RESUMO
Background: Retaining clients on antiretroviral therapy (ART) is challenging especially during the first year on ART. Mobile health (mHealth) interventions show promise to close retention gaps. We aimed to assess reach (who received the intervention?) and effectiveness (did it work?) of a hybrid two-way texting (2wT) intervention to improve ART retention at a large public clinic in Lilongwe, Malawi. Methods: Between August 2021 - June 2023, a quasi-experimental study compared outcomes between two cohorts of new ART clients: 1) those opting into 2wT with combined automated, weekly motivation short messaging service (SMS) messages and response-requested appointment reminders; and 2) a matched historical cohort receiving standard of care (SoC). Reach was defined as "the proportion clients ≤6 months of ART initiation eligible for 2wT". 2wT effectiveness was assessed in time-to-event analysis comparing Kaplan-Meier plots of 6- and 12-month retention between 2wT and SoC using a log-rank test. The effect of 2wT on ART drop out was estimated using multivariable Cox proportional hazard models, adjusting for sex, age and WHO stage at ART initiation. Results: Of the 1,146 clients screened, 645 were ineligible (56%) largely due to lack of phone access (393/645; 61%) and illiteracy (149/645; 23%): a reach of 44%. Among 468 2wT participants, the 12-month probability of ART retention was 91% (95%CI: 88% - 93%) compared to 75% (95%CI: 71% - 79%) among 468 SoC participants (p<0.0001). Compared to SoC participants, 2wT participants had a 62% lower hazard of dropping out of ART care at all time points (hazard ratio 0.38, 95% CI: 0.26-0.54; p<0.001). Conclusions: Not all clients were reached with 2wT. For those who opted-in, 2wT reduced drop out throughout the first year on ART and significantly increased 12-month retention. The proactive 2wT approach should be expanded as a complement to other interventions in routine, low-resource settings to improve ART retention.
RESUMO
BACKGROUND: Early retention of people living with HIV (PLHIV) in antiretroviral therapy (ART) programs is critical to improve individual clinical outcomes and viral load suppression. Although many mobile health (mHealth) interventions aim to improve retention in care, there is still lack of evidence on mHealth success or failure, including from patient's perspectives. We describe the human-centered design (HCD) process and assess patient usability and acceptability of a two-way texting (2wT) intervention to improve early retention among new ART initiates at Lighthouse Trust clinic in Lilongwe, Malawi. METHODS: An iterative HCD approach focused on patient and provider users' needs, incorporating feedback from multidisciplinary teams to adapt 2wT for the local, public clinic context. We present mixed-methods usability and acceptability results from 100 participants, 50 at 3-months and 50 at 6-months, post 2wT enrollment, and observations of these same patients completing core tasks of the 2wT system. RESULTS: Among the 100 usability respondents, 95% were satisfied with visit reminders, and 88% would recommend reminders and motivational messages to friends; however, 17% were worried about confidentiality. In observation of participant task completion, 94% were able to successfully confirm visit attendance and 73% could request appointment date change. More participants in 4-6 months group completed tasks correctly compared to 1-3 months group, although not significantly different (78% vs. 66%, p = 0.181). Qualitative results were overwhelmingly positive, but patients did note confusion with transfer reporting and concern that 2wT would not reach patients without mobile phones or with lower literacy. CONCLUSION: The 2wT app for early ART retention appears highly usable and acceptable, hopefully creating a solid foundation for lifelong engagement in care. The HCD approach put the local team central in this process, ensuring that both patients' and Lighthouse's priorities, policies, and practices were forefront in 2wT optimization, raising the likelihood of 2wT success in other routine program contexts.
