Abstract
IMPORTANCE: Long-acting injectable antipsychotics (LAIs) can help decrease the rate of nonadherence to medications in patients with schizophrenia, but these drugs are underutilized in clinical practice, especially in Asian countries. One strategy for the early prescription of LAIs is to administer the drugs during patients' first admission, when they have more time to absorb medication-related knowledge. OBJECTIVE: To estimate the prevalence of and risk factors for in-hospital use of LAIs among first-admission patients with schizophrenia in Taiwan and to examine the association of early discontinuation with readmission risk among patients receiving LAIs. DESIGN, SETTING, AND PARTICIPANTS: This cohort study included data from a claims database for patients with a first admission for schizophrenia at psychiatric wards in Taiwan from 2004 to 2017. Eligible patients were diagnosed with schizophrenia or schizoaffective disorder at discharge and aged between 15 and 64 years. Data analysis was performed from April to September 2022. EXPOSURE: In-hospital use of LAIs with or without early discontinuation. MAIN OUTCOME AND MEASURES: Readmission for any psychotic disorder following discharge from first admission, with risk estimated via multivariable survival regression analysis, including the Cox proportional hazards (CPH) model and accelerated failure time (AFT) model. RESULTS: Of the 56 211 patients with a first admission for schizophrenia (mean [SD] age, 38.1 [12.1] years; 29 387 men [52.3%]), 46 875 (83.4%) did not receive any LAIs during admission, 5665 (10.1%) received LAIs with early discontinuation, and 3671 (6.5%) received LAIs without early discontinuation. The prevalence of receiving LAIs increased by 4%, from 15.3% (3863 of 25 251 patients) to 19.3% (3013 of 15 608 patients) between 2004-2008 and 2013-2017. After controlling for sex, year, prior antipsychotic use, age at first admission, and length of stay, the CPH regression analysis revealed that the readmission risk increased among patients receiving LAIs with early discontinuation (adjusted hazard ratio [aHR], 1.25; 95% CI, 1.21-1.30) but decreased among patients receiving LAIs without early discontinuation (aHR, 0.88; 95% CI, 0.84-0.92) compared with patients not receiving LAIs. Results remained similar for the AFT model. CONCLUSIONS AND RELEVANCE: The incidence of in-hospital use of LAIs among patients with a first admission for schizophrenia has remained low. In this study, early discontinuation of LAIs was associated with readmission risk-specifically, early discontinuation with a higher risk while the lack of early discontinuation with a lower risk compared with treatment with oral antipsychotics alone-which suggests our results have implications for improving the efficacy of LAI administration among patients with a first admission for schizophrenia.