Severity assessment tools for predicting mortality in hospitalised patients with community-acquired pneumonia. Systematic review and meta-analysis
- James D Chalmers1,
- Aran Singanayagam2,
- Ahsan R Akram2,
- Pallavi Mandal2,
- Philip M Short3,
- Gourab Choudhury2,
- Victoria Wood1,
- Adam T Hill2
- 1University of Edinburgh, Edinburgh, UK
- 2Department of Respiratory Medicine, Royal Infirmary of Edinburgh, Edinburgh, UK
- 3University of Dundee, Tayside, UK
- Correspondence to Dr James Chalmers, Department of Respiratory Medicine, Royal Infirmary of Edinburgh, 51 Little France Crescent, Old Dalkeith Road, Edinburgh EH16 4SA, UK;
Contributors All authors had full access to the data in the study. The analysis was conducted by JDC and he takes responsibility for the integrity of the data and the accuracy of the data analysis.
- Received 13 December 2009
- Accepted 28 May 2010
- Published Online First 20 August 2010
Introduction International guidelines recommend a severity-based approach to management in community-acquired pneumonia. CURB65, CRB65 and the Pneumonia Severity Index (PSI) are the most widely recommended severity scores. The aim of this study was to compare the performance characteristics of these scores for predicting mortality in community-acquired pneumonia.
Methods A systematic review and meta-analysis was conducted according to MOOSE (meta-analysis of observational studies in epidemiology) guidelines. PUBMED and EMBASE were searched (1980–2009). 40 studies reporting prognostic information for the PSI, CURB65 and CRB65 severity scores were identified. Performance characteristics were pooled using a random effects model. Relationships between sensitivity and specificity were plotted using summary receiver operator characteristic (sROC) curves.
Results All three scores predicted 30 day mortality. The PSI had the highest area under the sROC curve, 0.81 (SE 0.008), compared with CURB65, 0.80 (SE 0.008), p=0.1, and CRB65, 0.79 (0.01), p=0.09. These differences were not statistically significant. Performance characteristics were similar across comparable cut-offs for low, intermediate and high risk for each score. In identifying low risk patients, PSI (groups I and II) had the best negative likelihood ratio 0.08 (0.06–0.12) compared with CURB65 (score 0–1) 0.21 (0.15–0.30) and CRB65 (score 0), 0.15 (0.10–0.22).
Conclusion There were no significant differences in overall test performance between PSI, CURB65 and CRB65 for predicting mortality from community-acquired pneumonia.