In Reply.—The P values stated in our article (Table 2) were based on one-tailed tests. As one of the cell sizes was zero in the thrombocytopenia comparison groups, we agree that Fisher's exact test would have been more appropriate. Obviously, more numbers are needed to arrive at meaningful statistical data. Neutropenia remains significantly more common in the group receiving sulfamethoxazole-trimethoprim, regardless of the method of analysis.
In our study, no attempt was made to elucidate the mechanism of neutropenia. The speculations of Drs Guess and Cohen may well indicate the operative mechanisms.
We agree that additional research on the incidence, severity, and mechanisms of hematologic reactions secondary to administration of sulfamethoxazole-trimethoprim is needed. Until more data are collected, however, we will continue to believe that it is prudent to monitor for potential abnormalities.