Prognostic assessment in community-acquired pneumonia by pneumonia severity scores and biomarkers

Introduction

Classical biomarkers like C-reactive protein (CRP) or the leucocyte count are only inaccurate tools for predicting the severity of community-acquired pneumonia (CAP). Procalcitonin (PCT) was found to predict 28-day mortality in CAP (area under the curve (AUC) 0.8) similar to the combination of PCT and CRB65 score (AUC 0.83) [1]. Novel prognostic biomarkers, such as Pro-ET1 and MR-ProADM, were shown to correlate with CAP severity [2]. We aimed to compare the diagnostic accuracy to predict mortality and ICU admission of clinical severity scores, biomarkers and their combination.

Methods

Nine hundred and twenty-five CAP patients enrolled in the ProHOSP trial [3] were analyzed by assessing clinical severity scores (SMART-COP, PSI, CURB65) and biomarker levels (PCT, MR-ProADM, Pro-ET1). Receiver operating characteristic curves for 30-day mortality and ICU admission were used to calculate and compare the different predictive values.

Results

The AUC for the prediction of 30-day mortality was 0.84 with SMART-COP, 0.82 with PSI, 0.72 with CURB65, 0.59 with PCT, 0.75 with MR-ProADM and 0.75 with ProET1. ICU admission was predicted best by SMART-COP (AUC 0.83), compared with the other severity scores and biomarkers (PSI: 0.68, CURB65: 0.65, PCT: 0.7, Pro-ET1: 0.73, ProADM: 0.72). The combination of SMART-COP and MR-ProADM was superior to SMART-COP alone (AUC 0.84, P = 0.04).

Conclusions

The combination of MR-ProADM with SMART-COP significantly improved the prediction of ICU admission. Prognostic biomarkers should complement the clinical assessment of patients with LRTI to improve allocation of healthcare resources to high-risk patients.