Ascertainment of Hepatitis B and C Infection from Linked Data Sources for Residents of New York City Diagnosed with Liver or Intrahepatic Bile Duct Cancer.

Abstract

Background: Chronic infection with hepatitis B or C substantially increases risk of hepatocellular carcinoma. However, central cancer registries do not routinely collect information on hepatitis diagnoses. We evaluated the extent to which information on hepatitis B or C diagnosis could be ascertained from linked external data sources for cancers reported to the New York State Cancer Registry.

Methods: We linked data for 14,747 New York City (NYC) residents diagnosed with liver or intrahepatic bile duct cancer during 2004-2018 to 2 data sources: (1) the NYC Viral Hepatitis Surveillance Registry, which collects information on reported probable and confirmed cases of hepatitis B and C from New York laboratories and health care providers, and (2) the New York Statewide Planning and Research Cooperative System (SPARCS), which captures hepatitis diagnosis codes from hospital inpatient stays and outpatient encounters. We determined whether documentation of hepatitis B or C was present in 1 or both data sources, assessed concordance between the data sources, and used multivariable-adjusted logistic regression to examine factors associated with discordance in hepatitis positivity.

Results: Of the 14,747 cancer cases included, 3,972 had documentation in either data source of hepatitis B (26.9%), 7,599 had documentation of hepatitis C (51.5%), and 9,753 had either diagnosis (66.1%). There was moderate to substantial agreement between the 2 data sources. The percent of NYC patients with any unrecorded hepatitis infection was 12.7% for the hepatitis registry and 7.8% for SPARCS, and discordance in hepatitis positivity was more common in certain individuals, including those aged ≥70 years at cancer diagnosis and those with intrahepatic bile duct cancer, Hispanic ethnicity (hepatitis registry only), and Black or Asian race (SPARCS only).

Conclusions: These results indicate that hospital discharge and public health surveillance data can be used to assess individual-level hepatitis B and C infection status in people diagnosed with liver cancer. Possible reasons for discrepancies between the data sources include incomplete reporting in the hepatitis registry, especially for earlier diagnosis years, differing case inclusion criteria, and differences in the linkage methods for the 2 data sources. This information can be used to enrich cancer registry data for epidemiologic analyses of hepatocellular carcinoma and other cancers.