Validation of predictive rules for ICU patients with community-acquired pneumonia

Introduction

The objective of the study was to validate specific rules and nonspecific scores for prediction of mortality in patients with severe community-acquired pneumonia (CAP).

Methods

The study included 300 patients with CAP admitted to six ICUs of the city during 2005–2006. On admission each of the patients was assessed using the specific rules PORT, CURB-65, CRB-65, SMART-CO, and SOFA and APACHE II scores with regards to pneumonia severity and mortality. All data were analysed and processed on receiver operating characteristic curves.

Results

See Table 1. The results of analysis demonstrated high predictive values of the specific rule CAP and SOFA score. The areas under the receiver operating characteristic curves (AUROCs) were compared. The APACHE II score did not have prognostic ability, because the difference in AUROC did not have statistical significance to the diagonal: 0.71 ± 0.17 (P = 0.2). PORT and SOFA scores have maximal sensitivity and specificity: 92.3 (63.9–98.7) and 81.0 (65.9–91.4).

Table 1

Conclusion

The specific rules PORT, CURB-65, CRB-65, SMART-CO and SOFA score are comparatively informative and valuable in predicting short-term mortality in severe CAP. The APACHE II score is of low specificity and cannot be used for prediction outcomes.