Because space constraints made it necessary to select the categories of toxicity to be reported in our article, hyperkalemia was not specifically reported. In response to Dr. Perazella's query, however, we specifically looked at the occurrence of hyperkalemia in patients receiving clindamycin-primaquine, dapsone-trimethoprim, or trimethoprim-sulfamethoxazole in the ACTG (AIDS Clinical Trials Group) 108 trial.

Table 1 summarizes all categories of hyperkalemia that occurred in study patients while they were actually receiving assigned therapy. No grade IV toxicities occurred. A single instance of grade III hyperkalemia occurred in a patient receiving dapsone-trimethoprim; this patient had the one case of hyperkalemia to be classified as dose-limiting. A comparison of the frequency of hyperkalemia of all grades shows a significant difference (P = 0.02, Kruskal-Wallis test).

We explored the occurrence of hyperkalemia by using several other methods. Comparison of mean potassium levels at baseline (P = 0.2), day 7 (P < 0.001), day 14 (P < 0.001), and day 21 (P = 0.1) shows significant differences only for days 7 and 14. Pairwise comparisons for these days show that the mean potassium level was significantly lower in the clindamycin-primaquine group than in the trimethoprim-sulfamethoxazole group on days 7 and 14 and that the mean potassium level was significantly lower in the dapsone-trimethoprim group than in the trimethoprim-sulfamethoxazole group on day 14.

In a separate analysis, we looked at the percentage of change from baseline in serum potassium level on days 7, 14, and 21. At day 7, the mean change was less in patients receiving clindamycin-primaquine than in the two groups receiving trimethoprim-containing regimens (P = 0.04). These differences, however, seemed to diminish with time, so that the mean change was lowest for the clindamycin-primaquine group, but not statistically significantly so, on days 14 and 21 (P = 0.2 and P > 0.2, respectively).

Our results confirm that hyperkalemia occurs more frequently in patients receiving trimethoprim-sulfamethoxazole or dapsone-trimethoprim than in those receiving clindamycin-primaquine. In our study, the daily dose of trimethoprim ranged from 12 to 20 mg/kg of body weight. In both groups, however, hyperkalemia appeared to be clinically significant only rarely. Although we do not have data on how many elevated serum levels were confirmed on repeated testing or on whether therapy was instituted for hyperkalemia, this side effect was thought to necessitate the substitution of alternate anti-P. carinii therapies in only one patient.