If the arrival of chemotherapy can be said to have contributed to the closure of tuberculosis sanatoria in the second half of the 20th century then antipsychotic drugs, beginning with chlorpromazine, significantly reduced the need for long-term inpatient care for patients with schizophrenia. For both illnesses compliance with prescribed drugs is essential but then the analogy breaks down because chronic treatment rather than achievement of cure is still common for schizophrenia so that adverse reactions to therapy and costs become even more important. Competitors to chlorpromazine soon arrived and by the turn of the century the Merck Manual (17th edition, 1999) was listing 16 other drugs aimed, like chlorpromazine, directly at dopamine type 2 receptors or with modified targets. When the U.S. National Institute of Mental Health’s Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) began in January, 2001, psychiatrists and patients faced difficult choices. So did the triallists because they had to decide not just which of the many drugs available to compare but also what measure of effectiveness to use.

CATIE (leaping to prominence at #2 in the current Top Ten ) compared one older drug (perphenazine) with four second-generation agents (olanzapine, quetiapine, risperidone, and ziprasidone). The main outcome measure was discontinuation of treatment, a criterion which may sound rather pessimistic but does reflect reality; it also combines physicians’ and patients’ perspectives on the drug’s effectiveness and tolerability. Most patients had discontinued treatment before the initially planned 18 months were up, there being a clear advantage for olanzapine (64% discontinuing compared with 74-82% for the other four drugs).

The difficulty of balancing efficacy and safety is illustrated by the analyses of withdrawal rates for lack of efficacy compared with withdrawals due to intolerability. For olanzapine, drug intolerance was more likely reason for stopping treatment than ineffectiveness; for the other four agents the reverse was true. The main worry with olanzapine was weight gain (and related metabolic changes)—indeed, this has been a concern with the second-generation drugs generally. The patients in the olanzapine group put on an average of two pounds in weight per month of treatment, and blood glucose, cholesterol, and triglyceride concentrations rose significantly. In 30% of patients there was a weight gain of more than 7%; the frequency of a 7% weight gain was 7-16% for patients on the other drugs.

A 2003 meta-analysis found that four second-generation agents were significantly more effective than first-generation drugs but six others were not (J.M. Davis, et al., Arch. Gen. Psychiatry, 60[6]: 553-64, 2003). If all second-generation agents are not the same, evidence-based decisions will require a large number of clinical studies, and many have indeed been completed—for example, the 2003 meta-analysis included 142 trials. CATIE is a large study of five drugs in almost 1,500 patients. Even though one effective second-generation drug was not included in the first phase (clozapine’s side-effects, the blood disorder agranulocytosis, limit prescribing) and the first-generation comparator was perphenazine rather than haloperidol, CATIE is clearly a significant clinical contribution to the long-running first/second generation argument. The CATIE Study Investigators have looked at the trial’s economic contribution too, but separately and with ratings adjusted for quality of life (R.A. Rosenheck, et al., Am. J. Psychiatry, 163[12]: 2080-9, 2006). On this basis, "there were no significant differences between perphenazine and any second-generation medication" but monthly healthcare costs were 20-30% less on average ($300-600) for the phenazine group than for the others. These two CATIE papers certainly inform decision taking but may not make life easy for policy makers.