OBJECTIVES: This study aimed to describe regional variation in antidementia and antipsychotic medication use among long-stay nursing home residents with dementia.

METHODS: A cross-sectional design was used. We identified 273,004 long-stay nursing home residents with dementia and continuous Medicare Fee-For-Service from August-November 2018 (living in 9,735 nursing homes located in 289 hospital referral regions (HRRs). Antidementia medication and antipsychotics were determined from Part D claims covering November 1, 2018. Using 2018 Minimum Data Set 3.0 linked to Medicare data, facility information, and Dartmouth Atlas files, we developed separate multilevel logistic models (outcomes: memantine, cholinesterase inhibitor (ChEI), antipsychotic active use) to estimate odds ratios (aOR) and 95% confidence intervals (CI) for resident, facility, and HRR characteristics. Variance components were derived to quantify drivers of prescribing patterns.

RESULTS: Residents with dementia had Part D claims for memantine (16.1%), ChEI (23.3%), and antipsychotics (20.9%). Resident and facility factors accounted for much of the regional variation in antipsychotic use (proportion change in variation (PCV) state: 27.80%; PCV HRR: 39.54%); and HRR healthcare utilization at the end of life (EOL) accounted for most geographic variation (memantine PCV state: 37.44%; ChEI PCV state: 39.02%). For antipsychotics, facility factors such as any use of physical restraints (aOR: 1.08; 95% CI: 1.05-1.11) or poor staffing ratings (aOR: 1.10; 95% CI: 1.06-1.14) were associated with increased odds of antipsychotic use. HRRs with the highest EOL healthcare utilization had higher odds of antidementia use than those in nursing homes in HRRs within the lowest health care utilization (aOR memantine: 1.68; 95% CI: 1.44-1.96).

CONCLUSIONS: Geographic variation in antipsychotic and antidementia medication use in nursing homes suggests the need for evidence-based protocols to guide appropriate use. Understanding the nursing home factors that may prevent the implementation of evidence-based protocols for the appropriate use of these medications would be useful.