SETTING: The ability of the pneumonia severity index (PSI) and the CRB-65 to identify patients with low vs. high risk for mortality among cancer patients with CAP has not been evaluated. DESIGN: Subjects with cancer, CAP/Ca(+), and without cancer, CAP/Ca(-), were identified from the Community-Acquired Pneumonia Organization database. Mortality for both groups was analyzed by comparing low vs. high risk for mortality for the PSI (Risk Class I, II and III vs. IV and V) and the CRB-65 score (scores 0 and 1 vs. 2, 3 and 4). RESULTS: A total of 2621 patients were included in the CAP/Ca(-) group and 280 in the CAP/Ca(+) group. In the CAP/Ca(+) group, no significant difference in mortality was detected in low vs. high risk populations, either for the PSI (P = 0.288) or for the CRB-65 score (P = 0.281). Analyzing the receiver operator characteristic curves, the concordance indexes for the CAP/Ca(+) group were respectively 0.53 and 0.54 for PSI and CRB-65. By fitting a multivariable logistic regression model, a significantly different trend in mortality was found between the CAP/Ca(-) and CAP/Ca(+) groups for both scoring systems. CONCLUSION: Clinical judgment will continue to be the physicians' primary tool in defining the site of care for cancer patients with CAP.
SETTING: The ability of the pneumonia severity index (PSI) and the CRB-65 to identify patients with low vs. high risk for mortality among cancerpatients with CAP has not been evaluated. DESIGN: Subjects with cancer, CAP/Ca(+), and without cancer, CAP/Ca(-), were identified from the Community-Acquired Pneumonia Organization database. Mortality for both groups was analyzed by comparing low vs. high risk for mortality for the PSI (Risk Class I, II and III vs. IV and V) and the CRB-65 score (scores 0 and 1 vs. 2, 3 and 4). RESULTS: A total of 2621 patients were included in the CAP/Ca(-) group and 280 in the CAP/Ca(+) group. In the CAP/Ca(+) group, no significant difference in mortality was detected in low vs. high risk populations, either for the PSI (P = 0.288) or for the CRB-65 score (P = 0.281). Analyzing the receiver operator characteristic curves, the concordance indexes for the CAP/Ca(+) group were respectively 0.53 and 0.54 for PSI and CRB-65. By fitting a multivariable logistic regression model, a significantly different trend in mortality was found between the CAP/Ca(-) and CAP/Ca(+) groups for both scoring systems. CONCLUSION: Clinical judgment will continue to be the physicians' primary tool in defining the site of care for cancerpatients with CAP.
Authors: Carmen Gonzalez; Tami Johnson; Kenneth Rolston; Kelly Merriman; Carla Warneke; Scott Evans Journal: Cancer Med Date: 2014-05-07 Impact factor: 4.452