Graduate Student/Research Assistant UMass Chan Medical School, United States
Background: Antipsychotics are often used to manage severe behavioral symptoms but increase mortality for older patients with dementia. Despite national efforts to reduce use, antipsychotics are still routinely used in nursing homes. Descriptive studies of geographic variation in usage patterns are limited.
Objectives: We sought to describe geographic variation in antipsychotic use among long-stay nursing home residents with dementia. We evaluated the association between resident, facility, and hospital referral region (HRR) characteristics and antipsychotic use.
Methods: A cross-sectional design was used. We identified 273,004 long-stay U.S. nursing home residents with dementia and continuous Medicare Fee-For-Service from August-November 2018 (living in 9,735 nursing homes located in 289 HRRs). Antipsychotic use was 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 (outcome: antipsychotic active use) to estimate odds ratios (aOR) and 95% confidence intervals (CI) for resident, facility, and HRR characteristics. Variance components of crossed random effects of HRRs and states were estimated to calculate percent change in variation (PVC) and were derived to quantify drivers of prescribing patterns.
Results: About 20.9% of residents with dementia had Part D claims for antipsychotics. Geographically, antipsychotic use varied with higher concentrations in the south and mid-west. Resident and facility factors accounted for most of the regional variation in antipsychotic use (PCVHRR=39.5%, PCVSTATE=27.8%), with gains achieved by adding HRR characteristics to the model. Resident factors such as < 65 years old (aOR: 1.23; 95% CI: 1.17-1.28), moderately impaired cognitive function (aOR: 1.33; 95%, CI: 1.30-1.36), very severe aggressive behaviors (aOR: 2.83; 95% CI: 2.56-3.12), MDS documented diagnosis of schizophrenia (aOR: 12.06; 95% CI: 11.68-12.45), or bipolar disorder (aOR: 4.06; 95% CI: 3.92-4.22) were associated with increased odds of antipsychotic use. Facility factors such as use of physical restraints (aOR: 1.08; 95% CI: 1.05-1.11), poor staffing ratings (aOR: 1.10; 95% CI: 1.06-1.14) or being in a rural area (aOR: 1.09; 95% CI: 1.06-1.12) were associated with increased odds of antipsychotic use.
Conclusions: Geographic variation of antipsychotics can be largely explained through resident and facility factors. Understanding these resident and nursing home factors can potentially help target interventions and provide useful insights regarding how best to assure appropriate dementia care in the nursing homes setting.
