ABSTRACT Background Antipsychotic medications remain widely used in nursing homes (NHs) despite guidelines discouraging potentially inappropriate use. While home‐level variation is known, less is understood at the unit and organization levels, or across other psychotropic and non‐psychotropic drugs. We examined prescribing variation across NHs in Ontario, Canada. We examined prescribing variation for psychotropic and non‐psychotropic drugs across NH units, homes, and organizations in Ontario, Canada. Methods We conducted a retrospective cohort study of NH residents using EMR data from April 1, 2022, to March 31, 2023. Outcomes included whether residents were prescribed any antipsychotic, antidepressant, statins, cardiovascular disease (CVD) medications, and trazodone for at least 1 day during the follow‐up period. Multi‐level logistic regression models adjusted for resident characteristics and included random intercepts for units, homes, and organizations. Results Among 16,896 NH residents across 179 homes and five organizations, 39% were prescribed antipsychotics, 65.5% antidepressants, 33% statins, 65% CVD medications, and 38% trazodone. Substantial variation was observed across all medications. For antipsychotics, the home‐level median odds ratio (MOR) was 1.65 (95% CrI: 1.52–1.78) and unit‐level MOR was 1.24 (95% CrI: 1.10–1.33). Trazodone exhibited the highest variation (home‐level MOR: 2.06, 95% CrI 1.75–2.37; unit‐level MOR: 1.87, 95% CrI 1.76–1.97). No significant variation was found at the organization level. Resident characteristics, like aggression and dementia, were associated with prescribing but did not explain unit and home‐level differences. Conclusions Prescribing variation extends beyond antipsychotics to other common pharmacotherapies. Unit‐level variation reveals additional unexplained variation not captured in previous home‐level studies, providing new targets for research and quality improvement interventions. Future research should explore potential unit‐ and home‐level contributors, such as staffing, practice culture, clinician decision‐making, and policies.
Dash et al. (2026) studied this question.