Stem MS, Rao P, Lee IJ, Woodward MA, Faia LJ, Wolfe JA, Capone A Jr, Covert D, Dass AB, Drenser KA, Garretson BR, Hassan TS, Margherio A, Oh KT, Raephaelian PV, Randhawa S, Sneed S, Trese MT, Yedavally S, Williams GA, Ruby AJ. Predictors of Endophthalmitis after Intravitreal Injection: A Multivariable Analysis Based on Injection Protocol and Povidone Iodine Strength. Ophthalmology Retina (2019). doi: 10.1016/j.oret.2018.09.013.
Intravitreal injections remain the most commonly performed procedure in medicine, and the numbers continue to grow in the treatment of age-related macular degeneration, diabetic retinopathy, and retinal vein occlusion. While injections are generally low-risk, endophthalmitis remains one of the most feared complications. Several studies have been done surveying retina specialists about their injection protocol, and there remains significant variation in injection protocol even within single practices, from use of gloves and lid speculums to anesthetic protocols. The only consistently proven preventative measure is the use of povidone-iodine. Even in this regard, the exact concentration of povidone-iodine that is needed remains debated.
In this retrospective cohort study, Stem et al evaluated whether several different variations in injection protocol in their practice were associated with an increased risk of endophthalmitis. Billing codes, specifically International Classification of Diseases and Current Procedural Terminology codes, were used to identify all intravitreal injections and cases of endophthalmitis within 15 days following an intravitreal injection during a three year period from January 1, 2014 to March 31, 2017. Any bilateral same-day injections were excluded for the multivariable analysis of risk factors.
Overall, 154,198 intravitreal injections were included in the study. Fifty-eight cases of intravitreal injection-related endophthalmitis were found during this study period, or 1 in 2,659 injections. Additionally, approximately half of these cases had positive cultures. These results are comparable to several prior studies of endophthalmitis rates and associated culture results. Of these injections, 98,960 intravitreal injections and 40 cases of endophthalmitis were included in the multivariate logistic regression model assessing risk factors. Factors evaluated were: concentration of povidone-iodine (5% vs. 10%), use of a lid speculum, use of gloves by the injecting provider, enforcement of a no-talking policy, use of Tetravisc, lidocaine gel, or subconjunctival lidocaine, injection site, conjunctival displacement, use of topical antibiotics, and intravitreal medication. After adjusting for possible confounders, use of 2% lidocaine gel (OR: 11.28, 95% CI 3.39-37.46, p<0.001) and 0.5% Tetravisc (OR: 3.95, 95% CI 1.15-13.50, p=0.03) were risk factors for post-intravitreal injection endophthalmitis. None of the other factors evaluated were significantly associated with endophthalmitis.
Overall, this study supports the results from multiple prior studies that speculum use, injection site, conjunctival displacement, and use of topical antibiotics are not associated with an increased risk of endophthalmitis. The use of a more viscous anesthetic, such as lidocaine gel or Tetravisc, may be associated with a higher risk of endophthalmitis due to increased microbial survival. Additional studies are needed to further investigate this potential risk factor.
Durga S. Borkar, MD
Duke Eye Center