Treatment costs and health care utilization for patients with bipolar disorder in a large managed care population

Objectives: This study measured the treatment cost of bipolar disorder (BPD), decomposed the cost into that portion which was directly BPD-related and that attributable to comorbidities, and compared health-care utilization and costs across groups of patients with different drug regimens. Methods: Using a multistate managed-care-organization claims database, a cohort of 67,862 BPD patients were selected and followed for the length of their enrollment between January 1, 1998 and December 31, 2002. All costs associated with the patients' medical claims were adjusted to 2002 dollars using the medical component of Consumer Price Index. Patients were classified into three groups based on their drug regimen: atypical antipsychotics (ATYP), atypical antipsychotics plus mood stabilizers (ATYP + MS), and mood stabilizers only (MS). The Charlson comorbidity index was used to control for comorbid conditions. Using both Poisson and log-linear regression analyses, numbers of hospitalizations, emergency room (ER) visits, and outpatient visits, as well as treatment costs per enrolled month, were regressed on age, sex, medication regimen, and clinical comorbidities. Results: The mean charge and reimbursement per patient-year were $12,797 and $6581, respectively. Of the treatment cost, 33% was BPD-related, and 67% was attributed to comorbidities. Compared to patients in the MS treatment regimen, higher treatment costs were associated with ATYP (Rate Ratio = 1.24, 95% CI 1.17-1.31) and ATYP + MS (RR = 1.52, 1.47-1.56). Moreover, higher costs were associated with key comorbidities like personality disorder (RR = 1.45, 1.37-1.53). Patients on the ATYP regimen had higher risks of hospitalization (RR = 1.44, 1.33-1.56) and ER visits (RR = 1.15, 1.04-1.27), but lower risk of outpatient visits (RR = 0.81, 0.76-0.86). Conclusions: Controlling treatment costs for BPD patients requires focusing on patients with key comorbidities and monitoring the association between treatment regimen and resource use.