In Reply: Dr. Rosenheck and his associates raise several important issues regarding our study, including how the lack of a control group might have influenced our findings. The presence of a control group might have shown that clozapine treatment had a much smaller impact on use of hospital resources than we found, but one set of data suggests otherwise: in any given year of our four-year study, the hospital utilization of the preclozapine group was higher than that of the clozapine group. However, because the study had no control group, the question remains whether an alternate treatment group would also have reduced hospital use.
In a 1993 report Rosenheck and associates (1) observed that patients hospitalized with schizophrenia who were followed prospectively decreased their use of hospital resources from baseline assessment, regardless of the specific treatment they received. This observation is consistent with the results of the controlled trial of clozapine versus haloperidol in VA patients with refractory schizophrenia mentioned in their letter. However, it is unclear that such a finding would pertain to the patient population we studied—outpatients who were evidencing sufficient clinical instability to warrant starting a new antipsychotic agent. In the absence of effective intervention, such patients might tend to show increased hospital use with follow-up.
Our study highlighted the point that cost evaluations cannot be generalized across utilization groups, yet Rosenheck and associates chose to compare our increased cost of $2,363 per patient during the first year of outpatient-initiated clozapine treatment with total inpatient and outpatient expenditures of just over $3,000 per psychiatric patient in the VA system. Our cost represents a 13.6 percent increase for patients with severe and persistent psychotic disorders who had a significant number of past hospitalizations. The VA cost undoubtedly includes a large proportion of clozapine-ineligible and even nonpsychotic patients. We think a more apt comparison with our study sample would be the group from the VA multisite clozapine study.
Although we did not find demonstrable savings associated with clozapine treatment during the first year, the net increase in total costs was small given the overall use of medical resources by patients in our sample. And as we noted, some data suggest that cost savings increase over time after initiation of treatment. Aside from a study by Meltzer and associates (2), in which treatment was initiated in the hospital, the best example of such data comes from the only other cost-effectiveness study of clozapine treatment initiated on an outpatient basis, by Aitchison and Kerwin (3). They found that over a three-year period after initiation of outpatient clozapine treatment, total costs decreased by 10.3 percent. A cost-effectiveness analysis that included clinical outcome measures confirmed that treatment was significantly more cost-effective than during the preclozapine period.