Antipsychotic use is an important issue in long-term care (LTC) homes due to their widespread use, the potential for serious adverse events and limited evidence about their efficacy in treating behavioural and psychological symptoms of dementia. Rates of antipsychotic use vary across LTC homes, but little is known about the contribution of physicians to this variation. This study documented the characteristics of physicians who regularly treated residents in Ontario LTC homes and examined variation in antipsychotic prescribing across physicians. In a population-based retrospective cohort of LTC residents, 637 (52.8%) of 1,207 LTC physicians cared for 46,365 (90.4%) of all residents. Overall, 27.3% of residents received antipsychotic therapy, but extremely high prescribers prescribed antipsychotics to 42.8% of their patients. Variation in physician antipsychotic prescribing persisted after controlling for clinical and behavioural resident characteristics. This variation was reduced by 47.1% when LTC homes were accounted for in multilevel cross-classification logistic regression models.

Antipsychotic use is an important issue in long-term care (LTC) homes due to their widespread use, the potential for serious adverse events and limited evidence about their efficacy in treating behavioural and psychological symptoms of dementia. Rates of antipsychotic use vary across LTC homes, but little is known about the contribution of physicians to this variation. This study documented the characteristics of physicians who regularly treated residents in Ontario LTC homes and examined variation in antipsychotic prescribing across physicians. In a population-based retrospective cohort of LTC residents, 637 (52.8%) of 1,207 LTC physicians cared for 46,365 (90.4%) of all residents. Overall, 27.3% of residents received antipsychotic therapy, but extremely high prescribers prescribed antipsychotics to 42.8% of their patients. Variation in physician antipsychotic prescribing persisted after controlling for clinical and behavioural resident characteristics. This variation was reduced by 47.1% when LTC homes were accounted for in multilevel cross-classification logistic regression models.