AHRQ OMB Questions 8-10-07

AHRQ OMB Questions

1. please revise all surveys so that the race/ethnicity question complies with the OMB standard – Change Made (see Survey attachments 4, 6, and 8 – changes highlighted in yellow)

2. please revise all surveys to include the standard PRA blurb - – Change Made (see Survey attachments 4, 6, and 8 – changes highlighted in yellow)

3. the supporting statement for the ICR says that no incentives will be provided except in the case of focus groups. Therefore, please take out all references to incentives in this IC. – Changed “incentives” to “non-monetary tokens of appreciation” valued at less than $4 – see Section A9, highlighted in green

4. how will AHRQ generalize the results of these studies, if at all? – While this is not a representative sample, due to resource limitations by AHRQ, we will be able to cross-reference the results with the demographic information that is available for the target audience members.

5. if one of the things AHRQ wants to assess is familiarity with AHRQ, sampling from people who have made repeated requests for information from AHRQ would seem to give you a skewed sample. –

Answer: We have modified the 5^th objective to more accurately reflect the desired information from the specified sample. Specifically, the purpose and five objectives for this data collection will now be:

…the purpose of this evaluation is to gather information from recipients of a specified publication, the 2005 and 2006 versions of Guide to Clinical Preventive Services (Guide), in order to (1) inform future product versions or enhancements and (2) examine the extent to which, and how, they have used the guides in clinical practice. To achieve this purpose, the project seeks to meet the following five objectives:

1. To determine the extent to which the target audience accepts these guides.

2. To determine target audiences’ attitudes toward these guides.

3. To determine how the target audience uses these guides and to what extent it has improved care.

4. To learn ways to strengthen the content and format of future versions of these AHRQ guides.

5. To determine the extent to which the target audience is awareness of and satisfaction with AHRQ.

In addition, our experience with surveying medical professionals is that while they are familiar with a publication or resource, they tend not to be familiar with the Federal agency that produced the product. For example, recent focus groups we conducted with primary care physicians found that while physicians were familiar with the National Institutes of Health (NIH), they were not familiar with specific institutes within NIH. Further, they were not familiar by name of other Federal agencies that provide resources to healthcare providers. Thus, we think it to obtain information about familiarity from respondents.

Also, we do not know whether these individuals who agreed to participate in the survey have made multiple requests for information from AHRQ. We only know the list consists of individuals who previously agreed to participate when they ordered the publication being evaluated and/or who self-select to respond to a Web-linked survey.

6. please revise all uses of the word “confidential” with “confidential to the extent permitted by the Privacy Act.” – Change Made (see highlights in yellow, on Pages 6 and 7)

7. the burden reported in table 1 does not seem to match the burden in attachment B1. Which is correct? - According to our records, the burden in Table 1 and in Attachment B1 are the same. A “total” line was added to Attachment B1, to reflect that they do match. (see pink highlight)

8. how will AHRQ assess non-response bias if an 80% response rate is not obtained? For every survey study, nonresponse is always present and needs to be considered. Non-response bias is something we consider whether or not the 80% response rate is obtained. The methods employed are designed to minimize non-response and maximize response rates (e.g.., use of incentives (in our case non-monetary gifts), repeated contacts (in our case follow-up emails, mailings or calls)). To address non-response bias, however, we will examine characteristics of responders versus non-responders available from the contact database (e.g., position, organization, zipcode). In addition, for non-responders we will make additional efforts to locate them if we were not successful originally (e.g., confirm the email address remains the same for the organization, determine if the individual is still with the organization through non-invasize means (e.g., search an organization’s website). If a pattern emerges for non-respondents (e.g., more non-responders held a certain position), we will seek to either obtain replacements or determine whether a focus group is necessary to better understand this subgroup of potential respondents.

Further, another strategy we use to mitigate the impact of non-response bias, when possible, is through the use of statistical procedures. For example, if it is determined that there was a low response rate for a particular profession, the data can be weighted in accordance with the profession composition of the total study population to obtain an estimate that compensates for the selective response.

9. the supporting statement says that skips will be used to minimize burden, however there do not seem to be any skips in the survey materials. Please revise. – There are skip patterns; please see green highlights in Survey Attachments 4,6, and 8