Accuracy of ICD9 and ICD 10 Codes for Clostridium difficile Among Ulcerative Colitis Patients

Abstract

Purpose: The incidence of Clostridium difficile (C. difficile) infections has increased among ulcerative colitis (UC) patients. Researchers have commonly used administrative databases to study associations between C. difficile infection and UC; however, the validity of diagnostic code for C. difficile in UC patients is unknown. The aims of this study were to: 1) evaluate the accuracy of administrative codes for C. difficile infections in UC; and 2) assess whether the frequency of testing for C. difficile and C. difficile diagnosis has changed over time. Methods: Population based surveillance was conducted in the Calgary Health Zone between January 1, 2004 and December 31, 2009 using the Discharge Abstract Database to identify adults (≥ 18 years) admitted to the hospital for UC flare-up (n=583). These patients were concurrently searched for ICD-9 (008.45) or ICD-10 (A04.7) codes for C. difficile diagnosis. In addition, the Calgary Laboratory Services (CLS) provided the results of all C. difficile toxin tests ordered for UC patients admitted to the hospital between 2004 and 2009 (n=389). Only tests performed during hospital admission and up to 7 days prior to admission to hospital were included in the analysis. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) with 95% CIs were calculated by comparing ICD9/10 codes for C. difficile in the Discharge Abstract Database with results of CLS stool C. difficile toxin testing. Subsequently, stool toxin tests were used to define the proportion of UC patients who were tested for C. difficile who had a positive result. Joint point regression analysis was used to test for temporal trends in C. difficile testing and positive C. difficile 2004-2009. Results: Over a 6-year period, 288 (74%) UC patients admitted to hospital were tested for C. difficile and 8 (2.8%) were positive. The sensitivity, specificity, PPV and NPV of administrative data in identifying C. difficile infection in hospital was 88% [47%-100%], 100% [98%-100%], 88% [47%-100%] and 100% [98%-100%], respectively. The proportion of UC patients who were tested for C. difficile and who had a positive result remained stable between 2004 and 2009 (Figure 1). Conclusion: Laboratory proven C. difficile infection occurred in 3% of hospitalized UC patients and administrative data accurately identified this. Approximately one quarter of hospitalized UC patients were not tested for C. difficile in hospital or 7 days before hospitalization, which has important implications for clinical care and quality assurance.