Scaling an intervention for the engagement of people with HIV who inject drugs into care in Viet Nam: an implementation-effectiveness, cluster-randomised trial

Abstract

Background: Governmental agencies often implement evidence-based interventions with a one-size-fits-all approach, which ignores site differences. We compared a tailored approach, accounting for sites' needs, to a standard approach to implement systems navigation and psychosocial counselling (SNaP) to improve HIV outcomes for people who inject drugs (PWID).

Methods: We conducted a hybrid type 3 implementation-effectiveness, cluster-randomised trial at HIV testing sites in ten provinces in Viet Nam. SNaP comprised two systems navigation sessions, in person or by telephone, and at least one psychosocial counselling session. Sites were randomised (1:1) to a standard approach (a package of 15 implementation strategies) or a tailored approach (standard strategies plus a menu of ten additional strategies). Study participants included navigators and counsellors of SNaP, site directors, other staff, and PWID with HIV. Sites were chosen based on the number of HIV tests performed, HIV positivity rate, and the size of the PWID population covered by the site. PWID with HIV were aged 18 years or older, not on antiretroviral therapy (ART) at enrolment, and had used injection drugs in the previous 6 months. The primary outcomes were fidelity (implementation outcome) and ART initiation (effectiveness outcome); secondary effectiveness outcomes were the current use of ART and medications for opioid use disorder and viral suppression. Analyses were conducted with generalised linear models or generalised estimating equations. The study was registered with ClinicalTrials.gov (NCT03952520) and is complete.

Findings: On Feb 27, 2020, 42 testing sites were randomised to two approaches (tailored: n=21; standard: n=21), and participants were recruited to the trial between May 8, 2020, and Aug 17, 2022, resulting in the inclusion of 277 staff (tailored: n=141; standard: n=136) and 723 PWID with HIV (tailored: n=364; standard: n=359). Clinic fidelity scores (range 0-200) were higher with the tailored approach (mean 151·4 [SD 21·4]) than the standard approach (114·1 [13·5]; adjusted mean difference 37·3 points [95% CI 26·5-48·1]). Nearly all follow-up participants receiving SNaP initiated ART (tailored: 345 [99%] of 347; standard: 334 [99%] of 338; adjusted risk difference 1·3 percentage points [95% CI 0·7-1·9]; intraclass correlation coefficient -0·01). Viral suppression among PWID was substantially higher in the tailored approach arm (adjusted prevalence difference 15·8% [95% CI 5·0-26·5]); no other secondary effectiveness outcomes were significant. There were 63 (9%) deaths among 685 follow-up participants, with 14 (2%) dying before baseline assessment.

Interpretation: A tailored approach with rapid needs assessment before and during implementation to address local testing site barriers in Viet Nam improved fidelity of SNaP delivery. The tailored approach had a small effect on ART initiation; viral suppression improved substantially.

Funding: US National Institutes of Health, US National Institute of Drug Abuse, and US National Institute for Allergy and Infectious Diseases.

Figure 1:. Trial profile

PWID with HIV were enrolled to assess effectiveness outcomes of ART initiation (primary outcome), and ART use, viral suppression, and MOUD use (secondary outcomes). The apparent attrition in cohort follow-up is primarily due to the study’s design: not all participants had the opportunity for multiple follow-up assessments because the duration of a participant’s follow-up was based on the timing of trial initiation within each clinic rather than the individual participant (appendix 2 p 8). ART=antiretroviral therapy. MOUD=medications for opioid use disorder. PWID=people who inject drugs.

Figure 2:. Fidelity score components, by implementation approach

The fidelity score for each clinic was calculated as the sum of the systems navigation score and the psychosocial counselling score (100 points each for a total of 200 points). Each of these component scores was based on the completion of the expected sessions (two systems navigation sessions per participant within 8 weeks of enrolment; one psychosocial session per participant within 4 weeks of enrolment) and the quality of those sessions, as established by audio-recording review.