Combining a Frailty Index Based on Laboratory Data and Pneumonia Severity Assessments to Predict In-Hospital Outcomes in Older Adults with Community-Acquired Pneumonia

Due to the increased morbidity, mortality, and cost of community-acquired pneumonia (CAP) in older people, strategies directed at improving disease evaluation and prevention are imperative. We independently compared the 30-day in-hospital mortality prediction ability of a frailty index based on laboratory data (FI-Lab) with that of the CURB-65 and the Pneumonia Severity Index (PSI) and then proposed combining them to further improve prediction efficiency. Retrospective cohort study. Patients aged ≥ 65 years (n = 2039) with CAP who were admitted to Jiangsu Provincial People's Hospital of Nanjing Medical University and Jiangsu Provincial Hospital of Chinese Medicine from January 2019 to June 2022. The 29-item FI-Lab, PSI and, CURB-65 were administered at admission. We defined frailty by the cut-off value of the FI-Lab score (> 0.43). Multivariable logistic regression analysis, together with the calculation of the area under the receiver operating characteristic curve (ROC-AUC), was conducted to identify stratified risks and relationships between the three indices and 30-day mortality. Participants were divided into the following three groups based on age: 65–74 years, 75–84 years, and ≥ 85 years. Hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality due to frailty were calculated. A total of 495 participants ranging from 65 to 100 years of age were ultimately included and divided into age groups (65–74 years, n = 190, 38.4%; 75–84 years, n = 183, 37.0%; ≥ 85 years, n = 122, 24.6%). A total of 142 (28.7%) of the 495 patients were defined as having frailty. All three scores tested in this study were significantly associated with 30-day mortality in the total sample. The ORs were as follows: 1.06 (95% CI: 1.03–1.09, P < 0.001) and 2.33 (95% CI: 1.26–4.31, P = 0.007) for the FI-Lab when the score was treated as a continuous and categorical variable, respectively; 1.04 (95% CI: 1.02–1.05, P < 0.001) for the PSI; and 3.70 (95% CI: 2.48–5.50, P < 0.001) for the CURB-65. In the total sample, the ROC-AUCs were 0.783 (95% CI: 0.744–0.819) for the FI-Lab, 0.812 (95% CI: 0.775–0.845) for the PSI, and 0.799 (95% CI: 0.761–0.834) for the CURB-65 (P < 0.001). The ROC-AUC slightly improved when the FI-Lab was added to the PSI (AUC 0.850, 95% CI: 0.809–0.892, P = 0.031) and to the CURB-65 (AUC 0.839, 95% CI: 0.794–0.885, P = 0.002). Older patients with frailty showed a higher risk of in-hospital mortality, with an HR of 2.25 (95% CI: 1.14–3.58, P < 0.001). The FI-Lab seems to generate simple and readily available data, suggesting that it could be a useful complement to the CURB-65 and the PSI as effective predictors of 30-day mortality due to CAP in older populations.