1-10-08 memo

Why has AHRQ decided not to include the linked sample component as part of this request?

AHRQ has not conducted the MEPS-IC linked sample component since 2002. There are two major reasons that collection is not anticipated. One is due to budget constraints that limited the funds available to do the collection. The second is because the compounded response rate for the link sample survey and resulting data quality for the years it was fielded were too low to produce reliable estimates.  The data have proved useful for modelling purposes, but no national estimates ever have been produced from the link sample data.  AHRQ may request clearance for the link sample collection in future years if collection improvements (particularly with regard to identifying the correct establishment based on household sample information) and budget allow.

How did AHRQ determine which questions lacked usefulness and quality and were therefore dropped from the questionnaire? (page 4 of part A)? What were the previous questions about services provided (i.e. the 10 questions relating to “services covered” categories)?

First, questions that must be kept from year-to–year are those used to produce either 1) tables of estimates published on the MEPS Web site or 2) estimates for Federal or State government agencies.  The remaining questions on the forms may be deleted or modified.

Each year, a review of the existing MEPS-IC forms is done to determine if there are any important new issues – as determined by AHRQ and other key researchers, plus Federal and State users -- regarding employer-sponsored health insurance that were not addressed in earlier questionnaires. Due to the need to minimize respondent burden by not creating too lengthy questionnaires, any addition to the forms must be balanced by the elimination of another question. So if an important new question is identified that generally means that an existing question must be deleted. Most deletions are made for that reason.

Questions most likely to be deleted include those 1) with the lowest research value, 2) with a low question-level response rate, and 3) related to practices that are so common or so unusual that the response is the same for virtually all establishments.

The services-related questions that were eliminated (these were check boxes indicating whether the service was covered – yes; no; don’t know) were:

• Routine mammograms

• Adult preventive care

• Well-baby/well-child care

• Orthodontic care

• Inpatient mental illness

• Outpatient mental illness

• Alcohol/substance abuse treatment

The MEPS-IC experience was there were many “don’t know” responses to these questions, due to the plan booklets and other documentation not providing this information (see the #2 reason for deletions, above). The MEPS-IC experience is consistent with Bureau of Labor Statistics findings that services are not well defined by health insurance plans in materials given to enrollees.

The BEA commented that an earlier release schedule for MEPS-IC data would be very useful. Is AHRQ able to provide for an earlier release schedule?

The BEA comment concerning the release schedule appears to be in the context of the MEPS-IC decision to “go current” – meaning that the collection year and survey year will be the same. The BEA notes that “If the new collection and dissemination method accelerates the timing by several months, this will allow BEA to use the data and lessen the magnitude of the revisions.” By converting from collecting the data retrospectively to collecting it currently, the MEPS-IC will be producing the private-sector data a full year earlier and the public-sector data approximately 9 months earlier. Therefore, BEA’s goal will be met.

 

Are the tables of estimates from the MEPS-IC reviewed by the DRB at Census or at AHRQ? 

The MEPS-IC private sector sample is drawn from the Census Bureau’s Business Register, which is covered by the confidentiality provisions of Title 13 and Title 26. Therefore, Census is responsible for any review before releasing the tables to AHRQ. AHRQ does not have access to the microdata from the survey except at Census under restricted conditions. AHRQ does review all tabular estimates before posting them to the MEPS website