Ehr 2

                                          OMB Questions and Comments

 

1. Part A: Item 11:  Could you clarify the arrangements under which RTI will have access to the Medicare claims database in order to link the survey results?

We have proposed to link the Medicare claims database to our survey results in order to verify the survey items asking physicians to estimate the ethnic and racial makeup of their patient panels. This is not essential to our charge of measuring the rate of adoption of electronic medical records in the U.S. Further, we now have concerns about the accuracy of the patient race and ethnicity data in the Medicare claims database. We have decided to delete the item asking for physician’s universal provider identification number and will no longer be linking the data to the Medicare claims data base.

2.  Item 12, “500 minutes” should read “500 hours.”

Thank you for catching that typo. We will make the correction.

3. Part B 1.1, Why are the three types of excluded physicians being excluded?  How confident are you that your frame allows you to do so accurately?

We propose to exclude physicians who list anesthesiology, radiology, or pathology as their primary specialties. We are proposing to exclude them because these physicians are primarily hospital based. They generally do not have ambulatory care practices, which is the focus of this survey. Because they are primarily hospital based, they are also unlikely to have been involved in the decision to purchase and implement an electronic medical record in an ambulatory care setting.

The AMA Masterfile include the primary specialty as a datapoint for every physician in the United States. It is the gold standard sampling frame for this population. We are confident that the data on primary specialty is correct. Using the Masterfile to screen for specialty is a commonly accepted practice in physician surveys.

4. 1.2, Specifically what type of information is available in the AMA Masterfile to support non-response bias analysis?

Each AMA Physician Masterfile record includes the physician's name, medical school and year of graduation, gender, birthplace, and birthdate. Additional data (residency training, state licensure, board certification, geographical location and address, type of practice, present employment, and practice specialty) are added from primary data sources or from surveying the physicians directly as the physicians' training and career develop.

5. 2.1, please provide a copy of the advance letter and any other mailing pieces available for review

That material was provided to Jon Kraemer via email on June 6^th. They are attached again for your reference. Titled “EHR letters.zip” and “Supportingdoc.doc”

6. 2.1, to what extent to the personal visit component of NAMCS contribute to its total response rate?

Please note that we are only seeking approval of our survey as CDC goes through a separate OMB clearance process for their NAMC tool. I do know that NAMC is traditionally performed through face to face interviews with the physicians. Due to this, their response rate is very high (approximately 63%). However, in order to ensure a large enough (and fully representative) sample size, ONC is funding the CDC to also conduct an additional mailing of the survey to 10,000 physicians in 2008 and each subsequent year. CDC anticipates the response rate for the mailed survey will be closer to 50% after using both mail and phone reminders.

7. 2.2.2, please provide more specific information on stratification plans as well as which commercial vendors you anticipate using.

This section refers to the use of geocoding to characterize the sampling strata. Geocoding is the process of assigning demographic characteristics of the neighborhood to an address. Private vendors obtain this data from the U.S. Census Bureau. Claritas is one of the premier vendors for geocoding. This have been conducting research on the demographics of the U.S. population for 35 years.

At no time will our sample be set to a private vendor for geocoding. RTI has the capability to purchase the geocoding data from a private vendor and applying to the AMA Masterfile to create the strata for sampling. Our sample will remain in house at RTI.

We have proposed a stratified random sample in order to ensure that we have an adequate number of physicians residing or working in areas of the country with low median income or with a higher proportion of racial and/or ethnic minorities.

8. 2.2, When you say that “approximately 50 percent of the AMA Masterfile records contain the physician’s practice address,” do you mean that the file contains (only) home address information for the remainder?  Will questionnaires be mailed to home addresses?

The remaining physicians have chosen to list their home address in the AMA Masterfile. We expect that our sample will include approximately 50 percent of physicians who have listed their home address. We intend to mail questionnaires to the address listed in the database, as this is the physician’s preferred mailing address.

9. What steps will HHS take to confirm AMA information/fill in gaps in the Masterfile?

We will telephone non-responders physicians to ensure that we have the correct address and other contact information. However, we will not take further steps to confirm AMA information. The AMA Masterfile is considered to be the most comprehensive sampling frame for physicians in the United States.

It is not a common survey research practice to confirm the information in the AMA Masterfile. The AMA regularly updates the contact information. We understand that NAMCS also uses the AMA Masterfile for sampling and does not confirm addresses.

10. 3, What is the basis for “no more than four calls” during the third mailing time period?

Research on survey methods indicates that after four follow up calls, the utility of further calls decreases markedly. Further calls are unlikely to generate survey responses. Further, there is some data that indicates that the reliability of these responses (after 4 follow up calls) is questionable.

11. 3.2 Why does this section say “up to three telephone attempts” rather than four?

This is a typographical error. Thank you for catching it. We will correct it.

12. Physician Survey Questionnaire: 

1. I identified quite a number of typos, so just a note that this needs to be scrubbed.

Thank you. We will ensure that the final survey has no typographical errors.

2. 600.  What is the rationale for asking for a rating of the impact “among physicians generally?”  On what basis will a respondent reply--personal opinion?  If this question has been used effectively in the past, please describe.

We have asked the question this way in the past in a survey of solutions to the problem of medical errors and found the data compelling (Blendon, DesRoches et al “Views of Practicing Physicians and the Public on Medical Errors”, NEJM, December 12, 2002). Further, the construction of the question is related to the mode of administration. Because this is a mailed survey, it is important that we construct the questions in such a way that all of the respondents can answer it. This allows us to know, as accurately as possible, the denominator for each question, which allows us to calculate accurate estimates. Skipping respondents out of question batteries introduces the possibility of significant error in the results.

3. Will both questionnaires be sent in one envelope?  If so, is there any concern about physicians not wanting the practice manager to see their responses?

Separate envelopes will be provided for the practice manager and physician survey, enabling both participants to keep their responses confidential.

4. I cannot seem to locate the Practice Manager’s survey – in ROCIS, the link to it is actually to a copy of the Physician questionnaire.  Please provide.

The practice manager’s survey is Appendix C. I am resending it to ensure that OMB has received it.

13. Schedule – could we get an updated schedule?

I have attached an updated schedule to this email entitled “EHR study calendar.doc”

13. Please provide additional discussion of the non-response bias analysis HHS is planning to conduct.

We intend to compare responders to non-responders on medical school and year of graduation, gender, birthplace, birth date, board certification, geographical location, type of practice, and practice specialty.

13. Please provide clarification of why HHS believes incorporation of questions into the NAMC will offer an effective apples to apples comparison relative to the baseline that will be established through the proposed survey.  We would like to see a side by side of the questions asked on both the 07 Survey as well as the NAMC survey to used in 08 and beyond.  Additionally, please provide demonstrable evidence/citations of similar sample surveys from year to year

The questions posed in NAMC to measure adoption are identical to the questions that will be used to measure adoption in this survey. I have attached questions 20-24 of the NAMC survey (same questions to be used in 2007 and all subsequent years). These questions are identical to the questions in our physician survey (appendix B) which will capture the information necessary to calculate the adoption rate. Therefore, this will be an “apples to apples” comparison.

SEND:

Appendix B, C

PDF of NAMC

EHR letters.zip

SupportDoc.doc

EHR study calendar