Evaluating a smartphone intervention to improve HIV prevention and care outcomes: results from the HealthMPowerment national randomized controlled trial

HealthMPowerment (HMP) is a digital intervention designed to improve HIV-related outcomes among young Black and Latinx men who have sex with men (YBLMSM) and transgender women (TW). We conducted a 12-month randomized controlled trial to evaluate HMP’s impact on routine HIV testing among participants who were HIV-negative or of unknown HIV status and consistent undetectable viral load (CUVL) among people living with HIV (PLHIV). We also examined intervention engagement, acceptability, and implementation fidelity of peer-referral network activation. A total of 750 participants, aged 15 to 29, who identified as YBLMSM or TW and reported condomless anal sex were enrolled in a U.S.-based trial. Participants were stratified by HIV status and randomized to an Information-Only Control Arm or one of two versions of the full-featured HMP app differing in recruitment and clustering strategy (Researcher-Created Networks vs. Peer-Referred Networks, which allowed participants to invite peers into the intervention). After excluding 10 non-randomized peer-referred participants, the primary analytic sample comprised 740 randomized participants. Among 516 randomized participants who were HIV-negative or of unknown HIV status at baseline, routine HIV testing was defined as two or more self-reported HIV tests at least three months apart. Among 224 randomized PLHIV, CUVL at 12 months was assessed using self-reported viral load status. Secondary analyses examined longitudinal viral suppression trajectories using generalized estimating equations (GEE), and implementation analyses described peer-referral activation. Of the 516 randomized HIV-negative or HIV-unknown participants, 16 reported a positive HIV test during the study period. Among the 500 who remained HIV-negative throughout (after removing seroconversions), 89.6% ( n = 448) had ever tested for HIV, and 69.4% ( n = 347) achieved routine testing (two or more tests ≥ 3 months apart) over 12 months. Routine HIV testing did not differ significantly between the Researcher-Created HMP Network Arm and the Information-Only Control Arm (66.0% vs. 72.8%; OR = 0.73, 95% CI: 0.50–1.06, p = .099). Among PLHIV, self-reported CUVL at 12 months was 60.8% in the Information-Only Control Arm, 80.0% in the Researcher-Created HMP Network Arm, and 72.0% in the Peer-Referred HMP Network Arm. In unadjusted endpoint analyses, participants in the Researcher-Created HMP Network Arm had higher odds of CUVL than those in the Information-Only Control Arm (OR = 2.58, 95% CI: 1.24–5.37, p =.011). However, after adjusting for baseline viral suppression, age, and income, study arm was no longer significantly associated with CUVL for either the Researcher-Created HMP Network Arm (adjusted OR = 1.98, 95% CI: 0.84–4.65, p =.117) or the Peer-Referred HMP Network Arm (adjusted OR = 1.44, 95% CI: 0.64–3.24, p =.385). Longitudinal GEE models did not detect significant between-arm differences in viral suppression trajectories over time. Median platform use was modest at approximately 24 min across 12 months. The peer-referral strategy showed limited implementation fidelity, with only 10 non-randomized peer-referred participants enrolled. HMP did not improve routine HIV testing relative to the Information-Only Control Arm. Among PLHIV, unadjusted 12-month CUVL differences favored the Researcher-Created HMP Network Arm, but these differences were no longer significant after adjustment for baseline viral suppression, age, and income, and longitudinal analyses did not show significant between-arm differences over time. These findings suggest that baseline viral suppression largely explained the observed endpoint differences and limit causal interpretation regarding intervention effects. The peer-referral findings may be best interpreted as an implementation fidelity challenge rather than evidence that peer-driven networks are ineffective. Future digital HIV intervention trials should prioritize objective outcome measures, strong implementation fidelity efforts, and lower-barrier strategies for activating digital social support networks. NCT03678181 Registered on September 18, 2018.