Background: The recent Infectious Disease Society of America/American Thoracic Society guidelines for the management of community-acquired pneumonia (CAP) in adults defined a predictive rule to identify patients with severe CAP to determine the need for intensive care unit (ICU) admission. We clinically validated this rule.
Methods: We analyzed 2102 episodes of CAP in consecutively hospitalized patients over a 7-year period. The predictive rule consists of at least 1 of 2 major severity criteria (septic shock and invasive mechanical ventilation) or at least 3 of 9 minor severity criteria. We assessed the association of the predictive rule with ICU admission and mortality.
Results: A total of 235 episodes of CAP (11%) occurred in patients who were admitted to the ICU, whereas the predictive rule identified 397 (19%) of 2102 episodes as severe CAP. The predictive rule and the decision for ICU admission agreed in 1804 (86%) of the episodes (kappa coefficient, 0.45), with a sensitivity of 71% and a specificity of 88%, similar to the 2001 American Thoracic Society guidelines (sensitivity, 66%; specificity, 90%) in predicting ICU admission. Severe CAP criteria had higher sensitivity (58% vs. 46%) and similar specificity (88% vs. 90%), compared with the 2001 American Thoracic Society guidelines in predicting hospital mortality. Invasive mechanical ventilation was the main determinant for ICU admission, followed by septic shock. In the absence of major criteria, ICU admission was not related to survival of patients with minor severity criteria.
Conclusions: The predictive rule to identify severe CAP is accurate for ICU admission and improved the prediction of mortality, compared with the previous American Thoracic Society guidelines. The need for ICU admission derived from minor severity criteria alone is uncertain and deserves further investigation.