Self-reported chronic disease has been widely used in aging and epidemiological research to estimate population-level prevalence of common chronic illnesses. The Health and Retirement Study (HRS), like other national health interview surveys, asks respondents at each interview wave whether they have been diagnosed with a variety of chronic conditions by a physician, and in re-interviews, respondents are given the option to dispute previous affirmative responses of disease diagnosis. In addition, the HRS not only provides access to self-reported survey data, but also linkage of publicly reported data to restricted-access administrative data for Medicare- eligible participants as well as pharmaceutical and biophysical data for a subset of respondents in selected years. This project proposes to examine the concordance between self-reported and restricted data derived chronic disease, examine the time consistency of self-reported chronic disease, and finally, assess the performance of adjudication procedures developed to handle inconsistent patterns of self-reported chronic disease. Our project team has been involved in developing procedures that examine additional self-reported data in the HRS on medication usage, receipt of specific treatments, and corroborating events to adjudicate inconsistencies across an individual's longitudinal record. We propose to assess how well unadjudicated (raw) self-reported data corresponds to health record data available in the linked HRS restricted files. Likewise, we also propose to assess how well adjudicated self-reported data resulting from our adjudication methodology corresponds to the linked HRS restricted health record data. We propose to analyze self-reported and restricted HRS data from 1998-2010. This project has the potential to inform and bring clarity to reported chronic disease prevalence over time and has implications for harmonization across longitudinal surveys. Should the adjudication procedure for self-reported data outperform the corroboration between raw self- reported data and health record data, there may be implications for using these self-reported data procedures in other longitudinal health interview surveys that do not have the capacity for linkage to restricted administrative and biophysical data. Investigators examining data from studies without supplemental health record data may be able to use these adjudication procedures to reconcile inconsistent longitudinal chronic disease patterns.
It is increasingly important to estimate population chronic disease trends in accurate, reliable, and cost-effective ways. This project will assess a methodology that uses a variety of health data to provide better information about patterns of self-reported chronic disease over time that are not clinically reasonable. The methods developed and assessed in this project could be very useful to health care researchers and provide valuable guidance for public health efforts in the accurate tracking of chronic disease in the population.