The 177 participants received induction therapy with 2 weeks of amphotericin B and high-dose fluconazole, followed by consolidation therapy with fluconazole. After 7 to 11 days of antifungal treatment, they were randomized to receive early ART (initiated ≤48 hours after randomization) or deferred ART (initiated 4 weeks after randomization). The ART regimens were predominantly AZT/3TC/efavirenz (80%) or d4T/3TC/efavirenz (19%).

The patients were randomized a median of 8 days after initiation of amphotericin B. The median time between diagnosis of cryptococcal meningitis and ART initiation was 9 days for the early-ART group and 36 days in the deferred-ART group. Mortality at 26 weeks — the primary study endpoint — was significantly higher in the early-ART group than in the deferred-ART group (45% vs. 30%; hazard ratio, 1.73; 95% confidence interval, 1.06–2.82). The between-group difference in mortality occurred early in therapy (2–5 weeks after diagnosis). Participants with low cerebrospinal fluid (CSF) white-cell counts (<5 cells/mm³) at randomization had particularly high mortality with early ART. No other participant characteristic, including a low CD4-cell count, affected between-group mortality differences. The timing of ART did not influence any of the secondary endpoints (survival at 46 weeks, cryptococcal immune reconstitution inflammatory syndrome, relapse of cryptococcal meningitis, fungal clearance, virologic suppression at 26 weeks, ART discontinuation for >3 days, or adverse events).