Importance of Frailty for Association of Antipsychotic Drug Use With Risk of Fracture: Cohort Study Using Electronic Health Records.

OBJECTIVE:To evaluate association of first- or second-generation antipsychotic (AP) drugs with fracture risk at different levels of frailty over the age of 80 years. DESIGN:Population-based cohort study. SETTING AND PARTICIPANTS:United Kingdom Clinical Practice Research Datalink including 153,304 patients aged 80 years and older between 2006 and 2015. METHODS:Rates of fracture and adjusted rate ratios (RR) were estimated by AP drug exposure category, adjusting for age, sex, frailty, number of deficits, and dementia diagnosis. RESULTS:Data were analyzed for 165,726 treatment episodes (153,304 patients; 61.3% women; mean age 83 years; 21,365 fractures; 681,221.1 person-years of follow-up). AP exposure was associated with increasing age, frailty, and dementia diagnosis. After adjusting for frailty and covariates, first-generation AP exposure was associated with risk of any fracture, RR 1.24 (95% confidence interval 1.07-1.43, P = .003). Second-generation AP exposure was associated with femur fracture (RR 1.41, 1.22-1.64, P < .001) but less strongly with any fracture (RR 1.12, 1.01-1.24, P = .033). Fracture incidence increased with frailty level. The number of person-years of first-generation AP treatment associated with 1 additional fracture at any site was 75 (42-257) for severely frail patients but 187 (95% confidence interval 104-640) for 'fit' patients. For second-generation AP, 1 additional femur fracture might result from 173 (111-323) person-years treatment in severe frailty but 365 (234-681) person-years treatment for 'fit' patients. CONCLUSIONS AND IMPLICATIONS:Frail patients are more likely to receive AP drug treatment, but their absolute risk of AP-associated fracture is substantially greater than for nonfrail patients.