ImportanceThere are limited data on discontinuation rates of antipsychotic medications (APMs) used to treat delirium due to acute hospitalization in the routine care of older adults.ObjectiveTo investigate discontinuation rates and patient characteristics of APMs used to treat delirium following infection-related hospitalization among older US adults.Design, Setting, and ParticipantsThis retrospective cohort study was conducted using US claims data (Optum’s deidentified Clinformatics Data Mart database) for January 1, 2004, to May 31, 2022. Patients were aged 65 years or older without prior psychiatric disorders and had newly initiated an APM prescription within 30 days of an infection-related hospitalization. Statistical analysis was performed on December 15, 2022.ExposuresNew use (no prior use any time before cohort entry) of oral haloperidol and atypical APMs (aripiprazole, olanzapine, quetiapine, risperidone, etc).Main Outcomes and MeasuresThe primary outcome was APM discontinuation, defined as a gap of more than 15 days following the end of an APM dispensing. Survival analyses and Kaplan-Meier analyses were used.ResultsOur study population included 5835 patients. Of these individuals, 790 (13.5%) were new haloperidol users (mean [SD] age, 81.5 [6.7] years; 422 women [53.4%]) and 5045 (86.5%) were new atypical APM users (mean [SD] age, 79.8 [7.0] years; 2636 women [52.2%]). The cumulative incidence of discontinuation by 30 days after initiation was 11.4% (95% CI, 10.4%-12.3%) among atypical APM users and 52.1% (95% CI, 48.2%-55.7%) among haloperidol users (P < .001 for difference between haloperidol vs atypical APMs). We observed an increasing trend in discontinuation rates from 2004 to 2022 (5% increase [95% CI, 3%-7%] per year) for haloperidol users (adjusted hazard ratio, 1.05 [1.03-1.07]; P < .001) but not for atypical APM users (1.00 [0.99-1.01]; P = .67). Prolonged hospitalization and dementia were inversely associated with the discontinuation of haloperidol and atypical APMs.Conclusions and RelevanceThe findings of this cohort study suggest that the discontinuation rate of newly initiated APMs for delirium following infection-related hospitalization was lower in atypical APM users than in haloperidol users, with prolonged hospitalization and dementia as major associated variables. The discontinuation rate was substantially higher in recent years for haloperidol but not for atypical APMs.