Daniel I. Steinberg, MD, FHM, FACP
How well do current risk stratification scales predict mortality in patients with community-acquired pneumonia (CAP)?
Included studies evaluated the accuracy of ≥ 1 of 4 scales for predicting mortality in patients (including nursing home residents) with radiographically confirmed CAP defined using signs, symptoms, and chest radiographs; enrolled patients between 1999 and 2009; included ≥ 100 patients; and reported mortality at 4 to 8 weeks by scale severity categories. Exclusion criteria included hospital-acquired pneumonia, outcomes reported only for specific types of pneumonia, community studies that did not include radiologic and laboratory tests, and randomized trials that compared different antibiotics. Scales evaluated were the Pneumonia Severity Index (PSI); Confusion, Urea nitrogen, Respiratory rate, and Blood pressure (CURB) scale; CURB-65 scale (CURB components plus age ≥ 65 y); and CRB-65 scale (CURB-65 components excluding urea nitrogen). Outcomes were sensitivity, specificity, and diagnostic odds ratios, evaluated using high-risk categories of each scale.
MEDLINE and EMBASE/Excerpta Medica (1999 to Oct 2009) were searched for English-language prospective studies. Reference lists of included studies were reviewed, and authors were contacted. 23 studies (n = 22 753, mean age 57 to 79 y, 45% to 74% men) met the selection criteria: 16 used the PSI (high-risk, class IV and V), 12 used CURB-65 (high-risk, score ≥ 3), 10 used CRB-65 (high-risk, score ≥ 2), and 5 used CURB (high-risk, score ≥ 2).
7.4% of patients died. Pooled analysis showed that the PSI had higher sensitivity and lower specificity than the other scales for predicting mortality. CURB and CURB-65 had moderate sensitivity and specificity, and CRB-65 had lower sensitivity and higher specificity than the other scales (Table). Significant heterogeneity, which was present for all scales, was unexplained.
In patients with community-acquired pneumonia, the Pneumonia Severity Index and CRB-65 scale are the most sensitive and specific, respectively, for predicting mortality. At the observed mortality rate, no scale is clinically more useful than the others.
Pooled test characteristics of high-risk categories of scales for predicting mortality in patients with community-acquired pneumonia*
* See text for expansion of scale names; other abbreviations defined in Glossary. +LR and −LR calculated from data in article using a random-effects model.
†High-risk categories: PSI class IV or V; CURB-65 score ≥ 3; CURB and CRB-65 score ≥ 2.
Steinberg DI. Review: Risk prediction scales have different strengths and weaknesses for predicting mortality in community-acquired pneumonia. Ann Intern Med. ;154:JC4–12. doi: 10.7326/0003-4819-154-8-201104190-02012
Download citation file:
Published: Ann Intern Med. 2011;154(8):JC4-12.
Infectious Disease, Pneumonia, Pulmonary/Critical Care.
Results provided by:
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use