Back

Risk Factors for Post-Injection Endophthalmitis: A Retrospective National Study in the IRIS(R) Registry (Intelligent Research in Sight)

Ghauri, S. Y.; Ross, C.; Gilbert, J. B.; Hu, D. J.; Gong, D.; Greenberg, P. B.; Eliott, D.; Elze, T.; Lorch, A.; Miller, J. W.; Krzystolik, M. G.

2026-01-13 ophthalmology
10.64898/2026.01.12.25336174 medRxiv
Show abstract

PurposeTo investigate the epidemiology of post-injection endophthalmitis (PIE) and evaluate the association of sociodemographic and clinical factors with incidence, timing of onset, and presenting visual acuity (VA) using the IRIS(R) Registry (Intelligent Research in Sight). DesignRetrospective cohort study. ParticipantsPatients with endophthalmitis after an intravitreal anti-VEGF injection in the IRIS(R) Registry from 2016-2023. MethodsOnly the first anti-VEGF injection per eye was included. Exclusion criteria were cataract surgery during the study, intravitreal corticosteroids within 30 days prior to PIE, uveitis, or cystoid macular edema. Mean best VA was recorded within 100 days prior to anti-VEGF treatment and at the time of PIE. Regression modeling evaluated associations between endophthalmitis and sociodemographic and clinical factors, and time to PIE. Linear regression assessed predictors of VA at the time of PIE, and descriptive statistics were used to analyze time to onset. Main Outcome MeasuresIncidence of post-injection endophthalmitis, time to symptom onset, and best VA at diagnosis. ResultsAmong 1,025,788 eyes treated, 600 (0.059%) developed endophthalmitis. Key risk factors included residence in U.S. territories (OR = 2.62; P = 0.038 vs. Northeast) and history of intravitreal corticosteroid injection (OR = 2.35; P = 0.004 vs. no history). The strongest protective factor was non-smoking (OR = 0.71; P = 0.023 vs. smokers). The median time from injection to onset of PIE was 5 days (interquartile range [IQR]: 3-8). The salient predictors of time to PIE included patient age (4.3 days sooner per decade older; P = 0.04), prior corticosteroid treatment (11.7 days sooner; P = 0.02), and a diagnosis of diabetic retinopathy (2.3 days sooner; P=0.03). Baseline VA before PIE was the only significant predictor of VA at the time of PIE diagnosis (0.67, P < 0.001). ConclusionPost-injection endophthalmitis was significantly associated with residence in U.S. territories and prior intravitreal corticosteroid exposure, while non-smoking status was protective. Most cases presented 3-8 days following anti-VEGF injection. Older age, history of prior corticosteroid treatment, and diabetic retinopathy were associated with earlier PIE. Baseline VA predicted VA at the time of PIE.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.