Accuracy in diagnoses recorded using the International Classification of Diseases (ICD) coding is the most important element ensuring the foundation of research using real-world data analyses. To evaluate the validity of ICD coding for diagnoses of disseminated intravascular coagulation (DIC) using the International Society on Thrombosis and Haemostasis (ISTH) overt DIC criteria and the Japanese Association for Acute Medicine (JAAM) DIC criteria as reference standards. This retrospective observational study included adult hospitalized patients diagnosed as having diseases potentially causing DIC extracted from a part of a large-scale database in Japan. The index test was a diagnosis of DIC based on the ICD-10 codes. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated using ISTH overt DIC criteria and JAAM-2 DIC criteria as the reference standards. We also conducted subgroup analyses according to the underlying diseases. We included 84,300 patients in this study. In the overall study population, sensitivity, specificity, PPV, and NPV of the ICD-based diagnosis for ISTH criteria were 26.28, 98.10, 35.12, and 97.14%, respectively. In subgroup analyses according to the underlying disease, sensitivity ranged from 9.48 to 52.08%, and specificity ranged from 96.94 to 99.47%. The accuracy of the ICD-based diagnosis for JAAM-2 criteria was similar to that for ISTH criteria. Identification of DIC patients using ICD-10 codes had relatively low sensitivity but very high specificity for DIC diagnostic criteria. Approximately 65% of patients identified by ICD coding are likely to meet the JAAM-2 DIC criteria.
Read full abstract