To the Editor: The study by Luchins and associates (1) in the August 1998 issue addresses an important question: are the high costs of clozapine pharmacotherapy offset by reduced hospital utilization by outpatients?
The authors show that although clozapine treatment resulted in increased medication costs of more than $6,000 per year, reductions in hospital use offset some of these costs, so that the net increase associated with clozapine was reduced to $2,300. In the Department of Veterans Affairs (VA), to take one example, the average total inpatient and outpatient expenditure per psychiatric patient per year is just over $3,000. Increasing total health system costs by more than $2,000 a year is thus not inconsequential. From this perspective, a more appropriate conclusion to the study might be that, in contrast to the savings generated for high-cost inpatients, clozapine increases total costs when used with outpatients.
In addition, although the authors acknowledge that their study lacks a control group, our review of the literature suggests that they may have underestimated the impact of their uncontrolled design on their findings. They correctly note that like uncontrolled studies (2,3,4), two experimental studies—the Connecticut State Hospital study (5) and our multisite VA trial of high hospital users (30 to 364 days a year)(6)—have all found significant reductions in hospital costs when clozapine is used with high-cost patients. However, the effects of clozapine in the controlled trials (5,6) are much smaller than those suggested by the uncontrolled trials. While uncontrolled trials reported savings of up to $50,000 per patient per year, the VA multisite trial, the only experimental cost-effectiveness study published thus far, showed nonsignificant savings of only $2,700 (5 percent) attributable to clozapine.
The differences in findings between the uncontrolled trials and the experimental studies are primarily attributable to the greater internal validity of the experimental studies. Over the 12-month course of the VA study, for example, clozapine patients showed a 72 percent reduction in hospital utilization, from 29 days per month early in the trial to only seven days per month at the end. However, the control patients also showed a 64 percent decrease in hospital use over same time period. In the absence of the control group, the effects of clozapine would have been substantially overestimated. Thus a controlled trial of outpatients would be likely to show an even greater increase in costs than reported by Luchins and associates.
Although the VA study included only high hospital users, subgroup analysis showed that costs savings attributable to clozapine declined to less than $700 a year among the one-third of high users with the lowest hospital use, an average of 58 days in the year before the study. Thus hospital use for this sample was almost double the 27 annual hospital days for the sample in the Luchins study.
Although data from our controlled trial, as well as from the Luchins study, suggest that clozapine is likely to be considerably more costly than conventional antipsychotic medications when used with outpatients, clozapine may yet prove to be cost-effective for treatment-refractory outpatients if it demonstrates substantially greater clinical effectiveness than conventional medications for such patients. Luchins and his colleagues have made a major contribution by so clearly demonstrating that cost evaluations cannot be generalized across utilization groups.
Dr. Rosenheck is director of the Northeast Program Evaluation Center of the Veterans Affairs Connecticut Health Care System and clinical professor of psychiatry at Yale Medical School in New Haven, Connecticut, where Dr. Charney is professor of psychiatry. Ms. Cramer is associated with the health services research program of the VA Connecticut Health Care System and is a lecturer in psychiatry at Yale.