The Impact of Frailty Syndrome on Endogenous Endophthalmitis Development and Outcomes: A Population-Level Analysis.

Purpose: Characterize the impact of frailty on endogenous endophthalmitis (EE) development and clinical outcomes among septicemic patients.

Design: Population-level, retrospective cohort study.

Participants: Adult inpatients within the National Inpatient Sample (years 2002-2014) diagnosed with bacterial septicemia.

Methods: Septicemic patients were classified as frail or nonfrail using the previously validated Johns Hopkins Adjusted Clinical Groups frailty-defining diagnoses indicator, and diagnosis of EE was abstracted from International Classification of Diseases 9 codes. We used multivariable logistic regression to generate odds ratios (ORs) for rates of EE development and in-hospital mortality based on frailty status. We also examined the association between frailty and blood culture-proven organism class, inpatient length of stay, and total charges billed to insurance.

Main Outcome Measures: Incidence of EE among septicemic patients; rates of EE development among frail and nonfrail patients; blood culture-proven microbe type, length of stay, and total charges billed to insurance.

Results: 9294 of 18 470 658 (0.05%) inpatients with bacteremia developed EE, 2102 (22.6%) of whom had at least 1 frailty-defining feature (predominantly malnutrition [68%]). Odds of developing EE were 16.7% higher for frail patients (OR, 1.167; 95% confidence interval, 1.108-1.229) when controlling for age, sex, race, concomitant human immunodeficiency virus/AIDS, pyogenic liver abscess, infectious endocarditis, cirrhosis, and diabetes with chronic complications. Frail EE patients had a 27.9% increased odds of in-hospital death, independent of age, sex, race, and Elixhauser comorbidity score (OR, 1.279; 95% confidence interval, 1.056-1.549). Higher rates of methicillin-resistant Staphylococcus aureus bacteremia (14.3% vs. 10.9%, P = 0.000016), gram-negative bacteremia (7.6% vs. 4.9%, P = 0.000003), and concomitant candidemia (10.4% vs. 7.0%, P = 0.0000004) were associated with frailty. Hospital stays were significantly longer (median, 14 days; interquartile range, 19 days; P < 0.00001) and total charges billed to insurance were significantly greater (median, $96 398; interquartile range, $154,682; P < 0.00001) among frail EE patients.

Conclusions: Frailty syndrome is independently associated with development of EE in the setting of bacterial septicemia; frailty-associated EE may occur in patients with malnutrition and particular bacterial subtypes, and it predisposes to higher rates of in-hospital death and health care resource usage.