BACKGROUND: Historically, “traditional reporting” of hepatitis C consisted of acute and chronic HCV cases reported by labs and healthcare providers by phone/fax, combined with only first reports from a subset of labs reporting data to the Washington State Department of Health via electronic lab reporting (ELR). These reports were manually entered into Public Health – Seattle & King County’s (PHSKC) local surveillance database, which was modified as part of the CDC-funded Hepatitis C Test and Cure (HCV-TAC) program to upload all ELR reports to the local surveillance database automatically (“a-ELR”). Six HCV-TAC healthcare facility (HCF) sites also submit lab and clinical data for HCV cases electronically to PHSKC quarterly (“partner-based reporting”).
METHODS: We used deterministic and probabilistic algorithms to identify potential matches across data sources (a-ELR, HCF, and traditional reporting). We manually reviewed potential matches to derive a dataset of unique individuals reported from any of the three data sources during the period January 2013 - November 2016. We calculated completeness of traditional reporting by counting the number of cases of HCV identified, divided by the total number of unique individuals identified by any of the three data sources. We determined the frequency that information captured via a-ELR and partner-based reporting contributed to establishing case classification. Lastly, we examined the frequency that data elements (patient last/first name, sex, date of birth, address, phone, SSN, and race/ethnicity) were reported by data source.
RESULTS: Of 34,224 unique person records identified by any reporting source, 28,614 (83.6%) were captured by traditional reporting. Of the total number of confirmed and probable HCV cases (n=24,787), 1322 would not have been identified without a-ELR. Of 727 records that initially had a case classification other than confirmed, 264 (36.3%) were confirmed by reporting from HCF. Across all parameters examined, data elements submitted by HCF were more complete relative to data captured by a-ELR and traditional reporting, with not more than 4% of records missing any of the data elements examined. While cases captured exclusively by ELR had fewer records with missing SSN, date of birth, and city/zip, completeness for all other data elements was higher with traditional reporting.
CONCLUSIONS: Electronic data collection and corresponding modifications to our local surveillance database have identified additional cases and improved determination of case classification, while reducing reliance on resource-intensive manual data entry. Capturing more complete demographics in ELR could enhance the utility of reporting.