DO WHITE-CELL COUNT SPIKES PREDICT AGRANULOCYTOSIS IN CLOZAPINE RECIPIENTS

The use of clozapine is limited by the risk of agranulocytosis. The in cidence of agranulocytosis after 1 year was .80 percent in 11,555 pati ents registered in the Clozaril Patient Management System (CPMS) who r eceived clozapine from February 1990 to April 1991. We noticed a tende ncy for white-cell counts to spike upward before agranulocytosis occur red. We analyzed the CPMS data to test whether a white-cell count spik e at least 15 percent above the previous measurement predicted agranul ocytosis within 75 days, with a 21-day lag to allow white-cell counts to decline to levels indicative of agranulocytosis. The occurrence of a spike, entered as a time-dependent covariate in proportional hazards regression, significantly predicted development of agranulocytosis (r isk ratio, 3.02; 95% confidence interval, 1.38 to 6.57). Spikes were s ensitive though nonspecific predictors, occurring frequently in patien ts who did not develop agranulocytosis. These results, while explorato ry, indicate the potential usefulness of these spikes as guidelines to govern the use of clozapine.