We are troubled by the conclusions of Laine and colleagues [1] that fluconazole is more effective than ketoconazole in treating candida esophagitis. Factors other than drug efficacy might explain the better response to fluconazole.

Drug resistance may have accounted for some of the difference in response. Nearly half of the patients in both groups had been treated with antifungal therapy during the previous 3 months (26% with ketoconazole; 0% with fluconazole). The reasons for previous antifungal treatment were not given, nor is it mentioned whether ketoconazole use was equally distributed between the two groups. When the analysis was restricted to patients who did not receive previous antifungal therapy, no significant difference was observed in the rates of clinical cure.

Perhaps the ketoconazole dosage (200 mg/d) in this study was inadequate. Although 35% of ketoconazole-treated patients were not clinically cured, only 21% had their dosage increased to 400 mg/d. We often start patients on the higher dosage for initial treatment of candida esophagitis. In addition, it would be important to know whether any of the 37% of patients not treated with zidovudine were taking didanosine (ddI), a drug that interferes with ketoconazole absorption. Before it was licensed, didanosine was widely available through an expanded access program and buyers' clubs.

We expect that fluconazole will be promoted by the manufacturer that sponsored this study as being the drug of choice for the treatment of candida esophagitis. Even if clinically superior, less expensive treatment is effective for many patients. The average wholesale price of a 3-week course of ketoconazole is $45.78 (200 mg/d) or $91.56 (400 mg/d). Routine use of fluconazole could also lead to the development of fluconazole-resistant candidal infections, now being seen among patients with AIDS at San Francisco General Hospital. Because candida esophagitis is rarely life-threatening, fluconazole could be reserved for patients who do not improve with ketoconazole therapy.