Frailty assessed by administrative tools and mortality in patients with pneumonia admitted to the hospital and ICU in Wales

Abstract

AbstractThe ideal method of identifying frailty is uncertain, and data on long-term outcomes is relatively limited. We examined frailty indices derived from population-scale linked data on Intensive Care Unit (ICU) and hospitalised non-ICU patients with pneumonia to elucidate the influence of frailty on mortality. Longitudinal cohort study between 2010–2018 using population-scale anonymised data linkage of healthcare records for adults admitted to hospital with pneumonia in Wales. Primary outcome was in-patient mortality. Odds Ratios (ORs [95% confidence interval]) for age, hospital frailty risk score (HFRS), electronic frailty index (eFI), Charlson comorbidity index (CCI), and social deprivation index were estimated using multivariate logistic regression models. The area under the receiver operating characteristic curve (AUC) was estimated to determine the best fitting models. Of the 107,188 patients, mean (SD) age was 72.6 (16.6) years, 50% were men. The models adjusted for the two frailty indices and the comorbidity index had an increased odds of in-patient mortality for individuals with an ICU admission (ORs for ICU admission in the eFI model 2.67 [2.55, 2.79], HFRS model 2.30 [2.20, 2.41], CCI model 2.62 [2.51, 2.75]). Models indicated advancing age, increased frailty and comorbidity were also associated with an increased odds of in-patient mortality (eFI, baseline fit, ORs: mild 1.09 [1.04, 1.13], moderate 1.13 [1.08, 1.18], severe 1.17 [1.10, 1.23]. HFRS, baseline low, ORs: intermediate 2.65 [2.55, 2.75], high 3.31 [3.17, 3.45]). CCI, baseline < 1, ORs: ‘1–10′ 1.15 [1.11, 1.20], > 10 2.50 [2.41, 2.60]). For predicting inpatient deaths, the CCI and HFRS based models were similar, however for longer term outcomes the CCI based model was superior. Frailty and comorbidity are significant risk factors for patients admitted to hospital with pneumonia. Frailty and comorbidity scores based on administrative data have only moderate ability to predict outcome.