Omino R, Mittal S, Kramer JR et al.
Department of Medicine, Baylor College of Medicine, Houston, TX, USA. Ronald.Omino@bcm.edu.
Digestive diseases and sciences. Mar 2017.
Administrative databases that include diagnostic codes are valuable sources of information for research purposes.To validate diagnostic codes for hepatocellular carcinoma (HCC) in chronic hepatitis B patients.We conducted a retrospective study of patients with chronic HBV seen in the national Veterans Administration (VA). HCC cases were identified by the presence of ICD-9 code 155.0. We randomly selected 200 HBV controls without this code as controls. We manually reviewed the electronic medical record (EMR) of all cases and controls to determine HCC status. We calculated the positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity for the HCC code. We conducted an implicit review of the false-positive cases to determine possible reasons for the miscoding.Of the 8350 patients with HBV, 416 had an ICD-9 code for HCC. Of these 416, 332 patients had confirmed HCC and 61 did not; HCC status was indeterminate for 23 patients. Of the 200 controls, none had HCC confirmed in the EMR. The PPV ranged from 85.3 to 80.0% and specificity ranged from 99.2 to 99.0% based on classification of indeterminate cases as true versus false positives, respectively. The NPV, sensitivity, and specificity were 100%. Two-thirds of false-positive cases were diagnosed with HCC prematurely as a workup of liver mass and latter imaging and/or biopsy were not diagnostic for HCC.The diagnostic code of HCC in chronic HBV patients in the VHA data is predictive of the presence of HCC in medical records and can be used for epidemiological and clinical research.