Comparative Prognostic Accuracy of Clinical and Inflammation- or Nutrition-Based Scores in Older Adults with...

This study aimed to assess the prognostic accuracy of the Glasgow Prognostic Score (GPS), modified Glasgow Prognostic Score (mGPS), and C-reactive protein/albumin ratio (CAR) in predicting 30-day mortality and intensive care unit (ICU) admission compared with the Pneumonia Severity Index (PSI) and CURB-65 in older adults with community-acquired pneumonia (CAP).

This retrospective, single-center cohort study was conducted in a tertiary emergency department. Patients aged ≥65 years with CAP were included. Exclusion criteria were hospital- or ventilator-associated pneumonia, pneumonia mimics, and immunocompromised status. GPS and mGPS were calculated using CRP >10 mg/L and albumin <35 g/L. ROC and logistic regression analyses were performed.

A total of 349 patients (mean age: 77.96 ± 8.42 years; 52.7% men) were included. The 30-day mortality and ICU admission rates were 19.5% and 27.2%, respectively. For predicting mortality, the GPS showed an AUC of 0.753 (95% CI: 0.690-0.816), sensitivity of 75.0%, specificity of 73.3%, PPV of 43.9%, and NPV of 92.4%. mGPS had an AUC of 0.747 (95% CI: 0.679-0.814), sensitivity 77.9%, specificity 73.3%, PPV 45.2%, and NPV 93.2%. The CAR yielded an AUC of 0.677 (95% CI: 0.604-0.751), sensitivity of 82.4%, specificity of 45.6%, PPV of 29.5%, and NPV of 91.4%. For ICU admission, the AUCs were 0.770 (GPS), 0.757 (mGPS), and 0.676 (CAR). The PSI demonstrated the highest predictive accuracy (AUC: 0.884 for mortality, 0.919 for ICU admission), followed by CURB-65 (AUC: 0.848 and 0.879, respectively). Independent predictors of 30-day mortality included acute confusion, lower PaO₂/FiO₂ ratio, low systolic blood pressure, reduced hemoglobin levels, and Alzheimer's disease or dementia.

The PSI and CURB-65 demonstrated superior prognostic accuracy. GPS and mGPS showed moderate performance, whereas CAR exhibited the lowest overall discriminative ability for both outcomes.