Keratitis-associated endophthalmitis: clinical features and factors associated with evisceration in a long-term multicenter study

Abstract Background Keratitis-associated endophthalmitis (KAE) is a rare but devastating form of exogenous endophthalmitis that usually occurs in eyes with advanced ocular surface disease and often results in limited visualization of the posterior segment. Methods We performed a multicenter retrospective observational study including all patients with infectious endophthalmitis who underwent ocular sampling from January 2003 to December 2025 at eight tertiary public hospitals in Galicia and Asturias (Spain). Consecutive cases of infectious endophthalmitis were identified through microbiology databases and clinical records. KAE was defined clinically by concurrent or recent microbial keratitis with signs of intraocular inflammation, supported by B-scan ultrasonography when posterior segment visualization was limited. Positive culture from aqueous or vitreous humour was not required; microbiological confirmation was defined as growth from any ocular sample. Demographic, clinical, microbiological, therapeutic, and outcome variables were extracted to describe the cohort. A multivariable penalized logistic regression model using Firth’s method was used to evaluate factors associated with evisceration. Results Among 674 sampled endophthalmitis cases, 71 eyes (10.5%) met criteria for KAE. Median age was 74.4 years (IQR 20.8), and most patients were of rural origin (74.6%). Culture positivity from any ocular sample was 71.8%, with 66 microorganisms isolated. Gram-positive cocci predominated (43.9%), followed by fungi (22.7%) and gram-negative bacilli (22.7%). Antimicrobial susceptibility testing was available for 59 isolates; resistance to at least one antimicrobial agent was identified in 35 isolates (59.3%), although no isolate was resistant to ceftazidime and only one isolate was vancomycin resistant. Functional visual outcomes were available for 44 anatomically preserved eyes; paired baseline and final visual acuity measurements were available for 39 eyes. Median visual acuity shifted from 2.80 to 2.30 logMAR ( p = 0.0002), although both values represent profound low-vision categories. Evisceration was required in 27 eyes (38.0%). No robust independent predictor of evisceration was identified in the exploratory Firth penalized logistic regression model. Previous topical corticosteroid exposure was not associated with evisceration. Conclusions KAE was associated with substantial morbidity, limited visual recovery, and a high risk of globe loss. This multicenter cohort highlights the microbiological heterogeneity of KAE, the frequent presence of antimicrobial resistance, and the need for early diagnosis, comprehensive microbiological characterization, and timely treatment strategies addressing both the anterior and posterior segments. The factors associated with evisceration should be interpreted as exploratory, and no robust independent predictor of globe loss was identified.