BACKGROUND AND OBJECTIVE: Our main objective was to assess the utility of the Pneumonia Severity Index (PSI) to decide the site of care home or hospital of patients with community-acquired pneumonia (CAP). PATIENTS AND METHOD: All CAP patients who came to the emergency department from 1 January to 31 December, 2000, were prospectively assessed with a protocol based on the PSI and additional admission criteria applied to classes I, II and III. Mortality within 30 days and poor outcome were used as endpoints. We tested the diagnostic efficacy of the PSI scale in predicting mortality or unfavourable events by calculating the area below the ROC curve. RESULTS: Of the 243 CAP patients included, 124 (51%) belonged to classes I, II and III, and 119 (49%) belonged to classes IV and V. One hundred and fifty six (64%) patients were admitted. Fifteen (6.2%) patients died, all of them belonging to classes IV and V. Forty four (18%) patients showed a poor outcome. Only one patient who was initially sent home had a poor outcome. The prognostic value of the PSI scale to predict mortality (ROC = 0.92; CI 95%, 0.88-0.95) was high. CONCLUSIONS: Our results confirm that the PSI scale is a good prognostic index in clinical practice for predicting mortality due to CAP. In order to use the PSI to decide the site of care of patients with CAP, not only the score obtained but also additional factors should be taken into account.
BACKGROUND AND OBJECTIVE: Our main objective was to assess the utility of the Pneumonia Severity Index (PSI) to decide the site of care home or hospital of patients with community-acquired pneumonia (CAP). PATIENTS AND METHOD: All CAP patients who came to the emergency department from 1 January to 31 December, 2000, were prospectively assessed with a protocol based on the PSI and additional admission criteria applied to classes I, II and III. Mortality within 30 days and poor outcome were used as endpoints. We tested the diagnostic efficacy of the PSI scale in predicting mortality or unfavourable events by calculating the area below the ROC curve. RESULTS: Of the 243 CAP patients included, 124 (51%) belonged to classes I, II and III, and 119 (49%) belonged to classes IV and V. One hundred and fifty six (64%) patients were admitted. Fifteen (6.2%) patients died, all of them belonging to classes IV and V. Forty four (18%) patients showed a poor outcome. Only one patient who was initially sent home had a poor outcome. The prognostic value of the PSI scale to predict mortality (ROC = 0.92; CI 95%, 0.88-0.95) was high. CONCLUSIONS: Our results confirm that the PSI scale is a good prognostic index in clinical practice for predicting mortality due to CAP. In order to use the PSI to decide the site of care of patients with CAP, not only the score obtained but also additional factors should be taken into account.