Dr. Keefe and Colleagues Reply

To The Editor: Thank you for the opportunity to clarify our recent article published in the Journal . We are grateful that interest in cognitive problems associated with schizophrenia, and how to treat them, has attracted broad attention in the field of schizophrenia research, including from experts in phenomenology and psychopharmacology such as Drs. Carpenter and Conley.

We are happy to report that we agree with Drs. Carpenter and Conley on almost all of the points they have made, although we were somewhat perplexed about the intention of their letter, since we made these points in our article. In fact, of the seven “causal explanations” they raise for the cognitive improvement found with the treatments that we tested, we discussed six in detail. The sole exception was their statement that industry-sponsored studies tend to report more favorable effects of drug treatment. We declined to discuss this important issue in our article because none of the effects we reported were particularly favorable. We repeatedly referred to treatment effects as “modest” and stated that “the amount of cognitive change we report here is consistent with what may be expected from practice effects and placebo effects” (p. 1069). In fact, we discussed in this article and other recent publications (1) that the small improvements found with atypical antipsychotics cannot be distinguished from practice or placebo effects (2) and atypical antipsychotics may even impair some cognitive functions (3) . For all these reasons, we certainly do not feel that we have overstated the positive cognitive benefit of atypical medications.

Drs. Carpenter and Conley criticize the design of our study because it only included three antipsychotic treatments and did not include control conditions such as a typical antipsychotic or placebo treatment group. We did not feel the necessity to repeat previous first-episode studies that compared risperidone and olanzapine with haloperidol (4 , 5) . Our primary aim was to compare the cognitive efficacy of atypical drugs in first-episode patients, which had not been done previously. The inclusion of a fourth treatment with a typical antipsychotic would have required a reduction in the sample size for the other treatments and thus a loss of statistical power. In addition, our perspective at the time the study was designed (prior to the Clinical Antipsychotic Trials of Intervention Effectiveness results [6] ) was that treatment with a typical antipsychotic raised ethical questions and was not relevant because of the tiny percentage of first-episode patients who in practice receive typical antipsychotics. The suggestion of treating first-episode patients with placebo in an antipsychotic trial continues to be ethically objectionable to us.

Drs. Carpenter and Conley also raise the question of whether the effect had “clinical importance.” First, their letter cites the weak relationship between cognitive improvement and changes in quality of life at 12 weeks. As stated in our article, we feel that it is unwise to emphasize these correlations because it is unlikely that any cognitive benefit or worsening will precipitate changes in functional outcomes in such a short period of time. Surprisingly, Drs. Carpenter and Conley ignore our data from the truer test of the relationship between cognitive change and functional change, which was after 52 weeks of treatment. These correlations were between 0.22 and 0.36, within the range of a medium effect size (7) , which is the traditional criterion for clinical importance (7) . Thus, we stand by our conclusion that these cognitive changes “may be clinically relevant” (p. 1068). However, as we stated in our article, “interpretation of this relationship is tempered by analyses indicating that symptom change and baseline cognitive scores also predicted the variance in functional outcomes. Thus, cognitive improvement may be a part of a general treatment response that is associated with improved functional outcomes” (p. 1068).

In summary, we do not see substantial differences between the concerns raised by Drs. Carpenter and Conley and our views about the study design and modest treatment effect. We certainly agree with their major point, which we tried to make clear in our article, in that our ability to improve cognition in schizophrenia remains very limited and we cannot, as a field, accept the modest improvement in cognitive test performance that atypical medications provide as satisfactory. We need to redouble our efforts to discover and evaluate new treatment options to improve cognition and functional outcomes in schizophrenia patients. This will be the best way, and probably the only convincing way, to move beyond worries about whether the modest cognitive changes of the level we reported are powerful enough to have “clinical relevance” and are more than what might be attributed to practice and placebo effects.