Topical Antibiotic Prophylaxis and Intravitreal Injections: Impact on the Incidence of Acute Endophthalmitis-A Nationwide Study in France from 2009 to 2018

Abstract

Background: The dramatic increase in intravitreal injections (IVTs) has been accompanied by a greater need for safer procedures. The ongoing debate about topical antibiotic prophylaxis after IVTs emphasizes the importance of large-scale studies. We aimed to study the role of topical antibiotic prophylaxis in reducing the risk of acute endophthalmitis after IVTs. Methods: Population-based cohort study, in France, from 2009 to 2018, including all French patients receiving IVTs of corticosteroids or anti-VEGF agents. Results: A total of 5,291,420 IVTs were performed on 605,434 patients. The rate of topical antibiotic prophylaxis after IVTs progressively decreased during the study period, with a sharp drop in 2014 (from 84.6% in 2009 to 27.4% in 2018). Acute endophthalmitis occurred in 1274 cases (incidence rate = 0.0241%). Although antibiotic prophylaxis did not alter the risk of endophthalmitis (p = 0.06), univariate analysis showed an increased risk after fluoroquinolone and aminoglycoside prophylaxis. This increased risk was not found in multivariate analysis. However, we observed an increased risk related to the use of fixed combinations of fluoroquinolones and aminoglycosides with corticosteroids (IRR = 1.89; 95% CI = 1.57-2.27%, antibiotics combined with corticosteroids). Conclusion: These results are consistent with the literature. Endophthalmitis rates after IVTs did not decrease with topical antibiotic prophylaxis. The use of a combination of antibiotics and corticosteroids doubles the risk of endophthalmitis and should be avoided. Avoiding antibiotic prophylaxis would reduce the costs and the potential risks of antibiotic resistance.

Keywords: IVT; antibiotics; endophthalmitis; intravitreal injection.

Figure 1

Proportion of IVT treatments and their relative endophthalmitis incidence rates. Due to the low proportion of IVTs with triamcinolone, pegaptanib, and bevacizumab, they are not displayed in this figure.

Figure 2

Changes in antibiotic prophylaxis use with IVTs from 2009 to 2018. The years 2009 and 2018 are truncated to allow for the required 42 days of lookback and follow-up.

Figure 3

Proportion of IVTs performed with antibiotic prophylaxis per year, and corresponding endophthalmitis rates per year; * 2014, 2nd quarter, publication of the French guidelines on the absence of any benefit from antibiotic prophylaxis and prefilled ranibizumab availability. The years 2009 and 2018 are truncated to allow for the required 42 days of lookback and follow-up.

Figure 4

Univariate and multivariate analysis of factors associated with acute endophthalmitis after intravitreal injections of corticosteroids or anti-VEGF (vascular endothelial growth factor) agents from 2009 to 2018. IRR = incidence rate ratio; CI = confidence interval. Missing data for agent class and specific agent, n = 259,802. Multivariate Poisson regression with variable selection based on significant association with the event and on the quasi-likelihood under the independence model criterion (QIC) proposed by Pan, used to compare GEE models [16]. ^a Since we recorded only one prescription of phenicol antibiotics, univariate analysis is not shown for this class. ^b Since no endophthalmitis occurred after intravitreal bevacizumab injections, these 13,562 injections were not considered for the by-agent analysis.