Behaviourally-informed two-way text messaging to improve return to HIV care in South Africa: evidence from a randomised controlled trial.

Introduction: One-way text messaging to re-engage people in HIV care has shown promise, but little is known about two-way messaging where the recipient is asked to respond. We evaluated a two-way text messaging intervention, informed by behavioural science, to increase re-engagement in care following missed antiretroviral therapy (ART) visits. Methods: We conducted an individual-level randomised controlled trial between February and March 2023 in Capricorn District, South Africa. Adults aged ≥18 years who had missed an ART visit by >28 days were randomised to (1) a standard one-way text message, or (2) behaviourally informed two-way text messages. Two-way messages asked recipients to reply by selecting from a list of reasons for their missed visit. Those who responded received another text message tailored to their response and designed using behavioural economics principles (affect heuristic, availability heuristic, loss aversion, social norms, providing slack, expanding choice). The primary outcome was return to care within 45 days, analysed as 1) intention to treat, and 2) restricted to successful message delivery. Results: 3,695 participants were randomised: 1,845 to the one-way message group and 1,850 to the two-way message group. 27.9% (515/1845) of participants sent a one-way message and 27.2% (503/1850) sent a two-way message returned for an ART visit within 45 days (proportion difference:-0.7%, p-value: 0.622). In an analysis restricted to participants whose text message was delivered, 28.3% (310/1094) in the one-way message group compared to 28.3% (304/1076) in the two-way message group returned to care (proportion difference:-0.09%, p-value: 0.966). 19.5% (210/1076) responded to the two-way message. The two most commonly reported reasons for missed appointments were being out of town (41.0%) and still having medication (31.0%.). Among those who responded, 27.1% (95%CI: 21.3-33.7) returned for an ART visit. Clients ≥50 years were less likely to respond to the two-way text message, (AOR 0.4; 95%CI: 0.2-0.9). Conclusions: Behaviourally informed two-way text messages did not improve return to care over one-way messages. However, they elicited reasons for disengagement, which could inform future outreach for missed visits. Additional research is needed on the mode, content and timing of two-way messages intended to increase return to care. Clinical Trial Number: PACTR202202748760768 & DOH-27-042022-6703.