In a nationwide cohort of 94,388 middle-aged and older adults who initiated DAA therapy in the largest provider of HCV care in the United States between 2014 and 2018, we found no evidence that any level of alcohol use was associated with decreased odds of achieving SVR and this finding did not differ by baseline stage of hepatic fibrosis. Moreover, 94% of all patients achieved SVR in this “real-world” setting of an older population with multiple comorbidities who are typically underrepresented in clinical trials. Taken together, our findings support providing DAA therapy without regard to reported alcohol consumption or AUD; furthermore, payors should no longer require alcohol abstinence for reimbursement of DAA therapy for HCV infection.

Our results support the current American Association for the Study of Liver Diseases/Infectious Diseases Society of America recommendations that current or prior alcohol use is not a contraindication to HCV DAA therapy. However, a recent analysis of administrative claims and encounters in the United States found that only 23% of Medicaid, 28% of Medicare, and 35% of private insurance recipients initiated DAA treatment within one year of HCV diagnosis.³⁰ Furthermore, this analysis found that Medicaid recipients were less likely to initiate DAA treatment if they resided in a state with Medicaid treatment restrictions including alcohol abstinence. Even though VA has not required alcohol abstinence prior to DAA treatment, pre-existing alcohol use still impacts who initiates DAA therapy.³¹ Given the increased risk of cirrhosis and hepatocellular carcinoma in those with chronic HCV and AUD, our findings suggest guidelines should encourage HCV treatment in those with AUD, rather than creating additional barriers to patients accessing HCV treatment. Integrating HCV treatment with AUD treatment services may prove to be a successful care model, especially because the hepatotoxic effects of alcohol on the liver have been shown to persist even after SVR.^32,33

While numerous studies from the interferon era examined alcohol use and SVR,^1–4 there are few such studies in the DAA era. Tsui et al examined alcohol use and SVR in the VA including the first 18 months of the DAA era.⁷ Although the authors found no difference in SVR across alcohol use categories in all primary analyses, after imputing missing HCV RNA values for 9% of their cohort, they observed a significantly lower likelihood of SVR in those reporting unhealthy alcohol use. However, Tsui et al also demonstrated that those with unhealthy alcohol use (defined as AUDIT-C ≥4) were more likely to have missing HCV RNA values and therefore assumptions required for multiple imputation may have been violated.²⁹

Furthermore, Tsui et al used patients reporting abstinence as the referent group; however, this is a heterogeneous group comprising very few lifetime abstainers and the majority of whom quit drinking after alcohol-related or health problems.⁸ To avoid misclassification, especially in a cohort of middle-aged and older adults, we used lower-risk consumption as the referent group and distinguished patients reporting abstinence by whether they had been diagnosed with AUD. Results from our study after imputing missing HCV RNA values did not change conclusions from our primary analysis; these analyses were included only to enable comparisons with Tsui et al since assumptions of missing at random were not met. However, after assuming everyone with missing outcome data did not achieve SVR, persons with high-risk consumption or AUD had 12% decreased odds of achieving SVR in multivariable analysis. We know in clinical practice that not every missing SVR value is a treatment failure and in fact many persons are found to have SVR when re-engaged in care;³⁴ therefore, we express caution in interpreting this single association.

This study has many strengths, including the availability of detailed, longitudinal, electronic health record data on a diverse nationwide cohort of HCV patients initiating DAA therapy and findings that were robust to multiple sensitivity analyses. Importantly, the VA has been more successful than most healthcare systems in the US in diagnosing and treating HCV infection given its unified electronic health record, nationalized healthcare system, and prioritization of HCV.³⁵ The result is that the VA has far exceeded other healthcare systems^30,36–40 by treating nearly 85% of patients with known chronic HCV infection in VA care.^10,41,42

We also recognize possible limitations. First, owing to the observational nature of the study, a degree of uncertainty persists due to the potential for residual confounding. Nonetheless, we mitigated potential confounding by comprehensively accounting for numerous demographic and clinical characteristics as well as liver- and treatment-related variables. Second, we assessed HCV RNA measurements in the 6-month period following the end of DAA treatment to define SVR, which may have resulted in potential misclassification of some patients who may have experienced viral rebound; however, this has been shown to be an extremely rare event.⁴³ Some patients may also have had their first evidence of SVR recorded beyond that window of measurement, which we would have classified as a missing outcome. Third, alcohol use measurement may have been influenced by both patient and provider-level factors,^44–47 including under-reporting level of alcohol use due to social desirability bias,⁴⁸ which may have resulted in misclassification of some patients with high-risk consumption at lower levels of consumption. However, the AUDIT-C tool is a widely validated questionnaire that has been part of a standardized triage practice within the VA and measured annually during routine healthcare visits since 2008. Furthermore, we categorized patients with AUD separately and used patients with lower-risk consumption as the referent category, which minimized the potential of misclassification of “sick quitters”.⁸ Fourth, we performed a complete case analysis, which will be unbiased if, conditional on model covariates, missingness is independent of the outcome.²⁶ To that end, we demonstrated that missing outcome data was not strongly associated with any covariate used in the model suggesting complete case analysis was appropriate. However, we additionally performed multiple sensitivity analyses, all of which were consistent with our findings from our complete case analysis. Finally, while individuals in VA care represent a diversity of backgrounds, women represented a small proportion of individuals in the cohort. However, given the current payor restrictions requiring alcohol abstinence prior to initiating DAA therapy, studying the relationship between alcohol use and HCV treatment outcomes, including SVR, may be limited in other US healthcare settings.

In conclusion, achieving SVR has been shown to reduce the risk of post-SVR outcomes, including hepatocellular carcinoma, liver-related mortality, and all-cause mortality. Our findings suggest that DAA therapy should be provided and reimbursed despite alcohol consumption or history of alcohol use disorder. Restricting access to DAA therapy based on alcohol consumption or AUD creates an unnecessary barrier to patients accessing DAA therapy and challenges HCV elimination goals.