To the Editor: We read with great interest the article by Gary S. Sachs, M.D., and his colleagues (1). They compared the efficacy of a mood stabilizer in combination with placebo, haloperidol, or risperidone in the treatment of acute mania. The authors concluded that risperidone and haloperidol were equally efficacious as adjuncts and more efficacious than a mood stabilizer alone. The validity of this study can be criticized on the grounds of use of a placebo arm, the selection of patients, and improper random assignment and blinding.

The use of a placebo arm remains a highly controversial issue from an ethical point of view and has been addressed in the Declaration of Helsinki. It states that the patients in a study should "be assured of the best proven diagnostic and therapeutic method," even in the control group (2). This clearly implies that a placebo should not be used as a control when superior existing treatment is available. It can be argued that existing studies of combination therapy in acute mania, several published treatment guidelines, and the widespread accepted use of combination treatment among our colleagues support the opinion that a mood stabilizer with an antipsychotic is the best treatment available to date. We, therefore, suggest that it might have been more relevant to omit the placebo group and compare the efficacy of a mood stabilizer in combination with a variety of different antipsychotics. After all, as Hill pointed out in 1963 (3), the key point is how a new treatment compares with existing treatment rather than whether it is better than nothing.

The clinical relevance of this study can be criticized on the basis of the selection of patients. The issue of selection of patients for randomized controlled trials is controversial and has been addressed in the Consolidated Standards of Reporting Trials (CONSORT) guidelines (4). This state that all patients assessed for a trial must be accounted for and recommends the inclusion of a diagram that explains the outcome for every patient involved in the trial (4). In the study in question, the information about recruitment of the subjects is lacking. We do not know how many subjects were initially assessed, how many subjects were excluded, and the reasons for exclusion. The information regarding the participation rate and the response rate has implications for generalizability and future research.

Ensuring that random assignment of patients to intervention groups is achieved is vital in eliminating the introduction of allocation and confounding bias to a trial. The authors did not describe the method of random assignment or concealment, although the CONSORT guidelines states that this information should be clearly reported. It has been reported that studies that use poor or unclear concealment compared with those that conceal appropriately may give a higher estimate of treatment effect (4). In the absence of adequate information on random assignment, this raises the possibility of allocation bias in the present trial.

Haloperidol causes marked extrapyramidal side effects, which can easily render investigators unblind. We question, therefore, whether the investigators really were blind to treatment. It has been found that trials that are not double blind yield larger estimates of treatment effect than trials in which the authors specify double blinding (5).