(SLA-013) Evaluating Mortality Risk and Post-discharge Antipsychotic Use in Patients with Hospital-Onset Delirium

Background: Previous studies have suggested that antipsychotic use in delirium may increase mortality risk; however, evidence specifically regarding post-discharge use is scarce. Additionally, when comparing against an untreated comparator, approaches such as clone-censor-weight can address immortal time bias and ensure an aligned time-zero.

Objectives: To evaluate the mortality risk and post-discharge antipsychotic use in patients with hospital-onset delirium.

Methods: We conducted a cohort study using Taiwan’s National Health Insurance Research Database between 2011 and 2021 and included patients with hospital-onset delirium. We excluded patients with schizophrenia or bipolar disorder to minimize misclassification of antipsychotic use. To address immortal time bias, we used a clone-censor-weight approach, cloning each patient into both post-discharge antipsychotic user and non-user groups. We applied censoring and weighting to accommodate treatment deviations. Our primary outcome was all-cause mortality within 180 days after discharge. We followed up patients from the discharge date to death or the end of the database, whichever came first. We conducted the as-started analysis to assess mortality related to post-discharge antipsychotic use. Additionally, we also conducted a time-varying exposure analysis to provide different causal estimands. We compared hazard ratios (HRs) with 95% confidence intervals (CIs) across methods. We further restrict patients with hospital-onset delirium superimposed on dementia.

Results: A total of 219,962 patients with hospital-onset delirium were included (mean age 76.1 years, standard deviation 16.2). Under the clone-censor-weight approach, post-discharge antipsychotic use was associated with a reduced mortality risk (HR 0.84; 95% CI 0.82–0.86). The time-varying exposure analysis yielded a similar finding (HR 0.92; 95% CI 0.89–0.94). Moreover, for those with hospital-onset delirium superimposed on dementia, the results were consistent to the primary analysis.

Conclusions: Our findings suggest that post-discharge antipsychotic use may be associated with a reduced mortality risk in patients with hospital-onset delirium. However, the choice of analytical approach should account for causal estimands of interests.