Background: Addition of assessment of comorbid diseases ('D') and oxygen saturation ('S') to the CRB-65 score has been recommended to improve its accuracy for risk stratification in community-acquired pneumonia (CAP). The aim of this study was to validate the resulting DS-CRB-65 score in a large cohort of patients with CAP.
Methods: A total of 4432 patients prospectively enrolled in the CAPNETZ cohort were included in this study. Predefined end points were 28-day mortality, requirement for mechanical ventilation or vasopressors (MV/VS) and requirement for MV/VS or intensive care unit admission (MV/VS/ICU). Receiver operating characteristic curve analysis was used to determine the accuracy of the CRB-65 score and the addition of D (extra-pulmonary comorbidities) and S (oxygen saturation <90% or partial pressure of oxygen <8 kPa). Binary logistic regression and the method of Hanley and McNeil were used to compare the criteria.
Results: The mortality rate was 4.0%, and 4.2% of patients required MV/VS and 6.6% required MV/VS/ICU. After multivariate analysis, D and S independently were added to the CRB-65 criteria for mortality prediction, but only S improved prediction of MV/VS and MV/VS/ICU (P < 0.001 for all). The area under the curve of the CRB-65 score was significantly improved by adding D and S for all end points (P < 0.02). Amongst patients who died or required MV/VS despite a CRB-65 score of 0, 64-80% would have been identified by the DS-CRB-65 score.
Conclusions: The addition of assessment of oxygenation and comorbidities significantly improved the prognostic accuracy of the CRB-65 score. Consequently, the DS-CRB-65 score may have a useful role in risk stratification algorithms for CAP.
Keywords: community-acquired pneumonia; mortality; prediction; risk score; risk stratification.
© 2015 The Association for the Publication of the Journal of Internal Medicine.