Paromomycin in Cryptosporidiosis

Sir—Hewitt and colleagues should be commended for publishing the findings of their placebo-controlled trial of paromomycin for the treatment of cryptosporidiosis [1]. Their results are similar to those noted in a previous placebo-controlled trial and in other open-label treatment studies [2, 3]. However, several weaknesses in the trial design and analysis may have led the authors to incorrect conclusions. The planned analysis was intent-totreat, but the data that were presented excluded information on individuals who dropped out during the study. Importantly, patients who dropped out were all in the placebo arm of the study, and all dropped out because of poor clinical response. Therefore, these patients who dropped out should have been included in the analysis as having failed treatment. Including patients who dropped out or died as having failed treatment, and defining “response to treatment” as either partial or complete response, we recalculated the results of the study as follows: at week 3, 8 (84%) of 17 patients in the paromomycin arm had responded to treatment, and 5 (28%) of 18 patients in the placebo arm had responded (P p ; x 2 test); at week 6, 10 (59%) of 17 .23 patients in the paromomycin arm had responded to treatment, and 8 (44%) of 18 patients in the placebo arm (who received placebo during the first 3-week period and paromomycin during the second 3-week period) had responded (; x 2 test). P p .40