Sir.—Sulfamethoxazole-trimethoprim has gained wide acceptance in the outpatient treatment of otitis media and urinary tract infections in children, as ampicillin-resistant strains of Haemophilus influenzae and Escherichia coli have become more prevalent.1 Asmar et al reported a prospective study in which neutropenia and thrombocytopenia were found to occur more frequently in children given sulfamethoxazole-trimethoprim than in those given amoxicillin (Journal 1981;135:1100-1103). We have two comments regarding this work.
The P values stated in the article seem to be incorrect. Application of the standard χ2 test for a 2 × 2 contingency table2(p13) to the data in Table 2 of Asmar et al yielded a two-tailed P = .012 for neutropenia and a twotailed P = .105 for thrombocytopenia. One-tailed P values would be half as large. It is worth nothing that Fishers' exact test would be more appropriate for the thrombocytopenia data, because one of the cell sizes is zero. The