Implications for Future Research of “Medication-Free Research in Early Episode Schizophrenia”

As noted by Bola,¹ treatment with antipsychotics early in the course of a schizophrenic illness has implications both for clinical care and research. This commentary will focus on the implications for clinical trials with patients who are early in the course of their schizophrenic illness. But first, I want to comment on the 1967 article that had inadequate detail to allow formal inclusion in the meta-analysis.² Although the trial was originally designed as a “first-episode” study, it included a number of patients who were acutely symptomatic but not experiencing their first episode. Further, virtually all of the patients initially randomized to placebo received medication following their participation in the trial, and we observed that the length of index hospitalization for these patients was longer than for those initially randomized to receive medication. Although this group required more time prior to hospital discharge, they were less likely to relapse during the next year. It seems unlikely that this was “caused” by the initial medication-free period since all received antipsychotic drugs prior to discharge. We speculated that patients initially treated with placebo received an extra measure of psychosocial care before discharge that contributed to their lower likelihood of subsequent rehospitalization.

In reviewing the Bola article, I am struck by the fact that only 1 study—that by May and colleagues³—lasted longer than 6 weeks. And bearing that point in mind, it is interesting that the 1 study that involved treatment for 6 to 12 months had a significant and positive effect size (.14), indicating better outcome for those treated with medication. With the exception of the Wirt and Simon study,⁴ all the other effect sizes are negative, indicating better outcome for those treated without medication.

The direction of these effects seems to parallel the findings of naturalistic studies that have examined the relationship between duration of untreated psychosis (DUP) and outcome in first-episode schizophrenia. The retrospectively reported durations range from several days to years, but the average duration is almost a year. Two recent meta-analytic studies conclude that a shorter duration of psychosis before treatment with antipsychotic medication is associated with a better long-term outcome and that a longer time before treatment adversely affects the course of illness.^5,6 The findings that Bola reports are consistent with that hypothesis.

One implication that has been drawn from the DUP literature is that the sooner medication is initiated, the better—even to the point of intervening before there is firm evidence of a psychotic illness. A second implication is that delay of medication initiation in the context of a clinical trial would be the equivalent of increasing the duration of untreated psychosis. The studies reviewed by Bola suggest that brief increases are not harmful. Further, treatment without medication is not the same as not treating psychosis. In all reported clinical trials that involve a delay of antipsychotic medication initiation, the comparison treatment condition included substantial psychosocial interventions, either informal or structured.

Is there a role for future studies early in the course of schizophrenia that involve delay of antipsychotic medication initiation or medication-free periods? The data regarding short-term response to medication suggest that trials designed to investigate antipsychotic effect of medications that involve established mechanisms of action would have known outcomes—medication better than placebo—and therefore may not be appropriate for participants who are early in the course of their illness. Longer studies that involve medication-free periods that last more than 6 weeks do not seem appropriate, even if they incorporate detailed protocols for initiation of rescue medication, because of the difficulties in implementing rescue medications. On the other hand, studies of medications that involve novel mechanisms of action may be particularly appropriate for study in this patient population because extensive prior treatment with more conventional medications might blunt the effects of these novel agents. In such cases, study designs that involve short delays or interruptions of treatment with medication could be highly desirable. Medication-naïve patients or those who have experienced relatively brief exposure to medication represent a unique population in which to study such medications because effects may not be detectable in chronically treated patients. Of course, such protocols must incorporate a high level of clinical care and specific stopping rules.⁷

Bola has rekindled attention to an important question in the early treatment of schizophrenia. This re-evaluation of an old literature suggests options for future research designs. This may be necessary if we are to substantially advance the treatment options for a patient population that sorely needs novel interventions.