ABSTRACT
OBJECTIVES: To determine the baseline accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of routinely collected co-morbidity data in patients undergoing abdominal wall hernia repair. METHODS: All patients aged > 18 who underwent umbilical, para-umbilical, inguinal or incisional hernia repair between 1 January 2015 and 1 November 2016 were identified. All parts of the clinical notes were searched for co-morbidities by two authors independently. The following co-morbidities were considered: hypertension, ischaemic heart disease (IHD), diabetes, asthma, chronic obstructive pulmonary disease (COPD), cerebrovascular disease (CVD), chronic kidney disease (CKD), hypercholesterolemia, obesity and smoking. The co-morbidities data from clinical notes were compared with corresponding data in hospital episode statistics (HES) database to calculate accuracy, sensitivity, specificity, PPV and NPV of HES codes for co-morbidities. To assess the agreement between clinical notes and HES data, we also calculated Cohen's Kappa index value as a more robust measure of agreement. RESULTS: Overall, 346 patients comprising 3460 co-morbidity codes were included in the study. The overall accuracy of HES codes for all co-morbidities was 77% (Kappa: 0.13). When calculated separately for each co-morbidity, the accuracy was 72% (Kappa: 0.113) for hypertension, 82% (Kappa: 0.232) for IHD, 85% (Kappa: 0.203) for diabetes, 86% (Kappa: 0.287) for asthma, 91% (Kappa: 0.339) for COPD, 92% (Kappa: 0.374) for CVD, 94% (Kappa: 0.424) for CKD, 74% (Kappa: 0.074) for hypercholesterolemia, 71% (Kappa: 0.66) for obesity and 24% (Kappa: 0.005) for smoking. The overall sensitivity, specificity, PPV and NPV of HES codes were 9, 100, 100, and 77%, respectively. The results were consistent when individual co-morbidities were analyzed separately. CONCLUSIONS: Our results demonstrated that HES co-morbidity codes in patients undergoing abdominal wall hernia repair are specific with good positive predictive value; however, they have substandard accuracy, sensitivity, and negative predictive value. The presence of a relatively large number of false negative or missed cases in HES database explains our findings. Better documentation of co-morbidities in admission clerking proforma may help to improve the quality of source documents for coders, which in turn may improve the accuracy of coding.
