Existing studies have not investigated the effectiveness of one long-acting injectable antipsychotic (LAI) versus another in preventing hospitalizations among patients with bipolar disorder (BD). This study was conducted to compare all-cause inpatient healthcare utilization and associated costs among BD patients who initiated LAIs.

This retrospective cohort analysis used the Truven Health Analytics MarketScan® Commercial and Medicaid claims database. Bipolar patients >18 years with at least one claim for one of the following LAIs were identified between 1 January 2013 and 30 June 2014 (identification period): aripiprazole, haloperidol, paliperidone, and risperidone. The first day of initiating an LAI was considered the index date. Logistic regression and generalized linear regression models were conducted to estimate risk of inpatient hospitalization and associated costs during the 1-year follow up.

A total of 1,540 BD patients initiated an LAI: 14.5 percent aripiprazole, 16.3 percent risperidone, 21.0 percent haloperidol, and 48.1 percent paliperidone. With the aripiprazole cohort as the reference group, the odds of having any inpatient hospitalizations were significantly higher in haloperidol [Odds Ratio, OR (95 percent Confidence Interval, CI): 1.49 (1.01 - 2.19)] and risperidone [1.78 (1.19 - 2.66)] cohorts. The paliperidone cohort also had a higher risk of having a hospitalization than aripiprazole, but the difference was not statistically significant (p>.05). Among LAI initiators having any inpatient hospitalizations, the adjusted mean all-cause inpatient costs were lowest in the aripiprazole cohort (USD26,002), followed by risperidone (USD27,937), haloperidol (USD30,411), and paliperidone (USD33,240). However, the cost difference was not statistically significant.

Our study findings highlight the value of aripiprazole in reducing all-cause inpatient hospitalizations and associated costs among patients with BD during the 1-year follow-up. It is worthwhile to note that bipolar diagnoses were identified from healthcare claims coded for reimbursement purposes, thus misclassification was possible. Future studies are warranted to understand the impact of LAI use in a longer period of time.