The predictive value of international classification of disease codes for chronic hepatitis c virus infection surveillance: The utility and limitations of electronic health records.
Abara W, Moorman A, Zhong Y, Collier M, Rupp L, Gordon S, Boscarino J, Schmidt M, Trinacty C, Holmberg S. The predictive value of international classification of disease codes for chronic hepatitis c virus infection surveillance: The utility and limitations of electronic health records.. Popul Health Manag. 2018; 22(2):110-115.
Popul Health Manag.
Surveillance of chronic hepatitis C virus (HCV) cases faces limitations that result in delays and underreporting. With increasing use of electronic health records (EHRs), the authors evaluated the predictive value of using International Classification of Diseases, Ninth Revision (ICD-9) codes to identify chronic HCV cases from EHR data. Longitudinal EHR data from 4 health care systems during 2006-2012 were evaluated. Using chart abstraction and review to confirm chronic HCV cases (“gold standard” definition), the authors calculated the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 2 case definitions: (1) >/=2 ICD-9 codes separated by >/= 6 months and (2) >/=1 positive HCV RNA (ribonucleic acid) test. Among 2,718,995 patients, 20,779 (0.8%) with ICD-9 codes indicating a likely diagnosis of chronic HCV infection were identified; 13,595 (65.4%) of these were randomly selected for review. Case definition 1 (>/= 2 ICD-9 codes separated by >/= 6 months) had 70.3% sensitivity, 91.9% PPV, 99.9% specificity, and 99.9% NPV while case definition 2 (>/= 1 positive HCV RNA test) had 74.1% sensitivity, 97.4% PPV, 99.9% specificity, and 99.9% NPV. The predictive values of these alternate EHR-derived ICD-9 code-based case definitions suggest that these measures may be useful in capturing the burden of diagnosed chronic HCV infections. Their use can augment current chronic HCV case surveillance efforts; however, their accuracy may vary by length of observation and completeness of EHR data.