Face Masking and Risk of Post-Intravitreal Injection Endophthalmitis: A Network Meta-Analysis of 2.6 Million Injections.
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
TOPIC: To compare face masking protocols for post-intravitreal injection endophthalmitis (PIE) prophylaxis. CLINICAL RELEVANCE: Though mask mandates are lifted, ophthalmologists may question whether continued investment into face masks will influence their PIE rate. METHODS/LITERATURE REVIEWED: We included comparative studies of PIE incidence by masking policy (i.e. standard care [no restrictions], no-talking, physician masking, or universal masking [patient and physician]). A frequentist network meta-analysis (Mantel-Haenszel method with fixed-effects) synthesized direct and indirect evidence. Subgroup analysis excluded studies that systematically introduced new prophylactic techniques (e.g., prefilled syringes) during the observation period. The ROBINS-I and GRADE tools evaluated risk of bias and evidence certainty. RESULTS: We analyzed 17 studies (2,595,219 injections, 830 events; 0.032%). For the any PIE outcome (17 studies; 2,595,219 injections), compared with standard care, PIE incidence was significantly lower with no-talking (OR: 0.56, 95%CI [0.39, 0.82], I2: 0%) and physician masking (OR: 0.72, 95%CI [0.53, 0.99], I2: 0%) policies, which remained consistent in the subgroup analysis. Although PIE rates between standard care and universal masking did not differ in the main analysis (OR: 0.83, 95%CI [0.67, 1.02]), subgroup analysis revealed a significantly lower rate of any PIE with universal masking (OR: 0.70, 95%CI [0.55, 0.91], I2: 0%) compared to standard care. For the culture-positive (14 studies; 2,347,419 injections), Streptococcus (10 studies; 1,966,903 injections), and culture-negative (15 studies; 2,213,322 injections) outcomes, PIE rates between pairs of interventions groups generally did not reach significance, likely involving limited study power. As one exception, the incidence of culture-positive PIE was significantly lower with a no-talking policy (OR: 0.45, 95%CI [0.23, 0.92], single direct estimate) compared to standard care, though this result did not persist in subgroup analysis. As well, in subgroup analysis, universal masking had a significantly lower incidence of culture-negative PIE than standard care (OR: 0.68, 95%CI [0.47, 0.98], I2: 0%). CONCLUSION: For all outcomes, by GRADE analysis, low- or very-low certainty evidence suggests that no-talking and physician masking policies may reduce culture-positive or clinical PIE rates, respectively, compared to standard care and universal masking. While data were only available for endophthalmitis, the overall comparative safety of these interventions remains unclear.
