Accuracy of International classification of diseases, 10th revision codes for identifying severe sepsis in patients admitted from the emergency department

Abstract

Objective: To determine the accuracy of International classification of diseases, 10th revision, Australian modification (ICD-10-AM) codes in identifying severe sepsis in patients admitted from the emergency department (ED). Design, setting and participants: A retrospective cohort study of ED patients transferred to the intensive care unit of a tertiary hospital within 24 hours of leaving ED, 2000- 2006. Main outcome measures: Clinical diagnosis of severe sepsis compared with diagnosis-based code (DB-C) categories based on ICD-10-AM codes in the Emergency Department Information Systems (EDIS) and Hospital Morbidity Data System (HMDS); sensitivity, specificity, positive predictive value (PPV) and negative predictive value of these databases. Results: In the study period, 1645 patients were transferred to the ICU from the ED, of whom 254 had severe sepsis. Single discharge ICD-10-AM codes recorded in the EDIS and the principal ICD-10-AM codes recorded in the HMDS that fell into D-BC categories for sepsis, pneumonia, viscous perforation, peritonitis, cholecystitis or cholangitis had a PPV of 85.0% (95% CI, 78.4%-91.6%; 96/113) and 88.2% (95%CI, 72.6%-82.6%; 112/127), respectively. The respective sensitivity was 37.8% (95% CI, 31.8%-43.8%) (96/254) and 44.1% (95% CI, 38.0-50.2) (112/254). In contrast, ICD-10-AM codes in the HMDS that code for infection and organ dysfunction had a PPV of 33.5% (95% CI, 30.0%-37.0%; 227/677) and sensitivity of 89.4% (95% CI, 85.6%-93.2%; 227/254). Conclusion: ICD-10-AM codes recorded in the EDIS or HMD had limited utility for identifying severe sepsis in patients admitted to ICU from the ED.