Prediction of Mortality in Community-Acquired Pneumonia in Hospitalized Patients

Introduction: Community-acquired pneumonia (CAP) is common and associated with a significant mortality. Currently recommended criteria to assess severity of CAP could be improved. Methods: We derived 2 new criteria CARSI [confusion, age (<65, >= 65 to <85 or >= 85), respiratory rate and shock index] and CARASI, where shock index is replaced by temperature-adjusted shock index based on previous observations. By using data of a prospective study performed in Norfolk and Suffolk, United Kingdom, we compare these new indices with the CURB-65 criteria. Results: A total of 190 patients were included (men, 53%). The age range was 18 to 101 years (median, 76 years). There were a total of 54 deaths during a 6-week follow-up, all within 30 days of admission. Sixty-five (34%) had severe pneumonia by CURB-65. Using CARSI and CARASI, 39 (21%) and 36 (19%) had severe pneumonia, respectively. Sensitivity was slightly less, but specificity was higher with CARSI and CARASI indices than that of CURB-65. Positive and negative predictive values in predicting death during 6-week follow-up were comparable among 3 indices examined. The receiver operating characteristic curve values (95% confidence interval) for the criteria were 0.67 (0.60-0.75) for CURB-65, 0.64 (0.60-0.71) for CARSI and 0.64 (0.57-0.71) for CARASI. Comparing receiver operating characteristic curves for CURB-65 versus CARSI, or CURB-65 versus CARASI, there was no evidence of a difference between the tools, P = 0.35 and 0.33, respectively. There was good agreement, which was strongly statistically significant (kappa = 0.56, P < 0.0001 and kappa = 0.54, P < 0.0001, respectively). Conclusions: Both CARSI and CARASI are useful in predicting deaths associated with CAP, including older patients, and may be particularly useful in the emergency and community settings.