BACKGROUND: Chronic hepatitis C virus (HCV) infection can lead to cirrhosis and hepatocellular carcinoma if left untreated. Over half of all hepatocellular carcinoma cases in the U.S. are due to HCV. Increases in HCV incidence in U.S. young adults since the late 1990s have been attributed to rising intravenous use of heroin and prescription opioids. HCV is curable, but medications are often prioritized to persons with existing liver damage and are unavailable to persons who inject drugs (PWID). HCV is a nationally notifiable condition and surveillance data, adjusted for under-reporting, can be used in transmission models to estimate and improve precision of population-level parameters and evaluate potential interventions.
METHODS: An ordinary differential equation HCV transmission model with preferential age mixing was fit to newly reported chronic HCV cases among Michigan residents aged 15-30 years reported to the Michigan Department of Health and Human Services during 2000-2013. Model fit to surveillance data was used to estimate the HCV transmission and relapse rates among PWID, adjusted for under-reporting. Latin hypercube sampling was used to evaluate if data were consistent with population-level parameters from the literature. The potential impact of primary (reducing injection initiation), secondary (reducing injection drug use relapse and increasing cessation), and tertiary (HCV treatment) interventions on incident and prevalent HCV cases was simulated.
RESULTS: Estimates of the injection drug use relapse rate were dependent on the injection cessation rate. In the best fitting parameter set, PWID stopped injecting drugs after 1.4 years and relapsed after 4.4 years, on average, but results varied widely. Results supported a tendency towards higher than expected rates of under-reporting and lower prevalence of past-year drug injection compared to estimates identified in literature reviews. Interventions building on primary prevention initiatives (e.g. reducing injection initiation) predicted the largest reductions in HCV incidence and prevalence. Initiatives relying on tertiary interventions (treatment) predicted prevalence reductions but did not reduce incidence. Treatment of both current and former PWID in the absence of other interventions predicted reductions in new chronic HCV cases in only some scenarios.
CONCLUSIONS: Public health surveillance data can inform transmission models to improve precision of parameters and evaluate interventions. On average, young drug users in Michigan quit injecting drugs after 1.4 years and relapsed after 4.4 years, but results were variable; measurement of one parameter could inform the value of the other. Primary interventions decreasing injection initiation could concurrently reduce prevalence and new cases.