MEPS household survey respondents tend to report some medical conditions more accurately than others
Research Activities, January 2010, No. 353
Household reports on medical conditions of household members, derived from the Medical Expenditure Panel Survey (MEPS) of the Agency for Healthcare Research and Quality (AHRQ), tend to be more accurate for certain types of conditions than others, concludes a study by AHRQ researchers. The MEPS is a national annual survey of approximately 15,000 households and a sample of the medical providers from whom they receive care. Steven Machlin, M.S., and colleagues recently analyzed MEPS data from 2002 to 2005 to determine the sensitivity and accuracy of household reporting of medical conditions.
The analysis is based on household-reported condition information matched to provider-reported condition information classified into 1 or more of 23 broad condition categories. Overall, household reports were more accurate for conditions that were highly salient, caused pain, required hospitalization and/or ongoing treatment, had specific recognizable treatment, altered lifestyle, and/or affected daily life. Among the 23 condition categories examined, reporting sensitivity rates ranged from 93.8 percent to 37.4 percent, with a median of 70 percent.
Ten of the 23 condition categories had rates of 75 percent or higher. With the exception of three conditions, all categories had sensitivity rates of 60 percent or above. Both pregnancy/delivery and diabetes had the highest reporting sensitivity rates of more than 90 percent. Other conditions with rates between 82 and 88 percent were mental health/substance abuse disorders, thyroid conditions, and hypertension. The three condition categories with the lowest rates of reporting were cerebrovascular disease, osteoporosis, and anemia. Only one-third of persons with actual anemia were identified by survey respondents in the household as having the condition. Reporting was generally more accurate when conditions were classified into broader categories, such as heart disease versus coronary artery disease. In some instances, respondents used general terms, such as "back disorders," for conditions that clinicians identified with more specific terminology, such as osteoporosis. Therefore, the researchers urge caution when considering using MEPS household data to analyze conditions that are defined at an extremely detailed level and/or that may be underreported (e.g., anemia).
More details are in "Sensitivity of household reported medical conditions in the Medical Expenditure Panel Survey," by Mr. Machlin, Joel Cohen, Ph.D., Ann Elixhauser, Ph.D., Karen Beauregard, M.S., and Claudia Steiner, M.D., M.P.H., in the June 2009 Medical Care 47(6), pp. 618-625. Reprints (AHRQ Publication No. 09-R073) are available from the AHRQ Publications Clearinghouse.