Background: Hospital-acquired pneumonia (HAP) is a common infection occurring in hospitalized patients. CURB-65 (Confusion, Uremic, Respiratory, Blood Pressure, Age over 65) scoring system was developed to assess severity and risk in community-acquired pneumonia (CAP) but its use has not been validated in HAP patients in Indonesia. Objective: To validate the performance of CURB-65 scoring system to predict mortality in Indonesian patients with HAP.
Methods: This is a validation study with retrospective cohort design. Data were taken from medical records in Cipto Mangunkusumo Hospital, Jakarta, Indonesia, from January 2006 to December 2012. Among 204 recruited HAP patients, 171 patients with complete data were included. Hosmer-Lemeshow test was conducted to test the accuracy and area under Receiver Operating Curve (ROC) was calculated to discriminate the performance of CURB-65 scoring system. The specificity and sensitivity value of CURB-65 to predict mortality in HAP patients were analyzed. Statistical analysis was performed using SPSS 16.0.
Results: The overall mortality proportion is 42.7%. Hosmer-Lemeshow test shows good accuracy (p=1.00). The area under the ROC is 0.376, showing bad discriminating performance of CURB-65 scoring system. Using the cut-off value of score 2, the sensitivity of CURB-65 is 71.2% and its specificity is 42.9% to predict mortality in HAP. Conclusion: CURB-65 has low specificity and sensitivity value to predict mortality in HAP patients.
Key words: CURB-65, hospital-acquired pneumonia