Use of multiple data sources to estimate hepatitis C seroprevalence among prisoners: A retrospective cohort study

Hepatitis C is a major cause of preventable morbidity and mortality. Prisoners are a key population for hepatitis C control programs, and with the advent of highly effective therapies, prisons are increasingly important sites for hepatitis C diagnosis and treatment. Accurate estimates of hepatitis C prevalence among prisoners are needed in order to plan and resource service provision, however many prevalence estimates are based on surveys compromised by limited and potentially biased participation. We aimed to compare estimates derived from three different data sources, and to assess whether the use of self-report as a supplementary data source may help researchers assess the risk of selection bias. We used three data sources to estimate the prevalence of hepatitis C antibodies in a large cohort of Australian prisoners-prison medical records, self-reported status during a face-to-face interview prior to release from prison, and data from a statewide notifiable conditions surveillance system. Of 1,315 participants, 33.8% had at least one indicator of hepatitis C seropositivity, however less than one third of these (9.5% of the entire cohort) were identified by all three data sources. Among participants of known status, self-report had a sensitivity of 80.1% and a positive predictive value of 97.8%. Any one data source used in isolation would have under-estimated the prevalence of hepatitis C in this cohort. Using multiple data sources in studies of hepatitis C seroprevalence among prisoners may improve case detection and help researchers assess the risk of selection bias due to non-participation in serological testing.