Patient-centred models of hepatitis C treatment for people who inject drugs: a multicentre, pragmatic randomised trial

Background: To achieve WHO targets for the elimination of hepatitis C virus (HCV) as a public threat, an increased uptake of HCV treatment among people who inject drugs (PWID) is urgently needed. Optimal HCV co-located treatment models for PWID have not yet been identified. We aimed to compare two patient-centred models of HCV care in PWID with active drug use. Methods: We did a pragmatic randomised controlled trial at eight US cities in eight opioid treatment programmes and 15 community health centres. PWID actively injecting within 90 days of study entry were randomly assigned (1:1) to either patient navigation or modified directly observed therapy (mDOT) using computer-generated variable block sizes of 2–6 stratified by city, clinical settings, and cirrhosis status. The randomisation code was concealed, in a centralised REDCap database platform, from all investigators and research staff except for an authorised data manager at the data coordinating centre. All participants received a fixed-dose combination tablet (sofosbuvir 400 mg plus velpatasvir 100 mg) orally once daily for 12 weeks. The primary outcome was sustained virological response (SVR; determined by chart review between 70 days and 365 days after end of treatment and if unavailable, by study blood draws), and secondary outcomes were treatment initiation, adherence (measured by electronic blister packs), and treatment completion. Analyses were conducted within the modified intention-to-treat (mITT; all who initiated treatment), intention-to-treat (all who were randomised), and per-protocol populations. This trial is registered with ClinicalTrials.gov, NCT02824640. Findings: Between Sept 15, 2016, and Aug 14, 2018, 1891 individuals were screened and 1136 were excluded (213 declined to participate and 923 did not meet the eligibility criteria). We randomly assigned 755 participants to patient navigation (n=379) or mDOT (n=376). In the mITT sample of participants who were randomised and initiated treatment (n=623), 226 (74% [95% CI 69–79]) of 306 participants in the mDOT group and 236 (76% [69–79]) of 317 in the patient navigation group had an SVR, with no significant difference between the groups (adjusted odds ratio [AOR] 0·97 [95% CI 0·66–1·42]; p=0·35). In the ITT sample (n=755), 226 (60% [95% CI 55–65]) of 376 participants in the mDOT group and 236 (62% [57–67]) of 379 in the patient navigation group had an SVR (AOR 0·92 [0·68–1·25]; p=0·61) and in the per-protocol sample (n=501), 226 (91% [87–94]) of 248 participants in the mDOT group and 235 (93% [89–96]) of 253 in the patient navigation group had an SVR (AOR 0·79 [0·41–1·55]; p=0·44). 306 (81%) of 376 participants in the mDOT group and 317 (84%) of 379 participants in the patient navigation group initiated treatment (AOR 0·86 [0·58–1·26]; p=0·44) and, among those, 251 (82%) participants in the mDOT group and 264 (83%) participants in the patient navigation group completed treatment (AOR 0·90 [0·58–1·39]; p=0·63). Mean daily adherence was higher in the mDOT group (78% [95% CI 75–81]) versus the patient navigation group (73% [70–77]), with a difference of 4·7% ([1·9–7·4]; p=0·0010). 421 serious adverse events were reported (217 in the mDOT group and 204 in the patient navigation group), with the most common being hospital admission (176 in the mDOT group vs 161 in the patient navigation group). Interpretation: In this trial of active PWID, both models resulted in high SVR. Although adherence was significantly higher in the mDOT group versus the patient navigation group, there was no significant difference in SVR between the groups. Increases in adherence and treatment completion were associated with an increased likelihood of SVR. These results suggest that active PWID can reach high SVRs in diverse settings with either mDOT or patient navigation support. Funding: Patient-Centered Outcomes Research Institute, Gilead Sciences, Quest Diagnostics, Monogram Biosciences, and OraSure Technologies.