Drs. Stroup, Rosenheck, Essock, and Lieberman Reply

To The Editor: We agree with Drs. Leucht, Heres, Hamann, and Kissling that pretrial medication should be considered when designing, reporting, and interpreting randomized controlled trials of antipsychotic medications. In some cases, stratifying by pretrial medication may be needed, as exemplified in the Comparison of Antipsychotics for Metabolic Problems study that is currently on the National Institute of Mental Health (NIMH) Schizophrenia Trials Network and examines the effectiveness of switching medications to address weight and metabolic problems (1) .

As Dr. Leucht et al. suggest, enrollment criteria that exclude those who have failed to respond to one study drug but not to others could bias the results of a clinical trial. Their proposal to reanalyze, where possible, earlier studies comparing haloperidol to newer antipsychotics could determine whether patients with previous poor responses to haloperidol fared worse while receiving haloperidol than those patients without such prior exposure.

Dr. Leucht et al. paraphrased us as stating “that most patients in CATIE were only moderately ill and relatively stable at baseline.” However, it was reported that “these individuals varied widely in clinical severity.” This range led us to examine the impact of severity, and we reported no significant interaction between baseline Positive and Negative Syndrome Scale scores and the “stay versus switch” variable. Therefore, although we agree that switching the medications of stable patients may risk destabilization, our reanalysis of CATIE did not provide evidence to support their assertion concerning the interaction between switching medications and clinical status.

Whether first-generation antipsychotics, such as haloperidol or perphenazine, are more effective than second-generation antipsychotics in individuals who have failed trials of newer drugs is an empirical question, as is the extent to which response to previous medications predicts response to subsequent medications. Unfortunately, the conventional wisdom of 2000 led us not to include a first-generation antipsychotic in phase 2 of the CATIE schizophrenia trial (2) . However, a recent report by Kane et al., which compared aripiprazole and perphenazine in individuals with schizophrenia who had a history of “antipsychotic resistance” and a poor response to a prospective trial of olanzapine or risperidone, supported the proposition that those who discontinue one of the newer drugs may benefit from a trial of a first-generation antipsychotic (3) .