Abstract
BACKGROUND: Determining the accuracy of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) codes for postpartum hemorrhage (PPH) is vital for reaching valid conclusions about the epidemiology of PPH. Our primary objectives were to assess the performance characteristics of ICD-9 PPH codes against a reference standard using estimated blood loss (EBL) among a cohort undergoing Cesarean delivery. STUDY DESIGN AND METHODS: We analyzed maternal discharge and EBL data from women who underwent Cesarean delivery at Kaiser Permanente Northern California facilities between 2010 and 2013. We defined PPH as an EBL of at least 1000 mL. In a secondary analysis, ICD-9 performance characteristics were assessed using an EBL of at least 1500 mL to classify severe PPH. RESULTS: We identified 35,614 hospitalizations for Cesarean delivery. Using EBL of at least 1000 mL as the "gold standard," PPH codes had a sensitivity of 27.8%, specificity of 97%, positive predictive value (PPV) of 74.5%, and a negative predictive value (NPV) of 80.9%. The prevalence of a PPH code (9%) was lower than the prevalence using a blood loss of at least 1000 mL (24%). Using a reference standard of EBL of at least 1500 mL, PPH codes had a sensitivity of 61.7%, specificity of 93.8%, PPV of 34.2%, and NPV of 97.9%. CONCLUSION: PPH ICD-9 codes have high specificity, moderately high PPVs and NPVs, and low sensitivity. An EBL of at least 1500 mL as a reference standard has higher sensitivity. Our findings suggest that, for women undergoing Cesarean delivery, quality improvement efforts are needed to enhance PPH ICD-9 coding accuracy in administrative data sets.