFOA (5th quintile)
4th quartile: 23.1 mg/dL increase for PFOS,
12 mg/dL increase for PFOA.
Ln PFHxS β = 1.1 (-0.7, 2.9)
Ln PFNA β =12.9 (0.7, 25.1)
Ln PFOS β = 0.3 (0.2, 0.5)
Ln PFOA β = 6.6 (2.7, 10.4)
3rd tertile β for PFOS & PFOA were <0.
Mean differences at age 15, 3rd tertile vs 1st tertile, for
PFOA and PFOS = 8.1 and 19.1, respectively.
PFOS: 5.9 mg/dL increase (0.1, 11.6), 3rd tertile
PFOA: 7.0 mg/dL increase (1.4, 12.6), 3rd tertile
PFOS: OR=1.53 (1.07, 2.19), 3rd tertile
PFOA: OR=1.49 (1.05, 2.12) 3rd tertile
4th quartile mean difference (vs 1st quartile), (mg/dL)
Boys
Girls
PFOS: 3.6 (-8.5, 15.7)
-0.4 (-9.3, 8.6)
PFOA: 5.0 (-2.3, 12.2)
3.3 (-4.2, 10.8)
PFHxS: -3.2 (-15.4, 9.0) -12.7 (-23.4, -2.0)
PFOS: 2.0 (-8.1, 12.1)
-4.1 (-12.9, 4.7)
PFOA: 6.8 (-1.3, 14.9)
-1.0 (-9.1, 7.2)
PFHxS: -3.3 (-14.6, 8.1)
-16.5 (-29.5, -3.4)
PFOS, 4th vs 1st quartile: 3.1% (0.0, 6.2) increase in
TSH, 2.3% (1.2, 3.3) increase in TT4
PFOA: OR=1.44 (1.02, 2.03)
PFOS: OR=0.80 (0.62, 1.08)

Reference,
Location

Study population

PFOS
serum
level
7.0

PFHxS
serum
level

Lin 2013
Taiwan
Qin 2016
Taiwan

212 aged 12-19

29.9
28.8
28.9
16.6

1.4
1.2
1.3

Geiger 2013
NHANES

102 boys, aged 12-15
123 girls, aged 12-15
225 total, aged 12-15
1,772 aged 12-18
1999-2008 data

Kataria 2015
NHANES

1,960 aged 12-18
2003-2010

12.8

2.0

PFOA
serum
level
2.8

0.5
0.5
0.5
4.1

3.5

Outcome

Difference detected (95% CI)

Free T4

5% increase in free T4 for PFNA(same level as
Pease)¥
PFOA
PFOS
PFHxS
2.8 (1.4, 5.6) 1.4 (0.9, 2.2)
1.65 (1.01, 2.69)
1.6 (0.7, 3.9) 1.5 (0.8, 2.9)
1.3 (0.7, 2.3)
2.2 (1.3, 3.6) 1.35 (0.95, 1.93) 1.4 (0.9, 2.1)
PFOA: .30 mg/dL increase
PFOS: .12 mg/dL increase
PFOA: OR=1.62 (4q vs 1q)
PFOS: OR=1.65 (4q vs 1q)
PFOA: .21 mg/dL increase (0.06, 0.37)
PFOS: .19 mg/dL increase (0.03, 0.34)
PFHxS: .05 mg/dL decrease (-9.22, 0.11)

Uric acid ≥6 mg/dL;
14.7% prevalence
(Odds ratios)
Serum uric acid (4th
quartile)
Hyperuricemia (16%)
Serum uric acid (4th
quartile
eGFR (4th quintile)
(mL/min/1.73 m2)

Lopez-Espinosa 2016
C8

1,169 boys aged 6-9 years

22.4

8.1

34.8

1,123 girls aged 6-9 years

20.9

7.0

30.1

Tsai 2015
Taiwan

95 children aged 12-17

7.12

Maisonet 2015b
Avon, UK

72 girls, 15 years of age∆

19.2

3.03

1.6

3.6

Percent difference
Ln testosterone
Ln estradiol
Ln IGF-1
Ln testosterone
Ln estradiol
Ln IGF-1
Ln SHBG
Ln FSH
Ln testosterone
Total testosterone

SHBG
Zhou 2016
Taiwan
Christensen 2011
Avon, UK

102 boys, aged 12-15

29.9

1.4

0.5

Ln testosterone

123 girls, aged 12-15
218 girls (puberty <11.5 yrs)
230 controls (aged 13 yrs) ∆

28.8
19.8

1.2
1.6

0.5
3.7

Ln estradiol
Early age at puberty

158

PFOA: -6.61 (-11.39, -1.83)
PFOS: -9.47 (-14.68, -4.25)
PFHxS: -0.32 (-4.44, 3.81)
PFOS
PFOA
PFHxS
-5.8 (-9.4, -2.0) -4.9 (-8.7, -0.8) -2.7 (-6.4, 1.2)
-4.0 (-7.7, -0.1) 4.3 (-0.4, 9.1) -1.3 (-5.5, 3.1)
-5.9 (-8.3, -3.3) -0.4 (-3.4, 2.7) -2.5 (-5.2, 0.3)
-6.6 (-10.1, -2.8) -2.5 (-6.7, 1.8) 0.2 (-3.5, 4.0)
-0.3 (-4.6, 4.2)
4.2 (-0.7, 9.4) 2.1 (-2.2, 6.5)
-5.6 (-8.2, -2.9) -3.6 (-6.6, -0.5) -2.1 (-4.8, 0.7)
PFOA: decline among girls
PFOS: declines both sexes
PFOS: decline among girls
Increase in total testosterone by about .20 nmol/L for
PFOS, PFOA, and PFHxS (95% CI: .01, .38)
Declines for PFOA and PFHxS in 3rd tertile but
increases in 2nd tertile
Declines in both sexes for PFOA; decline in boys,
PFOS
Increase in both sexes, PFOA
PFOA: OR=1.29 (0.86, 1.93)
PFOS: OR=0.83 (0.56, 1.23)

Reference,
Location
Lopez-Espinosa 2011
C8

Kristensen 2013
Denmark
Wang 2015
Taiwan

Study population

PFOS
serum
level

3,072 boys, ages 8-18

20

26

2,903 girls, ages 8-18

18

20

343 women, 20 years old ∆

21.1

3.6

120 at age 5∆ €

13.25

0.69

2.5

120 at age 8∆ €

12.28

0.69

2.5

Stein 2013
C8

320 children, 6-12 years

Lien 2016
Taiwan
Stein 2011
C8

282 children, 7 years old
Cord blood levels
10,546 children ages 5-18

PFHxS
serum
level

PFOA
serum
level

35

4.79
20.2

1.55
5.2

28.2

Outcome

Difference detected (95% CI)

Reaching puberty:
Odds ratio
# days delay
Odds ratio
# days delay
Reaching puberty,
Months delay
VIQ, PIQ, FSIQ ø

4th quartile PFOS
PFOA
0.46 (0.29, 0.71) 0.75 (0.49, 1.15)
190 days delay
69 day delay
0.55 (0.35, 0.87) 0.57 (0.37, 0.89)
138 days delay
130 days delay
3rd tertile: PFOS, 1.5 (-2.5, 5.4) months delay;
PFOA, 5.3 (1.3, 9.3) months delay
Age 5: VIQ
PIQ
FSIQ
PFOS:-1.7 (-4.0, 0.7) -2.2 (-4.7, 0.3) -1.9 (-4.3, 0.5)
PFOA: 0.9 (-1.4, 3.3) 1.0 (-1.4, 3.4) 1.2 (-1.0, 3.5)
PFNA: 0.7 (-1.3, 2.7) -1.4 (-3.4, 0.6) -0.2 (-2.1, 1.7)
Age 8: VIQ
PIQ
FSIQ
PFOS: -1.3 (-3.6, 1.1) -1.6 (-4.0, 0.7) -1.9 (-4.3, 0.4)
PFOA: 0.5 (-1.5, 2.5) -1.1 ( -3.2, 1.0) -0.4 (-2.5, 1.7)
PFNA:-2.1 (-3.9, -0.2) -1.5 (-3.5, 0.4) -1.5 (-3.4, 0.4)
PFOA evaluated. 4th quartile PFOA had higher IQ
scores than 1st quartile and decreased scores for
ADHD characteristics.
Slight, inconsistent results for PFOS and PFOA.

IQ, reading,
language, memory,
(etc.)
Hyperactivity
symptoms
ADHD
Learning problem

Stein 2014b
C8
Fei 2011
Denmark

320 children, 6-12 years

Hoffman 2010
NHANES, 19992000, 2003-2004

571 children aged 12-15

22.6

Ode 2014
Sweden

203 ADHD cases and 205 controls
(cord blood PFASs)

6.8

787 children, 7 years old∆

34.4

35

ADHD behaviors

5.4

HyperactivityѰ
CoordinationѰ

2.2

4.4

ADHD, ORs for IQR

1.8

ADHD

159

4th quartile, ORs: PFOS, 1.3 (1.0, 1.6); PFHxS, 1.6
(1.2, 2.1); PFOA, 0.7 (0.6, 0.9); PFNA, 1.2 (0.9, 1.5)
4th quartile, ORs: PFHxS, 1.2 (1.0, 1.4); PFOS, 0.9
(0.7, 1.0); PFOA, 0.9 (0.8, 1.1); PFNA, 0.7 (0.6, 0.9)
Inconsistent results (parents vs teachers; boys vs girls)
Conduct problem: PFOS OR=1.45 (0.77, 2.72);
PFOA OR=1.29 (0.67, 2.52)
Coordination problem: PFOS OR=1.39 (0.65, 3.00);
PFOA OR=1.14 (0.46, 2.81)
PFOS: OR=1.60 (1.10, 2.31)
PFOA: OR=1.35 (1.04, 1.77)
PFHxS: OR=1.19 (1.05, 1.34)
PFNA: OR=1.15 (0.93, 1.42)
PFOA, ≥75th percentile: OR=1.07 (0.67, 1.70)
(PFOS OR < 1.0)

Reference,
Location

Study population

Liew 2015
Denmark

215 ADHD∆

545 controls∆
Serum levels:
PFOS=27.4
PFHxS=0.9
PFOA=4.0

PFOS
serum
level
26.8

PFHxS
serum
level
0.8

PFOA
serum
level
4.1

213 ASD∆

25.4

0.9

3.9

Braun 2014
Cincinnati, OH

175 children tested at age 4 and/or
age 5∆

13.0

1.6

5.5

Strøm 2014
Denmark

876 adolescents∆

Outcome

Difference detected (95% CI)

ADHD

PFOA, 4th quartile: OR=1.14 (0.92, 1.40). OR=2.0
(1.5, 2.8) when all six PFAS included in model; PFOS
& PFHxS ORs <1.0. PFNA OR < 1.0, but when all 6
PFAS in model, the OR for PFNA=1.6 (1.2, 2.1)
PFHxS, 4th quartile: OR=1.07 (0.73, 1.56). When all
6 PFAS in model, OR=1.3 (0.8, 2.1). Per ln(PFHxS),
OR=1.10 (0.92, 1.33). OR<1.0 for PFOA, PFOS,
PFNA. When all 6 PFAS in model, PFOS OR=1.2
(0.7, 2.1)
PFOS: 1.6 (-0.8, 4.1) increase in SRS score per 2-SD
increase. PFHxS, 1.0 (-1.2, 3.3) increase; (decreased
SRS score for PFOA)
HRs < 1.0

ASD

SRS

ADHD (3.1%)
21.4

3.4
Depression (11.9%)

3rd tertile: PFOS HR=1.16 (0.69, 1.95); PFOA
HR=1.03 (0.61, 1.73)
Slight decrements for PFOS and PFOA

Chen 2013
Taiwan

239 children aged 2 years
Cord blood PFASs

7.4

2.6

Scholastic
achievement
Developmental delay

Forns 2015
Norway
Gump 2011
Oswego, NY
Vuong 2016
Cincinnati, OH

843 toddlers
Breast milk PFASs
83 children, aged 9-11

0.11

0.04

Developmental delay

8.79

3.67

3.28

Response inhibition

256 mother-child pairs (maternal
serum measured , 2nd trimester)
Children aged 5 and 8 years

12.6

1.4

5.3

Dong 2013
Taiwan

225 children, 12-15, w/asthma
231 children, 12-15, controls

33.9
28.9

2.5
1.3

1.2
0.5

Executive function:
Behavioral regulation
Metacognition index
Global Executive
composite
Asthma

160

PFOS associated with deficits in development scores,
especially for motor development: gross-motor
domain, IQR= -3.7 points (-6.0, -1.5); OR for poor
performance= 2.4 (1.3, 4.2). (Slight deficits to null
findings for PFOA)
PFOA, >median: OR=1.25 (0.81, 1.95);
PFOS, OR<1.0
All PFASs measured reduced inhibition
3rd tertile, ORs (clinical relevance):
PFOA
PFOS
PFHxS
1.36 (0.55, 3.35) 2.45 (0.91, 6.56) 2.03 (0.80, 5.18)
1.06 (0.43, 2.60) 2.17 (0.85, 5.51) 1.53 (0.63, 3.74)
1.25 (0.51, 3.08) 2.42 (0.92, 6.35) 2.31 (0.91, 5.88)
4th quartile, PFOA: OR=4.05 (2.21, 7.42)
4th quartile, PFHxS: OR=3.83 (2.11, 6.93)
4th quartile, PFOS: OR=2.63 (1.48, 4.69)
4th quartile, PFNA: OR=2.56 (1.41, 4.65)

Reference,
Location

Study population

Stein 2016a
NHANES

1,191 children aged 12-19,
1999-2000, 2003-2004

640 children (2005-2006)

PFOS
serum
level
20.8

PFHxS
serum
level
2.47

PFOA
serum
level
4.13

Outcome

Difference detected (95% CI)

MMR antibody

15.0

2.09

3.59

Allergic conditions

PFOS: among seropositives, a 13.3% decrease in
rubella antibody (-19.9, -6.2), and a 5.9% decrease in
mumps antibody (-9.9, -1.6).
(declines also for PFOA and PFHxS for rubella, &
PFOA for mumps)
PFOA: OR=1.28 (0.81, 2.04) for asthma (similar
findings for PFOS and PFNA); OR=1.35 (1.10, 1.66)
for rhinitis. For rhinitis, PFOS OR=1.16 (0.90, 1.50)
and PFNA OR=1.24 (0.97, 1.60).

Humblet 2014
NHANES

1,877 aged 12-19 years (1999-2008)

Goudarzi 2016
Japan

1,558 mother-child (aged 4 year)
pairs. Maternal serum PFAS at 3rd
trimester

4.9

0.28

2.1

Eczema, wheezing,
rhinoconjunctivitis

For total allergic diseases, 4th quartile ORs for PFOA,
PFOS, PFHxS and PFNA < 1.00. For wheezing,
PFOA OR=1.09 (0.73, 1.65) (elevation in boys only);
for PFNA, OR=1.11 (0.76, 1.63) (elevation in boys
only). For PFOS and PFHxS, ORs < 1.00

Granum 2013
Norway

99 prenatal blood samples;
50 children aged 3 years

5.5

0.3

1.1

Rubella

PFOA & PFHxS: β= -0.4 optical density (-0.64, -0.11)
PFOS: β= -0.08 optical density (-0.14, -0.02)
PFNA: β= -1.26 optical density (-2.32, -0.20)
(regression coefficients were also negative for
measles but confidence intervals were wide)

Gastroenteritis
Common cold

PFOA & PFHxS: ORs > 3.0 (CIs were very wide)
PFHxS: OR=1.71 (0.20, 14.8)

Food allergies,
sensitization (IgE)

4th quartile ORs
Self-reported allergies
PFOA: 9.09 (3.32, 24.9)
PFOS: 2.95 (1.21, 7.24)
PFHxS: 3.06 (1.35, 6.93)
PFNA: 1.73 (0.54, 5.52)

Buser 2016
NHANES

Current asthma
Wheeze

PFOS & PFOA sensitivity to mold; PFOA sensitivity
to rodents
PFOA, 3rd tertile, OR=1.18 (0.90, 1.53). (ORs for
PFOA, PFNA and PFHxS < 1.1.) For wheeze, all ORs
for the PFAS chemicals were <1.1)

Children aged 12-19, 2005-2006,
2007-2010

161

food sensitization
1.23 (0.57, 2.65)
0.74 (0.23, 2.40)
1.17 (0.56, 2.44)
0.51 (0.28, 0.92)

Reference,
Location

Study population

Wang 2011
Taiwan

244 children, aged 2 years
Cord blood

Dalsager 2016
Denmark

346 children aged 1-3 years.
Maternal serum PFAS <16 weeks
gestation

PFOS
serum
level
5.5

8.07

PFHxS
serum
level
0.04

PFOA
serum
level
1.71

Outcome

Difference detected

Atopic dermatitis

PFOS, 4th quartile: OR=2.19 (0.78, 6.17). OR for
PFOA and PFNA < 1.00

0.32

1.68

Fever

3rd tertile ORs (above median proportion of days)
PFOS: 2.35 (1.34, 4.11)
PFOA: 1.97 (1.07, 3.62)
PFHxS: 1.29 (0.72, 2.28)
PFNA: 1.49 (0.86, 2.59)
3rd tertile RRs (# days with fever)
PFOS: 1.65 (1.24, 2.18)
PFOA: 1.12 (0.82, 1.54)
PFHxS: 1.20 (0.89, 1.62)
PFNA: 1.12 (0.84, 1.49)
3rd tertile RR (fever & cough, # of episodes)
PFOS: 1.33 (0.99, 1.80)
PFOA: 1.11 (0.80, 1.56)
PFHxS: 1.13 (0.82, 1.55)
PFNA: 1.02 (0.76, 1.38)
3rd tertile RR (# episodes of diarrhea)
PFHxS: 1.71 (0.92, 3.16)
PFOS: 1.19 (0.67, 2.12)
PFOA: 1.08 (0.55, 2.13)

Fever & cough

Grandjean 2012
Faroes

532 children aged 5 years

16.7

0.63

4.06

Inadequate antibody
(<0.1 IU/mL), age 7)

ORs Tetanus (age 5)
PFOA: 4.2 (1.5, 11.4)
PFOS: 2.6 (0.8, 8.9)
PFHxS: 1.8 (1.1, 2.9)
PFNA: 1.6 (0.7, 3.6)

Grandjean 2016
Faroes

515 children aged 13 years

6.7

0.4

2.0

Diphtheria antibody

Geiger 2014b
NHANES

1,655 children, 1999-2000, 20032008 (aged 12-18)

% change per doubling of age 7 serum PFAS
PFOS: -31.1 (-49.8, -5.4)
PFOA: -9.4 (-31.1, 19.2)
PFHxS: -19.5 (-34.7, -0.7)
PFNA: -17.4 (-33.7, 2.8)
(Tetanus had increased % change for PFAS)
4th quartile OR:
PFOS: 0.69 (0.41, 1.17)
PFOA: 0.77 (0.37, 1.61)

Hypertension

162

Diphtheria (age 5)
3.3 (1.4, 7.5)
2.4 (0.9, 6.4)
1.5 (1.0, 2.3)
1.8 (1.0, 3.4)

Reference,
Location

Study population

Domazet 2016
Denmark

590 aged 9 years
444 aged 15 years
369 aged 21 years

Mora 2016
MA

1,006 children in early childhood
(3-6 years), 876 mid-childhood (611)

Karlsen 2016
Faroes

444 children at 18 months
371 children at 5 years
Maternal 2-week postpartum serum

PFOS
serum
level
42
21.5
10.5

PFHxS
serum
level

PFOA
serum
level
9.3
3.5
2.9

Outcome

Difference detected

adiposity

2.3

5.6

BMI, skinfold
thickness, total fat
mass index, waist
circumference

0.34
0.19

2.22
1.37

PFOS serum levels at age 9 was associated with
indicators of adiposity in adolescence and young
adulthood. PFOA serum levels at age 9 was associated
with decreased β-cell function in adolescence. Later
exposures were not associated with indicators of
adiposity or glucose metabolism.
Among girls, each interquartile increment of prenatal
PFOA was associated with 0.21 kg/m2 (-0.05, 0.48)
higher BMI, 0.76 mm (-0.17, 1.70) higher sum of
subscapular and triceps skinfold thickness, and 0.17
kg/m2 higher total fat mass index. Similar findings for
PFOS and PFHxS. No associations with boys.
3rd tertile ORs age 18 months
age 5 years
PFOS:
1.24 (0.98, 1.57) 0.94 (0.53, 1.66)
PFOA:
1.10 (0.84, 1.46) 1.88 (1.05, 3.35)
PFHxS:
1.24 (0.97, 1.58) 1.22 (0.73, 2.04)

24.7

4.68
8.04

Overweight

PFASs serum levels are in micrograms per liter (µg/L)
SHBG: sex hormone-binding globulin. (Note: Testosterone circulating in the bloodstream is mostly bound to SHBG. Endocrine disruptors
may bind to SHBG displacing reproductive hormones and affecting their bioavailability.)
FSH: follicle stimulating hormone
¶
5th vs 1st quintile
¥
PFNA=0.91 µg/L geometric in the Taiwan study, and 0.92 µg/L among Pease Tradeport children.
∆
The PFASs levels are for the mothers of this study population during their pregnancies.
€
89 of the children were tested at both ages (“paired children”).
Ѱ
high score on a screening test for hyperactivity/conduct problems; low scores on developmental coordination screening test.
ø
change in IQ with a doubling of PFASs level (Wang 2015)
Note: For the Nelson 2010 study, a monotonic dose-response was observed only for total cholesterol and PFOS among males 12-19. The
average differences in the table are based on the highest difference observed regardless of quartile observed (i.e., the largest difference could
appear in quartiles 2, 3 or 4). For PFHxS, the highest quartile was negative for total cholesterol but there was considerable inconsistency
between quartiles.
Note: For the Fei 2011 study, total scores for hyperactivity/behavior problem
163

Comments from the Pease CAP on the 5/23/17 draft Feasibility Assessment for Epidemiological
Studies at Pease International Tradeport, Portsmouth, NH, and ATSDR responses

A large number of comments on the 5/23/17 draft of the Feasibility Assessment from individual
members of the Pease CAP were received by ATSDR. ATSDR consolidated the comments and
provided the following responses.

Comment: Evaluate the health endpoints against the full spectrum of PFAS, not just PFOS and PFHxS.
Although PFOS and PFHxS were the highest PFCs in the Pease population, there are others that came
back higher in the Pease population when compared to the national averages such as PFOA and PFNA.
Current testing for PFAS in water does not include many PFAS often found in AFFF-contaminated
groundwater. In order to more thoroughly characterize PFAS exposures at Pease prior to May 2014,
would it be feasible to analyze a more extensive suite of PFAS, including novel PFAS, by researchers
who specialize in AFFF contamination in samples of groundwater (monitoring wells or from the Haven
well) and serum samples?
Response: The draft feasibility assessment emphasized PFOS and PFHxS because these were the two
PFAS with the highest concentrations in the Haven Well and were the most elevated in serum among
those who participated in the New Hampshire Pease blood testing program. Additionally, PFOS and
PFHxS were notably elevated compared to NHANES data. However, ATSDR intends to measure all
PFAS in serum and possibly urine for which analytical methods are currently available, and we expect
analytical methods to improve by the time a study is begun. We will not sample or analyze groundwater
but will rely on samples taken by the water utilities and/or DOD and other agencies to inform our
exposure assessments.
Comment: The Pease community wants a longitudinal study design, registry, and surveillance program
because PFAS are persistent and have long half-lives. The community wants to be monitored over time.
If a cross-sectional study is the best way to begin, it should be noted as such. Please explain the different
phases and step associated with what comes after the “first” study.
Develop a multi-site longitudinal preconception birth and children’s cohort. This would provide the
necessary sensitivity and statistical power to investigate the other proposed outcomes. Examples of
exposure and health outcomes that are useful and sensitive to the proposed study design are outlined
below:
• Preconception: Measures of maternal and paternal exposures and health including
fertility, miscarriage, thyroid function, and sex hormones
• Birth cohort: Measures of infant and maternal exposure and health including fetal growth,
birth outcomes, thyroid function, pregnancy induced hypertension, post-partum
depression, breastfeeding duration, and biomarkers in breast milk
• Children’s cohort: Measures of exposure and health from early life through puberty
including IQ/neurobehavioral, asthma and atopic dermatitis, rhinitis, antibody response,
ADHD, ASD, delayed puberty, thyroid function, and cholesterol
• Impacts to the immune system with children (especially with prenatal and early life
exposure)
164

Response: ATSDR appreciates the concern about long term effects of PFAS exposures and the Pease
community’s desire for longitudinal studies. Our first priority is to conduct cross-sectional studies of the
children and adults at Pease. A longitudinal study could then be conducted to evaluate future health
effects and make comparisons with the results of the cross-sectional study. If sufficient funding
becomes available, we will consider conducting prospective longitudinal studies (following participants
into the future). This was the approach for the C8 studies. In order to conduct longitudinal studies, the
contact information for individuals must be updated when they move so they can be re-contacted for
questionnaire interviews and additional blood samples. If funding is available, a “registry” or updated
mailing list could be established for this purpose. Collecting the person’s social security number, full
name and date of birth would facilitate linkages with cancer registry and hospital discharge databases. If
current contact information can be collected over time (e.g., if the participant can name 1-2 people who
will know where to find the participant), then children and adults could also be periodically surveyed to
obtain disease information.
For military service and civilian worker populations at military bases that had PFAS drinking water
contamination, ATSDR is considering conducting a retrospective longitudinal study to evaluate causes
of mortality and possibly cancer incidence (to see if past exposures resulted in current diseases). This
would be a data linkage study that does not involve contact with participants. This type of study is
discussed in the draft. ATSDR is not at this time considering a longitudinal surveillance program.
The draft feasibility assessment concluded that if multiple sites in addition to Pease were included in a
cross-sectional children study, then it was possible to evaluate the most of the diseases mentioned
among children with sufficient statistical power and precision. The assessment also concluded that
pregnancy-induced hypertension could be evaluated in a multi-site adult cross-sectional study. The
assessment did not evaluate the feasibility of studying infertility but did conclude that endometriosis
could be evaluated in a multi-site cross-sectional study.
The feasibility assessment concluded that immune effects in children such as asthma and atopic
dermatitis would likely require a larger sample size than can be achieved at Pease to achieve sufficient
statistical power. To study antibody responses to vaccines such as diphtheria and tetanus, a multi-site
study would be required.
The assessment did not address birth outcomes because such endpoints were not feasible at Pease and
may not be feasible even with a multi-site study. Studying birth outcomes may only be feasible at sites
where large populations such as entire cities were served with PFAS-contaminated drinking water over a
long period. Although ATSDR currently is not considering establishing a birth cohort and following it
longitudinally, other researchers, e.g., NIEHS-funded studies, have established birth cohorts exposed to
“background” PFAS levels that are being followed.
Comment: Consider studying: cardiovascular disease, thyroid disease, hypertension, osteoarthritis and
osteoporosis, endometriosis, liver disease (including non-alcoholic fatty liver disease), kidney disease,
rheumatoid arthritis, lupus, multiple sclerosis, thyroid function, cholesterol, lipids, potential endocrine
disrupting impacts, sex hormone impacts, cancers in both children and adults (including consideration of
incidence of brain cancers, RMS & PPB in children) and fertility issues from PFAS with both men &
women
Response: Effects of endocrine disrupting chemicals are wide-ranging and include adverse effects on
thyroid function, sex hormones, and fertility endpoints, as well as diseases such as endometriosis and
165

cancers of the breast and prostate. The feasibility assessment evaluated the possibility of studying some
of these endpoints at Pease. A larger sample size than can be achieved at Pease alone is likely needed to
evaluate thyroid function in children and adults as well as sex hormones in children and endometriosis in
adults with sufficient statistical power. The feasibility assessment concluded that lipids can be evaluated
in both children and adults at Pease. To evaluate cancers, a multi-site study would be required. The
feasibility assessment draft did not address early markers of fatty liver disease. We plan to address these
biomarkers in the final version. To evaluate liver function, liver disease, and kidney disease, a multi-site
study would be necessary.
Comment: Page 19 notes that the younger age limit of 4 was selected because of the appropriate age
range for testing IQ. However, younger children may experience elevated exposures in utero and via
breastfeeding from mothers who worked at the Tradeport prior to 2014 and continue to have elevated
body burden due to the long human half-lives of PFOS and PFHxS. Could the age range for children be
lowered from age 4 (even if child was not at daycare on Pease, many mothers were exposed) and
expanded to age 17? Is it feasible to include younger children (under age 4) when assessing endpoints
other than IQ?
Response: We realize that those < age 4 at the start of the study may have elevated PFAS serum levels,
but this fact alone is not sufficient to include those aged <4 in the study. Expanding the age range would
add few additional children into the study and would not change the health endpoints that are feasible to
study using the Pease population alone. However, we will consider expanding the ages for the children
study, but our decision will be based on the endpoints that will be evaluated as well as the degree of
difficulty convincing parents of very young unexposed children to allow their children to participate. We
will also consider whether it is appropriate to draw blood from those aged <4 years, the amount of blood
necessary to measure and evaluate the effect biomarkers.
Comment: Consider studying the endpoints listed as “not feasible to study” at Pease as part of a
broader/national study. For both the children and adult proposed studies, the second tier entitled
“Health-related endpoints that may be possible to study (although a larger sample size from the Pease
community will likely be needed,”
a)
b)

When will we know when a larger sample size will be needed, and
Please associate an estimated number that that larger sample size would need to be

For both the children and adult proposed studies, the third tier entitled “Health-related endpoints not
feasible to study,” would these endpoints need to be included in a national study? The health related
endpoints in this tier for both the children and adult studies are the big ticket health endpoints that our
community is worried about. This study design should be started right away.
a)
Are there current plans for a national study underway?
b)
Will the Pease CAP and community have input to that study?
c)
Who will be conducting that study?
d)
Can the national health study for these tier 3 endpoints be noted in writing in the Pease Study
design that a national health study will be conducted and Pease will be a part of it?
e)
If there are not plans in place for a larger sample size when the Pease study starts, will this data
still be proactively collected in the event a larger population is identified and participates in the study in
the future? Or will this data be collected at a later time and only when a larger population is identified
and enrolled in the study?
166

Response: Depending on available resources, ATSDR plans to conduct multi-site (or national) studies
of PFAS-drinking water contamination that would include the Pease population. This will permit the
study of health endpoints in the second tier. For the final version of the feasibility assessment, the
sample sizes needed for each endpoint in tier 2 and tier 3 will be included.
Our first step is to prepare protocols for pilot studies of children and adults at Pease that will be used to
inform the multi-site studies. The Pease CAP will have the opportunity to provide input for the study
protocols. ATSDR has not decided how the studies will be conducted. In the past, ATSDR has used
contractors to conduct the recruitment and data collection for studies. ATSDR then receives the data
from the contractors, analyzes and interprets the data, and prepares reports and/or journal articles. The
Pease population will be included in any multi-site or national study of PFAS-contaminated drinking
water. The resulting sample size should enable the evaluation of most if not all of the health endpoints in
tier 3.
A multi-site study that includes the Pease populations would include neurobehavioral testing. The serum
samples will be analyzed for all the effect biomarker endpoints listed in the feasibility assessment
including those that are listed as not feasible to study at Pease alone. In addition, the consent form for
the study will request permission to archive the serum samples so that additional effect biomarkers as
well as PFAS chemicals can be analyzed if necessary in the future. This will be made clear in the
protocol for the study.
Comment: Study former military/civilian population from PAFB and children (now adults) from
Discovery Enrichment Center or from schools for military children on base from years ago – these are
important to consider regarding latency of endpoints seeing that they could have been exposed at the
highest doses and for extended periods of time.
Response: We are considering studying former military and civilian workers at Pease and other military
sites. This is discussed in the draft feasibility assessment on page 41. ATSDR is in the process of
identifying military bases that had PFAS-contaminated drinking water and could be included in a
national study of military and civilian workers.
We can include those who attended daycare at Pease in the adult study if the daycare centers have
information sufficient to track and locate these individuals (e.g., full name, date of birth, sex, and if
possible, social security number). Similarly, if the schools have records of students that contain
sufficient information to track and locate these individuals, they could be included in the adult study.
Comment: Children endpoints considered a lower priority/possibly not include (not that they are not
important, but think it would help to lower # of endpoints to reduce time and financial commitments for
this study):
•
Overweight/obesity
•
IQ/neurobehavioral (would eliminating this also allow a lower population age range?) (also cohort
for these being studied by NIEHS?)
•
ADHD
•
Autism (longitudinal cohort being analyzed by NIEHS)
* Based on feedback, I also think it will be a challenge for parents to allow access to children’s school
records if they are a requirement for some of these endpoints, possibly impacting recruitment
numbers
167

Response: We believe that neurobehavioral effects may be important to study, especially effects on
executive function and attention. A few studies have found associations between exposures to PFAS
and these endpoints. Although NIEHS may be evaluating these endpoints, the cohorts have lower
PFHxS and PFOS exposures than the Pease population. As stated above, expanding the age range for
children will not add appreciably to the sample size, and it may not be appropriate to draw the amount of
blood necessary for PFAS and effect biomarker analyses from those aged <4 years. Studying autism
would require a larger population than at Pease to achieve sufficient statistical power, and access to
student records would be necessary. It is possible to evaluate ADHD via questionnaire, as was done in
the C8 study and NHANES study, although access to school records would be useful to confirm the
diagnosis. Overweight/obesity can be ascertained via questionnaire, and is worth studying because a
few studies have found associations.
Comment: With a broader national study, is it possible to not eliminate participants based on TCE being
a confounding factor, but instead use it as a study focus in combination with PFAS? (TCE seems to be
just as common as PFAS on military installations, so it may be important to study how being exposed to
a combination of both these classes may increase people’s risks to certain cancers/diseases).
Response: Before we attempt to study any interaction effects of exposures to both TCE and PFAS
chemicals, it is important first to establish the health effects of PFAS exposures. Most populations will
not be exposed to both TCE and PFAS chemicals. Even at military installations, TCE contaminated
drinking water is not very prevalent. Therefore, we will attempt to identify populations exposed to
PFAS-contaminated drinking water who are not also exposed to TCE and other solvents. On the other
hand, some PFAS-contaminated drinking water may also have disinfection byproducts such as
chloroform and trihaloacetic acids. These disinfection byproducts will be taken into account as possible
confounders in any analyses.
Comment: Consider that adult females that were exposed may have lower PFAS levels due to
menstruation and nursing
Response: In the protocol, we will discuss methods to take into account the elimination of PFAS from
the body due to menstruation, pregnancy and nursing, as well as other ways that PFAS may be
eliminated (e.g., blood transfusions).
Comment: Look at potential study focuses for the shorter chain PFAS that were measured in Pease
serum to expand data on potential effects
Response: We intend to measure all the PFAS chemicals that can be currently analyzed in serum and
urine. We also intend to archive serum and urine samples for future analyses of PFAS chemicals.
Comment: In my opinion, this assessment will be limited by the number of persons available at Pease,
and will not be the best means to provide data relative to all of the health impacts that our affected
families are interested in so they can best monitor with their health providers. As such, I believe that it is
imperative that this study be developed as a pilot program for a national study to include other affected
populations so that this opportunity to gather meaningful health information for the affected public is not
missed.
Response: We agree and plan to have Pease be a pilot study for the national study that will include other
sites with PFAS-contaminated drinking water in addition to Pease.
168

Comment: While this is not directly under the scope of ATSDR, we must advocate to all available
resources to find funding. In my own opinion, the Department of Defense has admitted fault, and also
has an obligation to remediate the site and contaminants. In my opinion that includes contaminants in an
affected population, not just in the physical property itself.
Response: No response needed.
Comment: In terms of the non-exposed populations for the comparison groups:
a)
Why are we restricting ourselves to just Portsmouth? Can we open this up to the Seacoast area?
b)
What protocols or precautions will be taken to ensure that the “non-exposed” population is really
non-exposed? For example, parts of Portsmouth have been affected by PFCs. There are low levels of
PFASs in two Portsmouth municipal wells (Portsmouth well and Collins well). These two wells are in
the southern well field with the three Pease wells (Smith, Harrison, and Haven well). If we are taking the
non-exposed from these parts of Portsmouth wouldn’t the results be skewed? We have learned at a
recent PFAS conference at Northeastern University in June 2017 that health effects are being discovered
in populations with low level PFAS exposure. Is there another community nearby that should be
considered as a control group that does not have known low level of PFAS in their drinking water?
The city of Portsmouth gets water from many sources (not just the Portsmouth & Collins well). Would it
be in the scope of ATSDR to perform water modeling on the low level PFAS in the two Portsmouth
municipal wells to determine how much PFAS exposure a Portsmouth resident was receiving while
drinking municipal water? Would this be factored into the control group design? Could this bias the
results of the control group?
Response: Except for exposure to PFAS-contaminated drinking water, the non-exposed population
should be as similar as possible to the exposed population on age, sex, socio-economic factors,
occupations, environmental exposures, and other potential risk factors. We anticipated that members of
the Portsmouth population who never worked or attended day care at Pease would be an excellent nonexposed comparison population. However, if we find that members of the Portsmouth population also
received PFAS-contaminated drinking water, then it may be necessary to identify another population
unexposed to PFAS-contaminated drinking water with similar characteristics as the Pease population.
ATSDR will take into consideration the PFAS concentrations in the Portsmouth public drinking water
supply before deciding on whether Portsmouth residents can be used as comparison populations for the
childhood and adult studies at Pease. If necessary, water modeling will be conducted to determine the
PFAS concentrations in the Portsmouth water supply.
Comment: NH recently admitted publicly that PFCs are migrating from the Coakley Landfill/Dump to
residential wells and that the Air Force disposed of industrial waste there. (Previously the AirForce said
it was only household waste that they disposed of at Coakley.)
Does this recent development change the definition/scope of the exposed population? If so, that may
change the endpoints that would be feasible to study as that would increase the sample size.
Response: If residential wells have been contaminated with PFAS leaching from the Coakley landfill,
then we would consider including in the study those residents with contaminated wells. However, at
present we are not aware of any wells currently impacted by the leachate.
169

Comment: Adult study age starts at age 18. Children study age end at 16. Please include anyone who
will be 17.
Response: The feasibility assessment now includes those aged 17 in the children study.
Comment: What about including the children who were either developing in utero or were breastfed by
mothers who worked on Pease who never attended daycare on Pease and will not be 4 when the study
starts? PFAS pass through the placenta and through breastmilk so these children were also exposed.
Response: Currently, the draft feasibility assessment includes children exposed in utero or via
breastfeeding if their exposures occurred prior to the closing of the Haven Well (i.e., exposure occurred
prior to June 2014) and their ages are between 4 and 16 years at the time of the study.
Comment: After all comments are compiled and the assessment is in a more final stage, what is the
review process for the Pease CAP and community to see the next draft? Will that need to be approved
by the CDC again?
Response: The final version will need to complete the clearance process including CDC review. Any
proposed changes to the feasibility assessment, and responses to the comments received, will be
discussed with the CAP prior to finalizing the document.
Comment: Will a detailed timeline be provided as to when the study would start, research would begin
and a report drafted?
Response: Timelines will be included in the protocols for the studies.
Comment: Is there any data/value that you can gain by receiving the blood samples from NHDHHS’s
blood sampling program? Given NH DHHS’s blood sampling program is still open, would it be
beneficial for NH DHHS to consent the new people and take more blood to be a part of this? Has it been
determined that if those PFAS samples from 2015 are still available? And assuming new consent can be
obtained from the participants, is there any valuable data that can be extracted from these blood samples
that were taken at a critical window in time shortly after the PFAS exposure was identified?
Response: Because the consent form for the NH blood sampling program did not mention analyses
other than PFAS measurements and did not request permission to provide ATSDR (or any other entity
other than the NH DHHS) with the blood samples, ATSDR cannot access these samples for further
analyses. However, the PFAS results for each individual who participated in the 2015 or 2016 blood
sampling program are useful for the proposed studies, and the consent form for the proposed studies will
request access to these results from the individual participant or, if necessary, from the NH DHHS. The
sample sizes would be too small to be beneficial. Also, the amount of blood being drawn is insufficient
for the effect biomarkers being considered.
Comment: Would it be feasible to archive blood samples from each participant for future analyses with
more extensive target analyte lists, and possibly other biological endpoints?
Response: Yes. We plan to include in the consent form a request for archiving the samples.
Comment: PBPK modeling? What is that?
170

Response: Physiologically based pharmacokinetic (PBPK) modeling is a mathematical technique for
predicting the absorption, distribution, metabolism and excretion (ADME) of chemical substances in
humans and animals. In the C8 study, PBPK modeling was used in conjunction with information on the
drinking water contaminant levels of PFOA, the residential history, water consumption habits, and the
cross-sectional serum PFAS levels in order to historically reconstruct PFAS serum levels. The
historically reconstructed PFAS serum levels were used to estimate cumulative exposures as well as
estimate PFAS serum levels during critical periods in the past (e.g, in utero exposure).
Comment: Should the study results show that a health endpoint was possibly or conclusively the result
of PFAS exposure, will a medical monitoring protocol be provided? Can this be written into the study as
a follow up/next steps? The community has continuously advocated for ongoing medical monitoring to
diagnose adverse health effects early and limit the impacts of the disease process on their health. This
very important point should be mentioned in the feasibility assessment as it is a significant community
request of ATSDR when addressing the health questions and concerns of the community. The point was
made clear several times in the feasibility assessment that a health study may not find conclusive
evidence of health impacts in the studied population. That point is all the more reason that medical
monitoring is critical for the impacted community despite a health study or in combination with a health
study.
Response: Medical monitoring is a separate issue and outside the scope of the feasibility assessment.
The feasibility assessment focused on whether epidemiological studies were feasible to conduct at
Pease, and if so, what studies might be feasible. As stated above, medical monitoring is a separate
program and is not part of an epidemiological study. The epidemiological study will be evaluating
endpoints that might be included in a medical monitoring program (e.g., lipids, liver and kidney
biomarkers). But these endpoints will all be evaluated in the same fashion (e.g., same method of
collection, same lab and same analytical methods) in order for the results to be comparable. This does
not occur with medical monitoring.
Comment: There are adults now that were exposed to PFAS at Pease while in daycare in the later
1990's and early 2000's. Where would this demographic fall into the study as they were exposed as
children, but are now adults in their 20's and never worked on Pease. This is a valuable population to
study as they were exposed many years ago and their data may offer insight into the potential long term
health effects many years after an exposure as a child.
Response: ATSDR will consider including these adults in the multi-site study when developing the
protocol for the pilot studies at Pease. However, there is concern that those who attended daycare and
are 18 years of age or older at the time of the study will be last exposed to the contaminated drinking
water 13 or more years ago, and therefore their PFAS serum levels will only poorly reflect their drinking
water exposures when they attended daycare. If ATSDR determines that it is possible to accurately
reconstruct historically their PFAS serum levels using modeling methods, then these adults could be
included.
Comment: Many of the children that attend daycare on Pease have one parent that works on Pease and
one parent that does not. Is it possible to identify these families and use the parent/adult that is not
working on Pease in the control group?

171

Response: As part of the development of the protocol for the adult study, ATSDR will consider
including the adult family member who did not work or attend daycare at Pease in the comparison
population.
Comment: The feasibility assessment states that comparison military bases would include those with no
PFAS-contaminated drinking water or drinking water contamination from other chemicals above the
U.S. Environmental Protection Agency’s maximum contaminant levels (MCLs). Does a military base
like this exist given the widespread use of AFFF by DoD in multiple branches of the military?
Are there considerations by ATSDR to study veterans from Pease and other DoD bases with known
PFAS expsoure?
Response: Although many bases used AFFF, much fewer had PFAS-contaminated drinking water. It
may be possible to conduct a study comparing bases with and without PFAS-contaminated drinking
water. Pease would be included in a study of military bases.
Comment: What happens if the study is funded and despite recruitment efforts, there is not enough
community enrollment to meet the sample size of 350 children and 1500 adults? Will the study start?
Response: Yes, the study will start. Our current position is that the Pease studies of children and adults
will be pilot studies for the larger national PFAS studies.
Comment: What is current plan to recruit study participants from other impacted PFAS sites? Residents
from several communities in Pennsylvania have expressed interest in being part of a national PFAS
study. Veterans and residents around the Wurtsmith Air Force Base in Oscoda, Michigan, have also
expressed interest in being part of a national study. How can these communities be recruited to be part
of a larger study that includes Pease community members to look at the endpoints that require larger
numbers to study?
Response: Our current plan is to conduct pilot childhood and adult studies at Pease and to include other
sites if sufficient resources become available. We would evaluate the same endpoints at Pease as we
plan to do with these additional sites.
Comment: The feasibility assessment states "The feasibility of successfully evaluating particular healthrelated endpoints (or effect biomarkers) could change depending on final study design and goals.” When
do these changes occur in the timeline of the study (before or during the study) and how is the CAP,
community, and participants informed of these changes?
Response: These changes would occur as ATSDR develops the protocols for the pilot studies at Pease.
The CAP will have an opportunity to comment on changes that occur as the protocols develop.
Comment: What treatment was put in place for TCE to make the wells "back in operation" in the fall of
1978? Why were the wells allowed to be in operation from 1978 through 1986 despite the TCE levels
"did not remain consistently below the current U.S. Environmental Protection Agency (EPA) maximum
contaminant level (MCL) in drinking water of 5 µg/L until January 1986"?
Response: These are questions better directed at the USAF staff with historical knowledge about Pease.

172

Comment: How has the AFFF use impacted the other two Portsmouth municipal wells (Portsmouth and
Collins well) in the southern well field where the three Pease wells reside? There are low level PFAS
chemicals detected in these wells based on sampling done by the Air Force since 2014.
Response: From information gathered, it does not appear the Portsmouth drinking water system was
contaminated with PFAS. All of the Portsmouth water sources, as well as two locations in the water
distribution system, were sampled for PFASs in May 2014 by the NHDES and during four rounds of the
USEPA’s UCMR3 performed between July 2014 and April 2015. The sample results were all below the
laboratory’s reporting limit for the PFASs tested. In June 2016 the NHDES requested all community
water systems to voluntarily collect a water sample for PFOA and PFOS and share the results.
Following this request, Portsmouth water operations staff sampled for PFOA and PFOS. A second round
of sampling was performed in November 2016 and the Greenland water-supply well that supplies the
Portsmouth water system with public drinking water was found to have an average level of 9 parts per
trillion (ppt) of PFOS. This level of contamination is one order of magnitude lower than the U.S. EPA
lifetime health advisory level (LTHA) of 70 ppt for combined PFOA and PFOS. It should be noted that
the levels were also flagged by the laboratory as “J” values, which means that they were an estimate.
Comment: Why was the Harrison well out of service?
Response: This question is better directed to the Portsmouth Department of Public Works.
Comment: The feasibility assessment states, "In 2013, sampling of monitoring wells at the former Pease
Air Force Base fire training areas detected PFOS and PFOA above these EPA provisional health
advisory levels."
I think it's important to note that the PFAS levels in the 2013 sampling of monitoring wells were in some
samples > 100,000 ppt for PFAS in the ground water.
Response: On page 10, the Feasibility Assessment now provides additional information concerning the
2013 sampling.
Comment: It is unclear to me why it took a year for the drinking water wells to be tested for PFAS
when significantly elevated levels of PFAS (> 100,000 ppt) were discovered in the monitoring wells on
the Pease Tradeport the year before. It would seem that the next logical and immediate step would have
been to test the drinking wells as soon as possible when the monitoring wells were discovered to be very
high.
Response: This comment is better directed to the US Air Force.
Comment: The feasibility assessment states "No water samples from the Pease Tradeport distribution
system for PFAS testing are available before 2014." Has this been confirmed with the DoD that they did
not sample for PFAS in the drinking wells prior to 2014?
Response: That is our understanding. We will check with the DOD to make sure.
Comment: Does ATSDR provide a list of contributors to the document (i.e. the specific staff that
contributed to the document)? Or is the document released from ATSDR as a whole? Did anyone else
other than ATSDR staff contribute to this document?
173

Response: No one other than ATSDR staff contributed to the document. The document is considered an
agency-wide, ATSDR report and will not list contributors.
Comment: How was the literature review and the data generated on the tables in pages 112 to 154
verified? Do multiple scientists validate this data? Is this part of what CDC reviews prior to clearing the
document for public view?
Response: The description of how the literature review was conducted is on page 77. The data are from
the published studies that were accessed by the literature search. ATSDR assumes that the data
contained in each published article has been validated by the article’s authors. As part of the agency
clearance process, CDC staff reviewed the entire document.
Comment: The feasibility assessment does not address an action plan in the event that the study shows
conclusive health effects in the studied population. Is it possible to add an action plan as to what the
community can expect for next steps if adverse health effects are identified in the study and they are
diagnosed with one of those health effects?
Response: The purpose of a feasibility assessment is to determine whether studies are feasible. A
feasibility assessment does not address actions to be taken once studies are completed. Once the studies
are completed, ATSDR will consider what additional steps need to be taken including whether medical
monitoring is appropriate.
Comment: Does ATSDR feel capable of conducting a national PFAS study on Pease and many other
sites impacted by PFAS contamination should funding become available?
Response: Yes
Comment: Does ATSDR collaborate with other scientists, agencies, and members from academia when
designing and performing health studies? If so, please describe that process? Involve intra and
extramural researchers at the National Institute of Environmental Health Sciences (NIEHS) in designing
and executing any health studies in order to leverage their expertise and experience conducting such
studies.
Response: On a few occasions, ATSDR has collaborated with other scientists and agencies on specific
health studies. But more often, ATSDR conducts studies by itself, relying on contractors to conduct
recruitment of study participants and data collection. ATSDR plans to have discussions with a group of
experts with PFAS study experience to receive technical assistance in the development of study
protocols.
Comment: Page 10 notes the use of PFAS in the manufacturing of AFFF “through 2001.” PFAS
continue to be used in AFFF, although newer formulations likely contain short-chain and other families
of PFAS compounds.
Response: We agree and have amended this sentence.
Comment: The PFNA concentration in the Haven well water was 17 ng/L in April and May 2014. In
2015, the New Jersey Drinking Water Quality Institute established a PFNA MCL for drinking water of
174

13 ng/L. Was the documentation for the New Jersey guideline consulted to identify potential health
endpoints linked to PFNA exposures?
Response: No, we did not consult the New Jersey guideline. Instead, we consulted the epidemiological
literature on PFAS to identify possible health endpoints for study. PFNA levels in serum in the Pease
population were similar to PFNA levels in the NHANES data. Based on the 2014 sampling of the Pease
supply wells, the concentration of PFNA in the distribution system would be less than 13 ng/L since the
Haven well provides about half of the supply, the other two wells had non-detects for PFNA, and the
water from all three wells were mixed at the treatment plant before entering the distribution system.
Comment: Why use a 2:1 ratio of exposed:unexposed participants in the children’s study but a 1:1 ratio
in the adult study?
Response: Exact figures for the number of children who attended the two day care centers at the Pease
Tradeport are not available, but it is possible that up to 1,000 children attended these centers from the
dates of opening through 2016. On the other hand, the Tradeport employs almost 10,000 workers at any
point in time since 1993, so the total number of workers who were employed at the Tradeport since its
opening through 2014 is likely to be over 20,000. This difference in the sizes of the daycare and worker
populations is also reflected in the demographics of the participants in the NH blood testing program at
Pease. Three-quarters of the participants in the NH blood testing of the Pease population would be aged
≥18 years in 2018 when a possible study might begin (N=1,092 excluding firefighters). On the other
hand, about 370 who participated in the program would be between the ages of 4 and 16 years in 2018.
So the number of exposed children that could be recruited for a study would be considerably smaller
than the number of exposed adults that could be recruited. Moreover, it was believed that it would be
more difficult to recruit unexposed children than unexposed adults. Therefore, we used a 2:1 ratio for
the sample size calculations for the children study and a 1:1 ratio for the adult study. Using a 1:1 ratio in
the sample size calculations for the children study would not have changed the list of endpoints that
were feasible to conduct at Pease. However, we will conduct sample size calculations using a 1:1 ratio
for children and include that in the report.
Comment: For the children’s study, would it be feasible to include fevers? As noted in Table A8,
Dalsager et al. 2016 found more frequent fevers in 1-4 year olds with higher prenatal exposures to PFOS
and PFOA and co-occurrence of fevers with coughing and nasal discharge.
Response: For the children study, the draft feasibility assessment recommended an age range of 4 – 16
years which is different than the age range of 1-4 years in the Dalsager et al 2016 study. The sample size
that was assumed to be feasible at Pease, i.e., 350 exposed and 175 unexposed, would be larger than the
Dalsager et al study. We will consider fevers as an endpoint and will attempt to do a sample size
calculation.
Comment: Page 17 notes the potential for reverse causation between elevated serum PFAS and certain
endpoints. A 2017 review by Rappazzo et al. notes that elevated PFAS exposures have been associated
in some studies with delayed age of menarche, which, in turn, may lead to relatively high serum PFAS
since blood loss during menstruation can decrease body burden. Should age of menarche be included in
regression models for girls in the children’s study?
Response: This comment is more appropriate for a protocol for the study. We will be obtaining this
information in the questionnaire because we are considering evaluating age at menarche as an endpoint
175

(i.e., delayed puberty). If age at menarche is a risk factor for a particular endpoint, then it would need to
be evaluated as a potential confounder and included in regression models if necessary.
Comment: Page 21 notes the role of NH DHHS in assisting with recruitment of participants. Has
DHHS committed to providing this assistance?
Response: We will approach NH DHHS once we obtain funding for a study.
Comment: Page 24 describes the method for estimating PFOA and PFOS over a child’s life for children
exposed at Pease. How will exposures in utero and through breastfeeding be included in these
calculations?
Response: This comment is more appropriate for a study protocol. In developing the protocol, we will
consult with researchers who have modeled PFOA serum levels and evaluate strategies for including in
utero and breastfeeding exposures in the historical reconstruction of serum levels.
Comment: Page 24 also notes that serum levels in the unexposed comparison group can assist in that
modeling. However, background levels in the unexposed group are shifting as population-wide
exposures to long-chain PFAS have decreased, and exposures to replacement compounds have
increased. Is it feasible to model the levels of PFAS across the lifetime of children in the unexposed
group?
Response: Yes. This was done in the C8 study for PFOA. We would take into account the decline in
PFOS and PFOA levels as indicated by NHANES data.
Comment: Page 29 describes the sample sizes in other pediatric immune system studies. The
Grandjean 2012 study evaluated a cohort of children that were all the same age, evaluated at ages 5 and
7. Antibody levels vary substantially over time following vaccination, and are best evaluated during a
relatively narrow window of time (e.g., 3 weeks post-vaccination). Is it feasible to evaluate antibody
response to vaccinations in children within a narrower age range?
Response: To evaluate vaccines such as rubella, tetanus and diphtheria, it would be necessary to
evaluate children before and after their booster shots at around age 5, as was done in the Grandjean
study. To obtain an appropriate sample size, it will be necessary to include children from other sites with
PFAS-contaminated drinking water in addition to Pease. During the NH blood testing program at Pease
in 2015, 28 children, 38 children and 54 children were aged one, two and three years, respectively.
These children would be in the age range 4-6 in 2018, the age range that receives booster shots for these
vaccines. Given that some children will not participate and others may have already received their
booster shot prior to the start of the study, it is unlikely that more than 60 children will be available for
study at Pease.
Comment: For evaluating immune system effects, would participants be excluded if they are receiving
immunosuppressant medications or have immune-related medical?
Response: This comment is more appropriate for a study protocol. Certainly if these conditions are the
endpoints under evaluation, we would not exclude children with these conditions. On the other hand, if
the evaluation is of immune biomarkers that would be affected by these medications, then they may be
excluded from the evaluation of these biomarkers.
176

Comment: Page 42 notes the considerations in developing a study of military personnel across sites. An
additional challenge is that the mixture of PFASs present at each site is likely different, due to different
formulations produced by various manufacturers. This is an important consideration because serum
levels PFOS, PFOA, and other known PFAS are proxies for the mixture of PFAS at each site. How can
the site-specific mix of PFASs be incorporated into a multi-site evaluation?
Response: Studies of military personnel across sites will utilize modeling methods in order to
historically reconstruct PFAS concentrations in drinking water. Ground water sampling data as well as
drinking water data will be used in the modeling effort. It is likely that the formulations used for AFFF
will be similar across military bases because of DOD requirements and bulk purchasing. Nevertheless,
we will use the modeling results to assess exposures at each base.
Comment: How will the results of blood testing for PFASs and other biomarkers be reported back to
participants in a way that helps them interpret their results for both exposure and clinical measurements?
Will community members have the opportunity to provide feedback about the format of the report-back?
Investigate interventions for improving health outcomes such as measures to reduce cholesterol, treat
thyroid disorders, nutritional supplements, and cancer screenings?
Response: How the results are reported back to study participants will be addressed in the protocol.
Community members would get an opportunity to provide feedback on the format of the reporting back
of individual results. Interventions is not within ATSDR’s mandate, but ATSDR could encourage other
researchers and other agencies (e.g., NIH) to pursue clinical trials for interventions.
Comment: The evidence presented in this feasibility plan focuses on the results of epidemiological
studies, yet these results are strengthened when considering results of toxicological studies in rodents
and other animals that can support biologically relevant pathways. Toxicological studies can also
provide compound specific health effects information, whereas interpretation of epidemiological studies
can be complicated by complex mixtures in exposure media. Is it feasible to use the results of
toxicological studies to prioritize health endpoints that seem most plausible, and to potentially identify
additional endpoints that have not previously been considered in epidemiological studies?
Response: Yes. We will consider toxicological information in order to identify and prioritize possible
endpoints.
Comment: As mentioned on page 41, assembling a multi-site cohort of military and civilian base
employees may be able to achieve sufficient power to detect meaningful risk estimates for cancer and
rare disease. This design has the benefits of efficiency and responsiveness to military and civilian base
employees, and the notable limitation of accurate exposure assessment. Therefore, is it feasible to
supplement investigation of these endpoints by also utilizing a prospective multi-site case-control study
nested within the cross-sectional design that has been proposed or the longitudinal surveillance program
and preconception birth cohort that have been suggested?
Response: A case-control sample is useful to obtain additional information on exposure status and/or
outcome status. However, it is not clear what would be accomplished by a nested case-control sample of
either the military and civilian employee cohorts or the cross-sectional study populations in the adult and
children study. For the military and civilian employee cohorts, we could ask cases and controls about
their water consumption during the period when they worked or were stationed at the bases, but it is
177

unlikely that the information will be accurate given the length of time since they last were at the bases.
We could also ask about their work or barracks locations when onsite, but we have found (from our
experience conducting studies at Camp Lejeune) that this information also is unreliable because of the
length of time that has passed. For the cross-sectional populations, we plan to collect serum to measure
PFAS levels, and we plan to estimate historical serum levels using information from the questionnaire
on water consumption and history at the site. So a case-control sample is unnecessary to improve the
exposure assessment. Moreover, a case-control sample is not necessary to improve outcome
ascertainment in the cross-sectional studies.
On the other hand, a case-control sample might be useful in the evaluation of specific birth defects and
childhood cancers in large populations exposed to PFAS-contaminated drinking water and comparison
populations. Residential histories and drinking water consumption prior to and during the pregnancy and
during the child’s early years of life could be obtained via interview of case and control mothers. The
accuracy of such information will depend on how far in the past the pregnancies occurred.
Comment: Aqueous film forming foam (AFFF) is a hugely complex mixture with a composition that
has varied over time and with company of origin. The mixture in groundwater will depend on these
factors as well as mitigating geological factors. This variability will also translate to variability in
exposures that will have occurred between sites, likely increasing the impact of exposure
misclassification bias. Therefore it is important to fully characterize the mixture of contaminants that
was in the drinking water as well as the most appropriate combination of exposure biomarkers.
Response: It is likely that military bases used similar formulations of AFFF. We will use recent ground
water and drinking water sampling results at these bases to estimate PFAS drinking water concentrations
in the studies of military and civilian worker personnel who were stationed or employed at these bases.
For the children and adult studies, we will analyze all the PFAS for which methods are available, in
serum and urine.

178