Abstract
BACKGROUND: Nursing home (NH) care in the United States now includes many short-term admissions to skilled nursing facilities (SNFs) for postacute care. OBJECTIVE: To demonstrate the potential of the Health and Retirement Study (HRS) linked to administrative data to study this group. RESEARCH DESIGN: Descriptive retrospective panel study. SUBJECTS: HRS respondents between 2002 and 2010 linked to administrative data from the Centers for Medicare and Medicaid Services (CMS). MEASURES: NH use was defined in 3 ways: by survey responses, Medicare SNF claims, and mandatory NH assessments. RESULTS: In total, 8.5% of observation periods (ie, time between 2 consecutive survey dates or 2 years before initial survey) reported by the survey and 26.0% reported by administrative data indicated some NH use. There was 98% agreement between survey responses and administrative data when there was no indication of a NH observation in the administrative data. However, there was only 33% agreement between survey responses and administrative data when a NH stay was indicated in the administrative data. NH stays associated with SNF care were responsible for the discrepancy-they were not consistently captured by the HRS survey. Rates of agreement were highest when a proxy respondent was used, and lowest among respondents who rated themselves in excellent overall health. Rates of agreement were higher later in the decade than earlier. CONCLUSIONS: The HRS-Medicare-linked files enhance the ability of the HRS to examine the growing use of NH for postacute care as well as offer a more comprehensive view of who uses NHs.