Identifying a National Cohort of Medicare Beneficiaries Residing in Assisted Living Settings: An Updated Method.

Abstract

Present an updated approach to identifying Medicare beneficiaries residing in licensed assisted living (AL) settings in the United States. Retrospective cohort study using a national list of licensed AL settings, US Postal Service (USPS) data, and enrollment, claims, and assessment data from the Centers for Medicare and Medicaid Services. A total of 403,326 beneficiaries residing in 29,905 licensed AL settings. We identified every ZIP+4 code affiliated with each AL address. We then identified all of the Medicare beneficiaries with that ZIP+4 on January 1, 2019, and excluded beneficiaries in nursing homes and hospitals on that date. We identified beneficiaries who were "definitively" and "very likely AL residents"according to the number of addresses corresponding to the ZIP+4 in the USPS data, the capacityof the AL setting, and the presence of a claim or assessment indicating services were delivered inAL. We compared beneficiaries excluded during our new capacity restriction step (ie, "possibly neighbors")to those included as being "definitively" and "very likely AL residents" using standardized mean differences. The cohort excluded (ie, "possibly neighbors") using our new step in the identification process appears to be younger and healthier than the cohorts we include as being "definitively" and "very likely AL residents." In addition, the cohort we identified through our added step of supplementing with claims and assessment data have similar demographics to the other cohorts we include, although they appear to be in poorer health. Leveraging licensed capacity information and supplementing with claims and assessment data produce greater confidence in the ability to accurately identify AL residents using ZIP+4 codes reported in Medicare administrative data.