Drs. Singh and Yu wonder whether the increased prevalence of gastrointestinal symptoms and worse survival in patients randomly assigned to receive clarithromycin, 2000 mg twice daily, could reflect undiagnosed adrenal insufficiency. We do not believe that the gastrointestinal toxicity observed in the high-dose clarithromycin arm was a result of adrenal insufficiency. Patients were randomly assigned to one of three treatment arms, and we did not observe significant imbalances in patient characteristics by treatment. Gastrointestinal toxicity is a well-known, dose-related complication of macrolide therapy. Earlier studies of high-dose clarithromycin in patients with HIV infection confirmed the higher incidence of these effects in doses of 2000 mg or 3000 mg twice daily [1]. As noted in our article, we could not attribute the differences in mortality to any one factor.

Dr. Musher asks why a placebo group was not included in our study and what the effect of treatment with antipyretic agents would have been. These issues were discussed extensively within the AIDS Clinical Trials Group when studies for the treatment of M. avium complex infection were being planned. By the time our study was designed, investigators agreed that the use of placebos for patients with bacteremic M. avium complex infection was inappropriate because it was known that the disease was associated with increased mortality [2] and that drug treatment could reduce bacteremia and alleviate symptoms [3].

Symptomatic treatment alone of a disseminated bacterial infection known to shorten survival is unlikely to be embraced by patients or physicians. Indeed, a large proportion of patients in the trial were already receiving antipyretic agents when clarithromycin therapy was initiated. Our study was designed as a dose-ranging study to determine the bacteriologic activity of clarithromycin against M. avium complex infection rather than as a definitive trial of optimal therapy for this infection. The reductions in fever and symptoms that accompanied bacteriologic responses in our study support the use of antimicrobial treatment. We did do quality-of-life assessments in a subset of our patients, but data on these assessments were deleted from the paper during the peer review process. We are currently studying combination therapy for M. avium complex bacteremia in which two or three drugs are being used. We believe data from these studies will lead to better management of patients with this disease. We do not believe that placebo-controlled or antipyretic-controlled trials are warranted.

Drs. van der Meer and Danner suggest that 1000 mg of clarithromycin twice daily was the better treatment in our study because it was associated with faster and more durable sterilization of the blood and because our proportional hazards analysis did not find that this dose was significantly associated with mortality. We agree that 1000 mg of clarithromycin twice daily is more rapidly active than 500 mg twice daily; this dosage has been used effectively by other investigators [4]. In our study, however, the survival differences between dose groups were statistically significant after 12 weeks. Six percent of patients receiving 500 mg of clarithromycin twice daily and 29% of patients receiving 1000 mg twice daily died during the first 12 weeks of therapy (P = 0.01). Given that 500 mg of clarithromycin twice daily is microbiologically active and that combination therapy is now routinely used, we believe that this dose is a reasonable standard. In some circumstances, however, clinicians may choose to treat patients with a higher dose of the drug.