Addition of long-acting injectable antipsychotics during manic episodes in bipolar disorder: A retrospective analysis of rehospitalizations.

Introduction: Bipolar disorder (BD) often necessitates hospitalization, especially during manic episodes. Long-acting injectable antipsychotics (LAIs) are theorized to enhance treatment adherence and decrease rehospitalization rates compared to oral medications. This study aimed to evaluate the real-world effectiveness of LAIs in reducing rehospitalizations among BD patients admitted for bipolar mania.

Methods: We conducted a retrospective cohort study using data from a tertiary psychiatry hospital in Taiwan spanning January 1st, 2006, to December 31st, 2017. We analyzed 2212 hospitalizations among 945 patients with bipolar mania. A mixed-effects Cox regression model compared rehospitalization hazards between LAIs, mood stabilizer plus oral antipsychotic (MS + OAP), and mood stabilizer only (MS) groups. Sensitivity analyses assessed robustness across various subgroup criteria.

Results: LAI treatment significantly reduced the hazard of rehospitalization within one year post-discharge compared to MS + OAP (HR = 2.29, 95 % CI = 1.56-3.36) and MS alone (HR = 2.66, 95 % CI = 1.68-4.21). This effect was consistent across different rehospitalization types-all-cause, bipolar disorder-specific, and bipolar mania-specific. Each additional previous hospitalization was associated with higher hazard of rehospitalization across the three rehospitalization types. Sensitivity analyses suggested LAIs' efficacy in manic episodes with and without psychotic symptoms and for patients with frequent hospitalizations. The LAIs included in the analysis are haloperidol, risperidone, fluphenazine, flupentixol, and zuclopenthixol.

Conclusion: Our findings suggest that the addition of LAIs for bipolar mania during acute inpatient treatment is associated with reduced rehospitalizations, particularly among patients with recurrent hospitalizations, making it a valuable option. However, the lack of outpatient prescription data limits our ability to further substantiate this concept, warranting future research.