Aqueous chlorhexidine versus povidone‐iodine for ocular antisepsis: A systematic review and meta‐analysis

Abstract This systematic review and meta‐analysis aimed to compare aqueous chlorhexidine (AqCHX) with povidone‐iodine (PI) for ocular antisepsis in preventing post‐procedural endophthalmitis (primary outcome) and to evaluate antimicrobial efficacy and patient‐reported pain (secondary outcomes). A systematic search of PubMed, Embase and Web of Science was conducted on 26 November 2024 and updated on 25 June 2025. Randomised controlled trials (RCTs) and non‐randomised studies (NRS) comparing AqCHX with PI for ocular antisepsis were included. Risk of bias was assessed using RoB 2 for RCTs and ROBINS‐I for NRS. Meta‐analyses were conducted using random‐effects models. Endophthalmitis and microbiological culture positivity were reported as odds ratios (ORs), binary pain outcomes as risk ratios (RRs) and continuous pain outcomes as standardised mean differences (Hedges' g), each with 95% confidence intervals (CIs). The certainty of evidence was assessed using GRADE. Eight studies were included, comprising four RCTs and four NRS. Across three large NRS, endophthalmitis rates were low in both groups (0.024% with AqCHX vs. 0.025% with PI), with no statistically significant difference between AqCHX and PI (OR 0.80; 95% CI 0.54 to 1.19; I 2  = 0%). For microbiological culture positivity, pooled results from three RCTs and one NRS showed no significant difference between AqCHX and PI (OR 1.06; 95% CI 0.61 to 1.83; I 2  = 0%). The risk of pain or pain‐related ocular symptoms was lower with AqCHX than with PI (RR 0.22; 95% CI 0.14 to 0.33; I 2  = 0%) and pain intensity also favoured AqCHX, although heterogeneity was substantial (Hedges' g  = −1.83; 95% CI −2.54 to −1.13; I 2  = 77.3%). Certainty of evidence was rated as moderate across all outcomes. AqCHX showed comparable effectiveness to PI for ocular antisepsis while being associated with better tolerability. However, conclusions regarding endophthalmitis are limited by the reliance on NRS and the very low event rate, leaving uncertainty about the magnitude of any difference between treatments. Further well‐designed, adequately powered studies are needed to clarify their comparative effectiveness and tolerability.