Purpose: Recent results from the large NIH-funded CATIE study suggest that the first-generation antipsychotic perphenazine is more cost-effective than newer atypical antipsychotics as a first-line treatment for schizophrenia. Despite limitations, these results have influenced the selection of antipsychotic agents in the United States. Moreover, some have argued that the CATIE results are sufficiently conclusive that no further comparative effectiveness studies of these drugs are indicated. We studied the value of performing additional studies of the costs and effectiveness of treatment with typical versus atypical antipsychotics that could provide greater precision than the estimates in CATIE.

Methods: We used CATIE data to perform a probabilistic value of information analysis based on the uncertainty in the lifetime survival, costs, and QALYs resulting from initial schizophrenia treatment with perphenazine, olanzapine, risperidone, and quetiapine. The expected value of research calculations were done under alternative assumptions about whether typical or atypical antipsychotics would be used in the absence of additional research. The value of research depends on the probability that the decision made in the absence of research would be incorrect and the expected net social loss associated with that incorrect decision.

Results: Using a conservative threshold of $50K/QALY and assuming that the CATIE results concerning the mean outcomes across treatment arms are correct so that the typical antipsychotics are the cost-effective first-line treatment in this population, the expected value of more precisely determining the cost-effectiveness of typical versus atypical antipsychotics in the US is $268 billion. This includes $155 billion accruing to the prevalent cohort of schizophrenia patients and $5.5 billion for each of the next 20 incident annual cohorts. The probability that this decision will be incorrect is estimated to be 47%. On the other hand, if we reject this implication of the CATIE study and assume that atypical antipsychotic are cost-effective, then the expected value of a more precise comparative cost-effectiveness study is estimated to be $961 billion, with the chance that this decision will be incorrect being 53%.

Conclusions: Irrespective of whether typical or atypical antipsychotic medications are used as a first line treatment, the expected value of more precisely establishing the cost-effectiveness of typical compared to atypical antipsychotics is enormous. Further studies of the comparative cost-effectiveness of typical versus atypical antipsychotics are needed.