Pharmacoepidemiology of clozapine in 202 inpatients with schizophrenia

To evaluate clozapine in a field trial for hospitalized patients with treatment-resistant schizophrenia.

The setting consisted of a large, state-operated, public psychiatric system. The protocol called for the treating psychiatrist to provide symptom- and adverse-effect ratings at four times following the start of drug therapy. The outcome criteria included the Sandoz study outcome measure of symptom improvement as well as discharge status for one year of follow-up. To assess the validity of the ratings, several measures of internal consistency were determined. Clozapine therapy was started in 227 patients, and symptom data are available for 202.

Overall, 33 percent (n = 66) of the patients were improved at the end of one year of treatment; 12 percent (n = 24) maintained symptom improvement at all three evaluation times. Modest, statistically significant improvement after 12 weeks compared with baseline Brief Psychiatric Rating Scale (BPRS) total scores was observed for the patients continuing medication (n = 152); the emergence of a previously unimproved group (n = 26) explains this modest improvement. However, in the analysis of all patients (n = 202), (including dropouts), there was no significant symptom improvement after 12 weeks. Lower baseline BPRS scores predicted significant symptom improvement after 12 weeks of treatment. Among those medicated for one year, the pattern of symptom improvement showed that the probability of late improvement was 0.26 for those previously unimproved, and the probability of a 12-week responder losing improvement was 0.23, resulting in a net group gain of 3 cases in 100. By the end of one year, 8 percent (n = 17) of the cohort was discharged, and 3 percent (n = 7) was transferred to another facility while continuing to receive clozapine. Of the 227 original patients started on clozapine therapy, medication was discontinued for adverse effects in 11 percent (n = 25): white blood cell count (WBC) decrease (but no agranulocytosis) in 5 percent (n = 12), seizures in 1 percent (n = 3), one patient with seizures and decreased WBC count, and other events (e.g., cardiovascular changes, fever, or possible neuroleptic malignant syndrome) in 4 percent (n = 9). Patient refusal was reported for 6 percent (n = 13) of those starting treatment.

Although only 19 percent of the patients exhibited improvement at 6 weeks, the response rate at 12 weeks (29 percent) for this naturalistic study cohort was similar to that in the major, double-blind, six-week, controlled, clinical trial of clozapine. The impersistence of response as symptoms were followed for up to one year is a finding that deserves rigorous evaluation.