Validation of the 2001 American Thoracic Society criteria for severe community-acquired pneumonia.

STUDY OBJECTIVE: Ewig et al. proposed a new definition of severe community-acquired pneumonia in 1999, which was adopted by the American Thoracic Society in 2001. We evaluated this definition in an independent population of emergency department patients.

DESIGN: We compared the 2001 American Thoracic Society definition of severe community-acquired pneumonia using emergency department data to intensive care unit (ICU) admission, use of mechanical ventilation, and administration of vasopressors.

SETTING: LDS Hospital, a tertiary care, university-affiliated hospital with 520 total beds and 68 ICU beds in Salt Lake City, UT.

PATIENTS: We studied 980 consecutive emergency department patients with a radiographically confirmed diagnosis of pneumonia between June 1995 and June 1999. Of these patients, 498 were admitted to the hospital, immunocompetent, and without a "do-not-resuscitate" order within 24 hrs of admission.

MEASUREMENTS AND MAIN RESULTS: Forty-seven patients met the criteria for severe community-acquired pneumonia in the emergency department and were admitted to the ICU. Three hundred eighty patients did not meet the criteria and were admitted to a hospital unit. Nineteen patients met the definition but were admitted to a hospital unit; only one required subsequent ICU admission. Two of the 19 died after a do-not-resuscitate order was entered >24 hrs after admission; the remainder recovered. Fifty-two patients were triaged to the ICU but did not initially meet the definition of severe pneumonia. Sixteen of these 52 patients required mechanical ventilation, 13 of the 16 within 24 hrs of admission to the ICU. The sensitivity for the 2001 American Thoracic Society definition in our population was 44%, specificity was 95%, positive predictive value was 71%, and negative predictive value was 88%.

CONCLUSION: The 2001 American Thoracic Society definition of severe community-acquired pneumonia had high specificity but lower sensitivity in our population compared with the derivation population. Additional factors not reflected in the definition may contribute to ICU admission and the need for mechanical ventilation.