To the Editor: The article by Reid and associates (1) in the August 1998 issue suggests that clozapine might have a protective effect against suicide for patients with schizophrenia, a finding of real significance. However, we were left with several questions that need clarification.
In reporting the calculation of the suicide rate, the authors stated that an average of 1,310 patients continuously received clozapine each year. When calculating this figure, did they subtract the portion of time patients for whom clozapine was prescribed were not actually taking the medication?
For a suicide to be included in the clozapine group, a patient had to have been compliant with clozapine treatment. However, membership in the other group was not dependent on similar treatment compliance, or on whether patients were receiving treatment at all. This discrepancy would appear to introduce a compliance and treatment bias that might be correlated with a lower risk of suicide (2). Such a bias appears to be supported by the observation that two of the patients in the comparison group who committed suicide had received clozapine in the past, and at least one of them had been terminated from clozapine treatment because of noncompliance.
Given the available information and using the formula provided by Johnson and Kotz (3), we calculated a 95 percent confidence interval for the clozapine group of .32 to 91.64 suicides per 100,000 patients per year. For the group of all patients with schizophrenia and schizoaffective disorder, we calculated a 95 percent confidence interval of 44.26 to 86.14 suicides per 100,000 patients per year. The confidence intervals overlap, making the rates statistically equivalent to each other (p>.05).
We have conducted a preliminary analysis comparing one- and two-year mortality rates following 1,300 episodes of care in Veterans Affairs hospitals between 1992 and 1995 during which patients were started on clozapine and 55,000 similar episodes. After adjusting for sociodemographic factors and proxy measures for clinical status, we found no difference in mortality between the two samples, although specific information about suicide was not available.
If our calculations using data from the Reid study are correct, prudence dictates that few conclusions can be drawn from the data as they are presented; a protective effect due to clozapine was not demonstrated. The study does, however, illustrate the need for more definitive studies.
Dr. Sernyak is with the psychiatry service at the Yale University School of Medicine in New Haven, Connecticut. Drs. Hoff and Rosenheck are associated with the Northeast Program Evaluation Center of the Veterans Affairs Connecticut Health Care System in West Haven, Connecticut, and the Yale School of Medicine.