Severity of illness scoring systems in community-acquired Legionella pneumonia

Abstract

Introduction Prognostic and severity-of-illness scoring systems are valuable tools for predicting mortality and choosing the site of care for patients with community-acquired pneumonia (CAP) [1]. Legionnaires' disease (LD) is a pneumonia caused by Legionella spp. and carries a higher mortality rate (5 to 30%) than CAP of most other etiologies. The aim of our study was to evaluate five scoring systems commonly used in CAP for predicting mortality in patients with Legionella pneumophila serogroup 1 infection admitted during a large LD outbreak [2,3]. Methods Patients with microbiologically verified LD (n = 103) and CAP patients with epidemiological association to the outbreak with no other bacteriological etiology identified (n = 32) were included. A clinical protocol was initiated during an early phase of the outbreak, and clinical and biochemical data were collected from patients on admission to the regional hospital. The five evaluated scoring systems were: pneumonia severity index (PSI), CURB-65 (confusion, uremia, respiratory rate ≥30, low blood pressure, age ≥65) and CRB-65 score, the modified American Thoracic Society (ATS) score, and the IDSA/ATS guidelines. The endpoint was defined as 28-day mortality. (Table presented) Results The overall mortality rate was 12% (16/135), and 19% (25/135) were admitted to the ICU. The discriminatory power was highest for PSI, CURB- 65 and CRB-65 with area under the receiver operator characteristic curve (AUC) of 0.79, 0.78, and 0.75, respectively. The AUC of the modified ATS score and IDSA/ATS guidelines were 0.61 and 0.69, respectively. Table 1 shows that a PSI class IV or V, and a CURB-65 and CRB-65 score ≥2 yielded the highest sensitivity for prediction of mortality, but the specificity and positive predictive value was low. Conclusions The PSI, the CURB-65 and CRB-65 scores proved sensitive in predicting mortality in patients with Legionella pneumonia admitted during an LD outbreak, but the low specificities and positive predictive values necessitate thorough clinical judgment in patients with a high severity score. The modified ATS score and IDSA/ATS guidelines, which are decision recommendations for ICU admission, were not sensitive in predicting mortality from LD in this study.