Four methods for estimating hepatitis C incidence using extant testing data

BackgroundExtant hepatitis C virus (HCV) antibody (Ab) and ribonucleic acid (RNA) test results are widely used to estimate HCV incidence, but the impact of cohort specification and case definition on validity and generalizability is poorly understood. MethodsUsing databases linked at ICES, a cohort of 15.8 million Ontarians aged 18-80 between 1999 and 2018 was used to estimate annual HCV incidence using four methods: the population-based method defined incidence as new annual HCV cases divided by population size estimates; the test-negative method defined eligibility at first negative test; the RNA-based method prioritized specificity by requiring RNA+ tests; and the antibody-inclusive method prioritized sensitivity by including all Ab+ tests. Method assumptions, potential misclassification, and sensitivity to analytic choices were assessed. ResultsRNA-based estimates were lowest and fluctuated around 30 cases per 100,000 person-years, while population-based and antibody-inclusive estimates were 1.5-fold higher, and test-negative estimates were 7.9-fold higher. Population-based estimates were influenced by changes in the HCV case definition used in Ontario from 1999-2018. The test-negative cohort had a high prevalence of HIV and substance use disorder, limiting generalizability of HCV incidence estimates. RNA-based estimates likely underestimated HCV incidence because 22% of Ab+ tests were unconfirmed by RNA testing, while antibody-inclusive estimates likely overestimated HCV incidence by assuming all unconfirmed Ab+ tests were true cases. ConclusionThese findings illustrate the influence of cohort definition and HCV case definition when estimating HCV incidence using extant testing data, which will support more accurate monitoring of HCV burden and progress toward elimination.