Thirty RCTs were included for review (n=6,303): two triple-blinded RCTs (n=931); five double-blinded RCTs (n=1,674); and 23 unblinded RCTs (n=3,698). Ten RCTs had adequate allocation concealment, 10 had adequate allocation generation and nine used intention-to-treat analyses.
Anti-fungals versus no treatment placebo: Empirical antifungal treatment did not significantly decrease all-cause mortality (RR 0.82, 95% CI 0.50 to 1.34; six RCTs, n=955) but did significantly decrease the risk of invasive fungal infections (RR 0.25, 95% CI 0.12 to 0.54; five RCTs n=800) compared to placebo, no treatment or pre-emptive treatment. There was no evidence of significant statistical heterogeneity for mortality. Sensitivity analyses according to study quality and time of commencing treatment did not significantly alter the findings. The number needed to treat to prevent one invasive fungal infection was 17 and the control event rate was 7.7%.
Comparisons between anti-fungal treatments: All-cause mortality was significantly lower with liposomal amphotericin B compared to other amphotericin B treatments (RR 1.57, 95% CI 1.10 to 2.23; four studies, n=1,299). One trial found significantly higher mortality with amphotericin B lipid complex compared to liposomal amphotericin B (RR 3.34, 95% CI 1.35 to 8.3). There were no significant differences between other types of amphotericin based formulations on all-cause mortality or fungal infections. There were no significant differences between azoles combined or separately and amphotericin B-based anti-fungal treatments or between caspofungin and liposomal amphotericin B in risk of all-cause mortality or invasive fungal infections.
Results for secondary outcomes were reported in the review.