Comparison of three predictive rules for assessing severity in elderly patients with CAP

Abstract

This study compares the ability of the Pneumonia Severity Index (PSI) and the British Thoracic Society CURB-65 and CRB-65 rules in predicting short-term mortality among elderly patients with community-acquired pneumonia (CAP). It is a population-based study including all people over 65 years old with a radiographically confirmed CAP in the region of Tarragona (Spain) between 2002 and 2008. Treatment setting and clinical variables were considered for each patient. PSI, CURB-65 and CRB-65 scores were calculated at the moment of diagnosis and 30-day mortality was considered as a main dependent variable. The rules were compared based on sensitivity, specificity and area under the receiver operating characteristic curve (AUC). Of the total 590 CAP cases, mortality rate was 13.6% (15.3% in hospitalised and 1.4% in outpatient cases; p = 0.001). Mortality increased with increasing PSI score (None in class II, 6,9% in class III, 14,4% in class IV and 29,5% in class V), CURB-65 score (7.5%, 14.5%, 26.7%, 53.3% and 100% for scores 1,2,3,4 and 5 respectively) and CRB-65 score (6.6%, 26.1%, 40.5% and 50% for scores 1,2,3 and 4 respectively). The three rules performed too similarly to predict 30-day mortality, with a ROC area of 0.727 [95% confidence interval (CI): 0.67-0.79] for the PSI, 0.672 (95% CI: 0.61-0.74) for the CURB-65, and 0.719 (95% CI: 0.65-0.78) for the CRB-65. Our data shows that the analysed rules perform equally well among elderly people with CAP which supports the recommendation for using the simplified CRB-65 severity score among elderly patients in primary care or emergency visits.