Enhancing in-hospital mortality prediction in older patients with sepsis: the role of frailty indices and multidrug-resistance status in non-ICU wards—a proof-of-concept study

Abstract

Background

Prognostic stratification in older patients with sepsis is challenging due to frailty and the role of multidrug-resistant (MDR) infections.

Aims

To test the predictive accuracy of different frailty measures, blood routine tests and MDR infection status for in-hospital mortality among older patients with sepsis.

Methods

Consecutive patients aged ≥ 65 years with qSOFA ≥ 2 and positive cultures admitted to a tertiary care hospital were enrolled. Frailty was assessed using the Clinical Frailty Scale (CFS), the Primary Care–Frailty Index (PC-FI), and a 50-item FI. A base logistic regression model including age, sex, WBC count, platelets, creatinine, hs-CRP, and lactate predicted mortality. Frailty indices and MDR status were sequentially added, and model performance was compared using the area under the Receiver Operating Characteristics (AUROC). A nomogram was developed to visualize mortality probabilities.

Results

Among 93 patients (median age 80, IQR [72–84] years, 63.4% males), in-hospital mortality was 16.1%. Deceased patients were frailer and had a higher number of comorbidities. By logistic multivariable regression, the base model achieved an AUROC of 0.771 for predicting in-hospital mortality. Adding frailty indices improved model performance to 0.800 (PC-FI), 0.817 (CFS), and 0.823 (FI). Incorporating MDR status further increased AUROC to 0.890 (PC-FI + MDR), 0.907 (CFS + MDR), and 0.922 (FI + MDR), outperforming the base model (p < 0.05 for all).

Conclusions

Incorporating frailty indices and MDR status of culture isolates into traditional prognostic parameters improves mortality prediction in older patients admitted with sepsis, enabling more accurate risk stratification and personalized treatment strategies.