Design: Cochrane systematic review of RCTs and quasi-RCTs
Population: 229 newborns with suspected or confirmed invasive fungal infection in 6 trials
Primary outcome: Death before hospital discharge
Effect: Caspofungin vs amphotericin B RR 0.43 (95% CI 0.13–1.48); all comparisons very low certainty

This Cochrane update found six randomised trials, with only 229 newborns, comparing antifungals for suspected or confirmed invasive fungal infection: amphotericin B against fluconazole, micafungin or caspofungin, and micafungin against fluconazole. Most were in high- and middle-income countries.

For every comparison, the evidence on death before discharge was very uncertain: fluconazole vs amphotericin B RR 0.73 (0.26–2.05), micafungin vs amphotericin B RR 1.20 (0.29–4.90), caspofungin vs amphotericin B RR 0.43 (0.13–1.48). No trial reported neurodevelopmental outcomes.

In practice, drug choice will continue to be guided by local fungal epidemiology, resistance patterns — Candida auris and fluconazole-resistant species are a growing concern in Indian neonatal units — renal function and availability, rather than by comparative trial evidence.

Choose neonatal antifungal therapy by local species and resistance data, since trials cannot separate the drugs.
Send cultures before starting antifungals and adjust once the species and susceptibility are known.
Consider amphotericin B deoxycholate as a reasonable first choice where resistance to azoles is common.
Look for and remove infected central lines; drug choice matters less than source control.
Consider lumbar puncture and eye and renal imaging for disseminated candidiasis.

Why it matters

It confirms that neonatal antifungal choice is currently a stewardship decision, not an evidence-ranked one.

The statistics, in plain English

All the confidence intervals are extremely wide — for example, 0.29 to 4.90 — meaning the true effect could be a large benefit or a large harm. 'Very low certainty' means the true effect is likely to be substantially different from the estimate.