Timing of Povidone–Iodine Application to Reduce the Risk of Endophthalmitis after Intravitreal Injections

Purpose To analyze comparatively the effect of different intravitreal injection (IVI) protocols on the incidence of endophthalmitis occurring after injection. Design Retrospective case-control series. Participants Twenty-seven retina specialists in a large vitreoretinal practice performed 37 646 IVIs. Methods Multivariate analysis was used to identify risk factors for development of endophthalmitis occurring after injection. Before all injections, a technician applied 5% povidone–iodine (PI) to the eyelids and conjunctiva. There were 4 distinct aseptic protocols with regard to reapplication of PI by physicians: physicians who did not reapply PI, reapplication of PI without the use of a lid speculum, reapplication of PI before speculum placement, and reapplication of PI after speculum placement. Other analyzed variables included the use of gloves, a caliper to mark the injection site, and the class of medication (steroid vs. anti–vascular endothelial growth factor). Main Outcome Measures Cases of presumed infectious endophthalmitis. Results Thirty-three cases of presumed infectious endophthalmitis occurred after 37 646 injections (0.088%). The method of PI application was found to be a statistically significant predictor of the incidence of endophthalmitis ( P  = 0.031). When compared with the incidence of endophthalmitis for physicians who did not reapply PI (0.124% [20/16 155]), there was no statistical difference for reapplication of PI without the use of a speculum (0.110% [6/5472]; P  = 0.584) or reapplication before speculum insertion (0.122% [5/4067]; P  = 0.863). However, reapplication of PI after insertion of the lid speculum was associated with a significantly decreased incidence of endophthalmitis (0.017% [2/11 952]; P  = 0.004; odds ratio, 0.113). Use of gloves ( P  = 0.119) or a caliper to mark the injection site ( P  = 0.496) and the class of medication ( P  = 0.740) were not found to be statistically significant risk factors for endophthalmitis development. Conclusions The application of PI after placement of the lid speculum reduced the incidence of endophthalmitis occurring after injection approximately 7-fold compared with other aseptic protocols. Preventing the eyelid from contacting the injection site after the final application of PI is an important step in improving the safety of intravitreal injections.