Cost-Effectiveness of Schizophrenia Pharmacotherapy

To The Editor: The study by Robert A. Rosenheck, M.D.  and colleagues represents a large and ambitious attempt to study the cost-effectiveness of pharmacotherapy for the treatment of schizophrenia (1) . The authors acknowledge the caveats for policy making. However, we are concerned that drug plan administrators will nonetheless use these results to promote prior authorization and step-therapy policies that restrict access to atypical antipsychotic medications. Such policies will be based on both the imperative to contain costs and the conclusions of this analysis, which are unwarranted based on limitations in the study design.

The purpose of cost-effectiveness analysis is to compare the relative efficacy of interventions in order to aid decision making (2) . Thus, the study will be considered a commentary on policy options. However, the study deviates from the relevant policy choices in important ways. First, the trial was not limited to new episodes of schizophrenia. However, such patients are generally the only ones affected by Medicaid and Medicare policies, which largely grandfather currently treated patients. Second, it bases comparisons between typical and atypical medications on a select population—those both willing to switch antipsychotics and without tardive dyskinesia. The generalizability of this group is questionable, since those with tardive dyskinesia may benefit more from newer medications (3) .

Further, it is telling that patients and their clinicians almost universally switched to another atypical antipsychotic agent, suggesting customization of individual treatment. To this end, the unique circumstances of a clinical trial limit generalizability. While trial participants were free to switch medications at any time, typical prior authorization policies are burdensome on provider time and have unknown impacts on medication adherence and clinical outcomes. Inclusion of these impacts and use of a societal perspective in the analysis might reveal quite different implications.

The authors acknowledge that the results should not be used to limit treatment of patients to first-generation antipsychotics. However, their abstract—which is all many busy policymakers will ever read—states that perphenazine treatment “was less costly than treatment with second-generation antipsychotics with no significant differences in measures of effectiveness” (p. 2080) with no such caveat. Such statements will be all too alluring for pharmacy benefit managers struggling with costs. In fact, within 2 weeks of publication, we received requests for advice about proposed policies to provide typical agents without cost-sharing while charging high copayments for atypical medications in a large drug benefit plan. Consequently, we have serious concerns that this research will support potentially harmful access restrictions on effective medications for vulnerable patients.