Rates, Predictors, and Barriers of Deprescribing Antipsychotics for Behavioral Symptoms in Dementia

Abstract Antipsychotic medications (APMs) are commonly used to manage behavioral and psychological symptoms in dementia (BPSD), but it is recommended that APMs should be discontinued in a timely fashion when possible. We aimed to investigate the discontinuation rates, predictors, and barriers of discontinuing APMs for BPSD. We used Optum Clinformatics Data Mart database (01/01/2005 to 08/31/2024). Eligible patients were adults aged ≥65 years without prior psychiatric disorders taking APM continuously for ≥3 months, with a dementia diagnosis prior to the use of APM. The primary outcome was APM discontinuation, defined as an absence of refill for >30 days. We estimated cumulative incidence of discontinuation by calculating one minus the Kaplan-Meier estimate at 1, 3, 6, 12, and 24 months following the index date. Our study included 15,886 patients, with 524 (3.3%) haloperidol users (mean [SD] age, 82.94 [5.33] years; 327 women [62.4%]) and 15,362 (96.7%) atypical APM users (mean [SD] age, 82.18 [5.70] years; 9575 women [62.3%]). The cumulative incidence of discontinuation at 6 months of follow-up was 30.4% (95% CI, 29.5%-31.2%) among atypical APM users and 48.6% (95% CI, 43.0%-53.7%) among haloperidol users (P < .001 for the difference between haloperidol vs. atypical APMs). In the multivariate-adjusted Cox proportional hazard model, use of haloperidol, venous thromboembolism, Black or Asian race, and use of antiarrhythmics were associated with a significantly higher rate of APM discontinuation, and alcohol abuse or dependence, hip/pelvic fracture, use of anticoagulants, and use of dementia drug were associated with a significantly lower rate of APM discontinuation.