Comparative efficacy and safety of antiviral therapy for herpes zoster ophthalmicus: a systematic review and Bayesian network meta-analysis

Abstract Background Herpes zoster ophthalmicus (HZO) is reactivation of varicella zoster virus in the ophthalmic division of the trigeminal nerve. About half of untreated patients develop ocular complications. Acyclovir (oral, intravenous or topical), valacyclovir and famciclovir are all used, but no trial has compared more than two regimens. We compared these drugs and routes for acute HZO. Methods We searched PubMed, Embase and CENTRAL from inception to 15 March 2026 for randomised trials of antiviral treatment for acute HZO. Trials of herpes simplex eye disease and of long-term suppression were excluded. Systemic and topical acyclovir were analysed as separate treatments. Outcomes were treatment failure, resolution of ocular disease and adverse events. We fitted Bayesian random-effects network meta-analyses, which give odds ratios (ORs) with 95% credible intervals (CrIs). Risk of bias was assessed with RoB 2 and certainty with GRADE. Results Seven trials (933 participants, published 1983–2001) were included. Compared with placebo, systemic acyclovir reduced treatment failure (OR 0.19, 95% CrI 0.05–0.66; low certainty) and topical acyclovir did not (OR 1.00, 0.19–5.09; very low certainty). Topical acyclovir had higher odds of failure than systemic acyclovir (OR 5.15, 1.24–25.53; low certainty). Neither valacyclovir (OR 0.90, 0.19–4.99) nor famciclovir (OR 1.00, 0.24–4.87) differed from systemic acyclovir (low certainty). Systemic acyclovir also increased resolution of ocular disease (OR 8.50, 1.07–75.26; low certainty). Adverse events did not differ between treatments. No inconsistency was detected. Three trials were at high risk of bias and none at low risk. Conclusions Systemic antiviral therapy reduces ocular complications in acute HZO, and topical acyclovir ointment is not an adequate substitute. Valacyclovir and famciclovir perform similarly to oral acyclovir. The evidence is limited to seven small, older trials, so adequately powered head-to-head trials are needed. PROSPERO: CRD420261423206.