Relevance of frailty to mortality associated with the use of antipsychotics among community‐residing older adults with impaired cognition

PURPOSE: To examine the association between new antipsychotic use and mortality over 6 months among community-based older adults with cognitive impairment, and variation in risk by frailty and sex. METHODS: We conducted a retrospective cohort study of older (aged 66+) home care clients in Ontario, Canada, using linked administrative health and clinical databases. Included were clients with dementia and/or significant cognitive impairment assessed during April 2008 to March 2013. Frailty was defined using a validated 72-item index. Exposed were those newly dispensed an antipsychotic in the 6 months post cohort entry, with no such claims in the year prior to drug index date. Two-stage matching defined unexposed clients and their index date (matching on age, sex, frailty, assessment year, and propensity score). Outcome was time to death following index date. Cause-specific hazards models were used, and number needed to harm at 6 months was estimated from cumulative incidence function curves. RESULTS: Among 4955 matched exposed-unexposed pairs, new antipsychotic users showed a significantly increased hazard of mortality at 1, 3, and 6 months relative to unexposed, with the highest risk observed in the first month (hazard ratio [HR] = 2.08 [95% CI, 1.79-2.43]). At 1 month, risk was significantly higher for robust (HR = 3.72 [95% CI, 2.45-5.66]) vs frail (HR = 1.74 [95% CI, 1.40-2.17], P = .002) clients. The number needed to harm was 22.7 and did not vary by frailty but was lower for men (14.9) than for women (35.0). CONCLUSIONS: Risk of antipsychotic-associated mortality was highest in the first month following exposure, varied significantly by client frailty, and was greater among men than among women.