Tracing the journey: Patient engagement in long-term antiretroviral therapy care in Malawi

Introduction Retention in antiretroviral therapy (ART) care is essential for ensuring sustained viral suppression, improving health outcomes, and preventing the onward transmission of HIV [–]. ART retention, however, is challenging, particularly in low- and middle-income countries (LMIC) [–]. People living with HIV (PLHIV) in LMICs often face well-documented structural and individual barriers to HIV care, such as economic constraints, transportation difficulties, stigma, and mental health [–]. Previous studies reported that early disengagement, lack of retention within the first six months on ART, is particularly predictive of longer-term disengagement and loss to follow-up (LTFU) [,]. Traditionally, ART retention, defined as actively receiving ART, has been assessed cross-sectionally at 6, 12 or 24 months after ART initiation. At these time points, PLHIV are categorised as either ‘retained’ or ‘not retained’ [–]. This time-specific measure of ART retention has significantly shaped global policies but fails to capture the complexities of client behaviour between these time points. ART engagement is not uniform, but dynamic []. PLHIV frequently cycle in and out of care as they navigate changing personal, social, and economic circumstances []. ART engagement differs across times of ART follow-up and by client demographic groups [,]. Understanding ART engagement patterns - given that PLHIV frequently cycle in and out of care - is crucial for refining interventions and developing additional strategies to improve and sustain long-term retention. With the scale-up of electronic medical records systems (EMRS) which provide comprehensive, longitudinal data on PLHIV receiving ART, there is an opportunity to map client retention and ART engagement patterns in greater detail []. The Lighthouse Trust (LT) is a local non-governmental organisation providing HIV services and offers various support services to promote retention at its ART clinics, Lighthouse (LH) and Martin Preuss Centre (MPC), in Lilongwe, Malawi. These retention support services are provided by ART “Buddies”, two-way texting (2wT) and the Back-To-Care (B2C) program. ART Buddies are PLHIV who offer peer social support, encouragement, and reminders for upcoming and missed visits during the first 12 months of ART and are considered the standard of care (SoC) since 2019. 2wT is a hybrid intervention introduced in 2021 and combines an automated weekly blast of a non-HIV-related motivational message and specific, interactive, response-requested, ART visit reminders []. Preliminary findings demonstrated 2wT is effective, with a 91% ART retention rate at 12-months compared to 76% with SoC buddy support. Finally, B2C is a reactive ART retention support program offered to all clients as part of SoC (buddy and 2wT) in which clients who miss appointments by 14 days or more are traced by retention officers through phone and, if needed, home visit []. The objectives of this study were: 1) Describe characteristics of ART clients and ART program outcomes by type of retention support: Buddy (differentiating between clients with and without phone access) and 2wT; 2) Describe on-time clinic visit attendance over the first 24 months on ART; 3) Visualise ART engagement patterns over 24 months to understand how ART clients cycle in and out of care; and 4) Assess ART engagement patterns by sex, age group, and type of retention support at six-month intervals following ART initiation. intro sec005 Methods Study design This retrospective descriptive cohort study used routine program data from the LT clinics (LH and MPC) in Lilongwe, Malawi. The study included all ART clients who initiated ART between 01 January 2020 and 31 December 2021, allowing a 24-month ART follow-up period by 31 December 2023. sec007 Settings LT clinics use point-of-care (PoC) EMRS for client management []. Clients diagnosed with HIV are registered in the PoC EMRS and referred to an ART Buddy for pre-ART initiation counselling and psychosocial support. Clients are then assessed to determine their WHO HIV clinical stage. ART appointments are typically scheduled monthly for the first three months. After six months, appointments are scheduled every three or six months, depending on the client’s clinical condition. In addition to their standard ART supply, clients receive an additional two-days’ worth of ART for each appointment to prevent treatment interruptions. Clients transferring to LT clinics from other ART facilities are also registered in the EMRS and receive the same ART and adherence support. MoH ART outcomes are 1) Retained on ART, for clients who continue treatment; 2) LTFU, for clients who miss clinic appointments for at least 60 days; 3) Stopped ART treatment, for clients who discontinue treatment on their own volition or by ART provider’s decisions on account of suspected treatment failure and adverse events; 4) Transfer-out, for clients who moved their ART care to another ART facility; and 5) Died, for clients who have passed away due to any cause. sec008 ART retention support Following pre-ART counselling, clients are offered the choice between early retention support by SoC Buddies or through the 2wT platform, as previously described []. Phone ownership and receipt of a confirmation registration message are a prerequisite for 2wT participation. Clients who opt for SoC Buddies receive phone calls reminding them of clinic appointments and in-person adherence counselling during clinic visits in the first 12 months of ART. Clients who opt for 2wT support receive personalised visit reminders on days 3 and 1 before appointments and on days 2, 5 or 11 after missed appointment visits, if applicable. Clients are asked to respond to visit reminders with “yes” or “no” to confirm attendance and may reschedule an appointment within 13 days. Additionally, the 2wT-supported clients receive weekly motivational messages on various health topics. Clients can interact with a 2wT officer via SMS during routine clinic hours. Both Buddy-supported and 2wT-supported clients who miss appointments for 14 days or more are followed up by the B2C team via phone or home visits. During the study, the 2wT retention support was offered for up to 24 months post-ART initiation. sec009 Study definitions Clients were divided into 3 ART retention support groups: Buddy with phone (client had a recorded mobile phone number in the EMRs); Buddy without phone (client did not have a recorded mobile number in the EMRs), and 2wT (mobile phone required and used for 2wT enrolment). We defined three outcome measures: On-time clinic visit attendance, retention and ART engagement patterns. In addition to assessing interval-level engagement, we determined an overall engagement pattern for each client over the full 24-month period using the same definitions. For instance, a client who attended all scheduled visits within 13 days of the scheduled date was considered “continuously engaged” throughout the first 24 months of care. a : A visit-level outcome defined as the proportion of clients with a scheduled ART refill appointment who attended the appointment within 13 days of the scheduled date. Clients often had multiple visits per 6-month interval. b was defined as the proportion of clients retained on ART at 24-month post-initiation among those who initiated ART between 01 January 2020 and 31 December 2021. Follow-up time during the first 24 months on ART was divided into six-month intervals (>0–6, > 6–12, > 12–18, and >18–24) to assess temporal patterns in engagement. Each six-month interval was treated as an independent observation period. Interval-specific retention was defined as the number of clients who received ART during a given interval and were retained on ART, divided by the number of clients who received ART during the same interval. Clients who transferred out, stopped ART treatment, or were LTFU were not censored within the interval and were eligible to re-enter the cohort in subsequent intervals if they returned to care. c were assessed to deepen understanding of client behaviours of cycling in and out of care. ART engagement patterns were derived from client’s MoH ART program outcome at the end of each six-month interval and within-interval appointment attendance patterns. Clients who disengaged (MoH ART outcomes: LTFU, transferred out, stopped ART treatment, or died) from ART care at a facility at the end of any interval were assigned that outcome for that entire interval. Clients with the MoH ART outcome “retained on ART” at the end of any interval (consistent with the definition of retention above) were further categorised based on their within-interval appointment attendance pattern. These clients were considered: 1) Continuously engaged if they attended all appointments within 13 days of the scheduled date; 2) Cyclically engaged if they returned late (between 14 and 59 days) for at least one appointment in the interval; 3) Re-engaged, if they missed at least one appointment by 60 days or more in the interval but returned to care before the end of that six month interval. Clients who were LTFU, transferred out or stopped ART treatment were not censored at the end of the six-month interval to allow for potential returning to care. simple sec010 Statistical analysis Study data from the PoC EMRS at LH and MPC were extracted between 10 March and 20 May 2024. Data were analysed using Stata version 18. Descriptive statistics (frequencies and medians with interquartile range (IQR)) were used to describe the study population. On-time clinic visit attendance was assessed overall and monthly. ART retention and engagement pattern proportions were presented for each six-month interval, stratified by sex, age group, and type of retention support: Buddy support with or without phone and 2wT. We used Z-test for proportions and chi-square tests for trends to explore differences and trends in MoH ART outcomes, retention and ART engagement patterns across demographic and ART retention support groups. A 95% confidence interval and p-values were presented to aid interpretation of observed differences. ART engagement patterns were visualised using a Sankey chart, illustrating trajectories for those retained on ART (continuously engaged, cyclically engaged, re-engaged) and those who disengaged (LTFU, stopped ART, transferred-out, died) at 6, 12, 18, and 24 months after ART initiation. To enhance visualisations simplicity and interpretability, two Sankey charts were created based on a qualitative segmentation approach: one for common patterns illustrated by ≥21 clients and another for rare patterns with 20 or fewer clients. Calculations of common pattern counts were based on the entire follow-up period for each client, including clients who disengaged 24 months before. sec011 Ethical considerations Ethical approval was obtained from the Malawi National Health Sciences Research Committee (Protocol #23/10/4258). As this study used routine program data without client identifiers, participant informed consent was waived. sec012 materials|methods sec006 Results Study population A total of 6,990 PLHIV initiated ART between 1 January 2020 and 31 December 2021 at LH and MPC. Of them, 687 (10%) were excluded from analysis because they had received ART for six months or more at other sites prior to enrolment in care at LH and MPC. The remaining 6,303 (90%) were included: 1,030 (16%) in the Buddy support group with phone, 4,850 (77%) in the Buddy support group without phone, and 423 (7%) in the 2wT support group (phone required) (). Of those included, 58% were female and 42% male. Sex, age group at ART initiation, on-time clinic visit attendance and MoH ART outcomes at 24 months (χtest, p 6–12 months left 1 1 left 1 1 left 1 1 left 1 1 Total patients 4,565 1,944 2,621 Retained on ART 3,831 (84%) 1,617 (83%) 2,214 (84%) Continuously engaged 2,977 (78%) 1,261 (78%) 1,716 (78%) Cyclically engaged 615 (16%) 250 (15%) 365 (16%) Re-engaged 239 (6%) 106 (7%) 133 (6%) LTFU 466 (10%) 216 (11%) 250 (10%) Stopped ART 59 (1%) 32 (2%) 27 (1%) Transferred out 186 (4%) 65 (3%) 121 (5%) Dead 23 (1%) 14 (1%) 9 (0%) >12–18 months left 1 1 left 1 1 left 1 1 left 1 1 Total patients 4,046 1,722 2,324 Retained on ART 3,586 (89%) 1,542 (90%) 2,044 (88%) Continuously engaged 2,843 (79%) 1,204 (78%) 1,639 (80%) Cyclically engaged 513 (14%) 228 (15%) 285 (14%) Re-engaged 230 (6%) 110 (7%) 120 (6%) LTFU 283 (7%) 117 (7%) 166 (7%) Stopped ART 40 (1%) 15 (1%) 25 (1%) Transferred out 120 (3%) 42 (2%) 78 (3%) Dead 17 (0%) 6 (0%) 11 (0%) >18–24 months left 1 1 left 1 1 left 1 1 left 1 1 Total patients left 1 1 3,775 1,629 2,146 Retained on ART 3,367 (89%) 1,461 (90%) 1,906 (89%) Continuously engaged 2,647 (79%) 1,160 (79%) 1,487 (78%) Cyclically engaged 521 (15%) 216 (15%) 305 (16%) Re-engaged 199 (6%) 85 (6%) 114 (6%) LTFU 267 (7%) 108 (7%) 159 (7%) Stopped ART 24 (1%) 9 (1%) 15 (1%) Transferred out 101 (3%) 41 (3%) 60 (3%) Dead 16 (0%) 10 (1%) 6 (0%) hsides groups ‡ All outcomes are assessed at the end of each six-month interval. Total patients refers to clients with at least one ART visit during the 6-month interval; Statistical analysis: Pairwise comparisons were conducted using 95% confidence intervals from Z-tests with, (*) indicates statistically significant results (p 6–12 months, 89% at >12–18 months, and 94% at >18–24 months, compared to 81%,81%, 81%, 80% in ages 35–49 years and 74%, 74%, 76% and 75% in ages 18–34 years, respectively (χfor trend, all p-values 6–12 months left 2 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 2,442 1,813 310 left 1 1 left 1 1 Retained on ART 1,972 (81%) 1,577 (87%) 282 (91%) 0.058 left 1 1 Continuously engaged 1,467 (74%) 1,271 (81%) 239 (85%) 12–18 months left 2 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 2,113 1,640 293 left 1 1 left 1 1 Retained on ART 1,811 (86%) 1,500 (91%) 275 (94%) 0.15 left 1 1 Continuously engaged 1,381 (76%) 1,218 (81%) 244 (89%) 0.002 left 1 1 Cyclically engaged 279 (15%) 213 (14%) 21 (8%) 0.065 left 1 1 Re-engaged 151 (8%) 69 (5%) 10 (4%) 18–24 months left 2 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 1,929 1,567 279 left 1 1 left 1 1 Retained on ART 1,674 (87%) 1,431 (91%) 262 (94%) 0.235 left 1 1 Continuously engaged 1,260 (75%) 1,151 (80%) 236 (90%) 0.002 left 1 1 Cyclically engaged 282 (17%) 217 (15%) 22 (8%) 0.033 left 1 1 Re-engaged 132 (8%) 63 (4%) 4 (2%) 6–12 months, and 86% at >12–18 months), compared to the Buddy support group with phone (79%, 79%,80%) and without phone (66%, 62%,69%) (z-test, all p-values 6–12 months, as compared to LTFU rates in those same two periods in the Buddy support group with phones (15% and 9%) and 2wT group (8% to 4%) (z-test, p-values 6–12 months left 1 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 537 left 1 1 3,663 365 left 1 1 left 1 1 Retained on ART 366 (68%) 3,131 (85%)** 334 (92%)** 12–18 months left 1 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 397 left 1 1 3,304 345 left 1 1 left 1 1 Retained on ART 318 (80%) 2,955 (89%)** 313 (91%) 18–24 months left 1 1 left 1 1 left 1 1 left 1 1 left 1 1 Total patients 344 left 1 1 3,107 324 left 1 1 left 1 1 Retained on ART 278 (81%) 2,793 (90%)** 296 (91%) <0.001 Continuously engaged 197 (71%) 2,205 (79%)** 245 (83%) left 1 1 Cyclically engaged 54 (19%) 430 (15%) 37 (12%) left 1 1 Re-engaged 27 (10%) 158 (6%) 14 (5%) left 1 1 LTFU 43 (12%) 208 (7%) 16 (5%) left 1 1 Stopped ART 4 (1%) 19 (1%) 1 (0%) left 1 1 Transferred out 17 (5%) 74 (2%) 10 (3%) left 1 1 Dead 2 (1%) 13 (0%) 1 (0%) left 1 1 hsides groups βBuddy support with phone access: Defined as having a phone number recorded in the Electronic Medical Records (EMR) system.All outcomes are assessed at the end of each six-month interval. Total patients refers to clients with at least one ART visit during the interval.Statistical analysis: χ² tests were used to examine associations between types of retention support; Pairwise comparisons were conducted using 95% confidence intervals from Z-tests with, (*) indicates statistically significant results. Comparison groups: Buddy support without phone access vs. buddy support with phone access; Buddy support with phone access vs. two-way texting support. Denominator (total patients) is the number of clients who received ART during a specific six-month interval. t004fn001 float pone.0323983.t004 portrait sec020 sec017 results sec013 Discussion Understanding ART engagement patterns among PLHIV is vital for viral load suppression and resource allocation. This study describes demographics, on-time clinic visit attendance, retention and ART engagement patterns during the first 24 months of ART among clients receiving ART at two large, public clinics in Malawi. Overall, 53% of clients were retained at 24 months post-ART initiation, with the highest retention observed in the 2wT support group (70%) compared to those in the Buddy support group with phone (58%) and without phone (27%). Only 41% attended all clinic visits on time, with the lowest attendance among those without phones (21%). Of all clients, 47% disengaged from care (became LTFU, stopped ART treatment, transferred out and died), with half of them being LTFU. Most clients (84%) followed a limited set of common engagement patterns, while 16% had more varied, infrequently observed patterns. Only 29% of the total cohort remained continuously engaged throughout the 24-month period. Clients aged 50 + had the highest proportion of continuous engagement. Cyclically engaged was most common among those without a phone. We discuss several implications of these findings on practice and policy. First, this study highlights the complexity of clients’ ART engagement behaviours, with significant variability over the first two years of ART. Engagement patterns varied across six-month intervals. While most clients (84%) demonstrated common engagement trajectories with minimal variability over time, a smaller group (16%) had varied and infrequent patterns. These findings highlight the potential for a differentiated approach to retention interventions, tailoring support intensity based on clients’ evolving needs. During the early phase of ART (0–6 months), a universal intensive support package – including enhanced counselling, Buddy or 2wT support – can be implemented, as early retention is critical for long-term ART adherence. Beyond six months, once client’s engagement patterns become apparent, a stratified strategy to retention support could be considered. Shifting from a uniform to a dynamic, data-informed retention approach has the potential to improve efficiency and retention outcomes; however further research is needed to evaluate the effectiveness and feasibility of such approaches. Second, our study suggests that clients with phones had better ART engagement and on-time clinic visit attendance than clients without phones. 2wT participants received motivational messages, two visit reminder messages before each visit, and up to three additional messages if they missed an appointment []. Likewise, although not all ART buddies were able to remind clients of upcoming visits, ART retention buddies contacted their clients after missed visits. Regardless of contact method, clients with phone access to 2wT or buddies had higher engagement and more timely clinic visits, likely reflecting early identification of retention problems and swifter response to follow-up efforts. The observed disparity in retention by phone status may also be associated with socioeconomic and psychosocial differences between the retention groups. ART clients with phones tend to have higher education levels, greater economic ability and better health literacy [–], all of which can improve the ability to engage with healthcare services []. Additionally, those who provided their phone numbers may also be more likely to have disclosed their status while those without phone numbers may face more stigma or fear violence. Third, demographics still matter. Older ART clients showed more stable engagement compared to younger clients, a finding consistent with other studies []. Although the reasons are not fully understood, higher retention in older adults could reflect established routines, fewer competing priorities, and stronger health-seeking behaviours. These findings highlight the importance of age-tailored interventions to improve engagement in younger populations. Also, as there are more female clients, yet more men than women have phones, both proactive and reactive retention efforts must address the persistent gender dimension to engagement and retention in care. Fourth, while cyclically engaged or re-engaged patterns are associated with worse clinical outcomes, including risks of treatment failure and unsuppressed viral loads [], not all cyclical engagement and re-engagement instances result from treatment interruptions. Some clients may obtain ART from other facilities while they travel or temporarily relocate for work [], underscoring the need for improved communication with clients, better documentation of emergency ART provision and simplified transfer processes between clinics. Lastly, as suggested by previous studies [,], the interval-based engagement indicators capture ART continuity more effectively than standard fixed-point retention measures. Unlike traditional program retention indicators, which only measure whether clients are in care at a single point in time, this approach identifies patterns of continuous engagement, cyclical attendance, and re-engagement after gaps in care. All clients who return to care are included in the interval in which they receive ART, ensuring that no participants are excluded from the analysis. This approach is particularly relevant for monitoring progress toward the 95-95-95 targets, as sustained engagement and continuity of care are essential for achieving viral suppression. Incorporating interval-specific engagement indicators into routine program monitoring, especially where EMR systems are available, can provide a more accurate picture of client adherence, identify periods of potential disengagement, and support targeted interventions to maintain long-term ART continuity. Strengths and limitations This study is among the few to describe retention as a dynamic and evolving ART engagement, reasserting that client retention is not a binary state but a complex process. Differentiating client engagement patterns by type of retention support and phone access contributes to novel insights that can inform future retention service delivery strategies. Our study also had limitations. First, using a phone number as a proxy for phone access can be problematic for several reasons. Although phone numbers were confirmed at 2wT enrollment, some became inactive or did not work consistently throughout the study. Similarly, clients in SoC buddy support may have changed numbers during the study, and those categorised as without phones at enrolment may have obtained a phone later or may have had phone access later but did not share a number due to privacy concerns. These phone-based factors could have caused errors in client categorisation, potentially affecting observed ART engagement patterns. Second, clients were limited to those receiving care at two large urban sites, so observed cyclical and re-engagement patterns may not be representative of those found in smaller, rural or peri-urban clinics. Third, while utilising a large study population offers several advantages, some statistically significant differences of less than 5% between groups may not be meaningful for practice or policy change. Lastly, appointment rescheduling of fewer than 13 days was not accounted for in the classification of engagement patterns. This may have led to misclassification of some continuously engaged clients as having cyclical engagement patterns. sec022 conclusions sec021 Conclusion ART engagement among study clients was dynamic and heterogeneous over time. The majority of ART clients followed a consistent and stable retention trajectory, with minimal variability. Engagement patterns varied by retention support type: clients who opted into 2wT showed higher levels of continuous engagement across retention periods. Strikingly, the finding that only 29% of clients were continuously engaged during the first 24 months underscores a substantial gap in continuous ART retention. These findings support the urgent need for further research to better identify clients at higher risk of disengagement and to explore tailored retention strategies that best match clients’ changing needs over time in ART care. conclusions sec023