Enabling Adherence to Treatment (EAT): a pilot study of a combination intervention to improve HIV treatment outcomes among street-connected individuals in western Kenya

The United Nations Human Rights Council identifies homelessness as a “global human rights crisis,” driven by inequalities in wealth and land access [1]. Although definitions vary, international bodies recognize that homelessness extends beyond not having a house [1‐4]. Homelessness may be included within the broader experience of being street-connected, where as a result of insecure or inadequate housing, the street plays a central role in an individual’s everyday life or identity [1, 5, 6]. Street-connected individuals (SCI) include adults, children, and adolescents who may be unsheltered or inadequately sheltered (sleeping in a temporary or unsafe location), and may be part of a street-based community or economy, for example, for drug use or sex work.

Across varied global contexts, SCI experience profound violations of their right to health and right to life, with high burdens of morbidity and mortality from infectious and chronic disease, mental illness, substance use, and violence [4, 7‐10]. For example, research in high-, middle-, and low-income settings shows high rates of HIV among street-connected adults [4, 8, 11] and youth [7, 12‐14]., In low- and middle-income settings, including sub-Saharan African nations with highly generalized HIV epidemics, drivers of HIV risk among SCI include age, gender, employment, substance use, sexual and gender-based violence, and survival sex (sex in exchange for food, shelter, or other material necessities) [7, 11, 12]. The scope of the challenge to address HIV risks among SCI is enormous: up to 238 million people in sub-Saharan Africa are homeless or inadequately housed, including a large number of children and youth [15].

SCI in sub-Saharan Africa experience social and structural barriers to accessing HIV treatment, including discrimination, stigmatization, criminalization, and poverty [1, 16‐19]. SCI are also likely to experience food insecurity and to lack a safe place to take medications [20‐22]. Thus, it is not surprising that the limited available research suggests that SCI in sub-Saharan Africa have poor HIV treatment outcomes [23]. For example, a 2019 intervention in urban western Kenya that connected SCI under 30 years of age to HIV testing and treatment found that less than one-third of participants achieved viral suppression over two years of follow up [23]. This suggests that even in the context of an intervention to support treatment, the rate of viral suppression among SCI is less than half the rate of viral suppression among the local general population [24]. Given the multiple inequities that limit SCI’s access to healthcare and HIV treatment, SCI are likely to benefit from interventions that combine behavioral and biomedical supports and address structural barriers to HIV treatment [25]. However, the majority of research on improving HIV treatment outcomes among SCI is based in high-income settings [26, 27]. There is an urgent need for evidence to guide the design and implementation of interventions that improve HIV treatment outcomes among SCI in sub-Saharan Africa.

One potential strategy for supporting HIV treatment among SCI in sub-Saharan Africa involves modified Directly Observed Therapy (mDOT). mDOT is an intervention originally developed to promote adherence to tuberculosis treatment, where all or some doses of medication are taken under direct supervision by a healthcare professional [28]. mDOT interventions have been adapted to support adherence to antiretroviral therapy (ART) for HIV, and have been shown to be particularly effective in populations with high rates of non-adherence [28, 29]. In addition to supervised dosing, mDOT may address other barriers to ART adherence by providing SCI with a safe care environment, a secure place to store medications, and facilitating regular contact with healthcare providers. mDOT has also been combined with other evidence-based strategies to support adherence to treatment in resource-constrained settings, including peer support and food [29]. An mDOT intervention has not yet been piloted with SCI in an urban setting in Kenya.

This pilot study describes the design, implementation, costs, and HIV outcomes of the Enabling Adherence to Treatment (EAT) intervention, a combination intervention providing mDOT and a daily meal alongside pre-existing peer navigation services, to SCI in an urban setting in western Kenya. The goal of this pilot study is to provide proof of concept for the use of the EAT intervention to support engagement in HIV care and improve rates of viral suppression among SCI in urban settings in western Kenya.

Our pilot study provides proof of concept that EAT, a combination intervention providing mDOT, food, and peer navigation services, was feasible to implement among SCI in an urban setting in Kenya and can support SCI’s engagement in HIV care and achievement of viral suppression. We also identified factors to guide implementation of trials of EAT at a larger scale or in different contexts. The estimated cost for 19 months of the EAT intervention was $57,448.64 USD.

Our study demonstrates that EAT can feasibly be implemented with this unique population and within the existing HIV care infrastructure, and provides preliminary evidence that EAT can support SCI’s engagement in HIV care and achievement of viral suppression. This pilot study was not designed to detect significant changes in HIV treatment outcomes. However, we observed that by the end of follow up all participants had initiated ART; the proportion of participants active in HIV care increased from 59% to 87%; and the proportion of participants who met the AMPATH definition of virally suppressed increased compared to the 19 months before the EAT intervention, from 33% to 59%. Among participants who were enrolled in care prior to EAT baseline, there were statistically significant increases in the proportion who were active in care at AMPATH and who had a viral load measure at end of follow up compared to baseline. There was also a statistically significant increase in the proportion of participants who were virally suppressed at their last viral load measure within the 19 months of EAT follow up compared to within the 19 months before EAT enrollment. The proportions of participants who were active in care and initiated ART are comparable to results of other interventions among populations at high risk of HIV in Kenya, for example, a program for female sex workers living with HIV where by end of follow up, 79% of participants were engaged in care and 73% were on ART [43]. Although our sample size is small, the proportion of participants virally suppressed at their last viral load test is comparable to published rates of viral suppression among adults and children in various settings across Kenya (ranging from 39.7% among a rural residents of another western county, to 51% and 60% respectively among children and adults in Uasin Gishu) [24, 44‐46]. The proportion of participants virally suppressed after EAT is also similar to other interventions trialed among other sub-Saharan African populations at high risk of HIV, including female sex workers, [47] and adolescents [48, 49]. Although our findings suggest that participants who enrolled in HIV care concurrently with EAT had a statistically significantly shorter time to initiation of ART and to the first episode of viral suppression compared to participants who enrolled in HIV care before EAT, we cannot attribute this change to EAT alone, because guidelines for initiating ART have changed over time and we did not control for heterogeneity between the two groups. Additionally, EAT’s preliminary successes may be partly attributable to its unique context, embedded in AMPATH and linked to an established peer navigator program and adolescent health centre [23, 50]. However, based on our evidence that EAT can feasibly be implemented in this unique population and shows promise in improving HIV outcomes, future research should include controlled trials powered to evaluate whether EAT can improve adherence to ART and viral suppression and how EAT performs relative to other interventions.

One major goal of this proof of concept study was to identify implementation challenges or unexpected factors that could guide implementation of future trials. First, we found that although most participants were on drug regimens requiring at least daily dosing, most participants did not adhere to the dosing schedules assigned at their first visit. Irregular visit timing may be related to competing demands on SCI’s time, including work, [51] or barriers like costs of transportation. Some SCI may not need or want daily meals or treatment observation. Improvements in rates of viral suppression during EAT even though many participants did not attend daily may reflect success of the other components of EAT that supported adherence, like follow up from PN after a missed medication refill. These findings are in line with evidence that interventions using non-daily visits to encourage adherence and interventions using PN can improve HIV treatment outcomes [29, 52]. Future studies could focus on exploring how different combinations of elements in the EAT intervention function to meet participants’ needs. We also recommend that future trials of EAT explore whether (or for whom) a daily, weekly, or drop-in mDOT schedule would be most effective, and fund participants’ transport to reduce barriers to attendance. Finally, although EAT was originally conceived as an intervention for street-connected youth and eligibility was extended to adults for ethical and practical reasons, over half of the participants living with HIV in this pilot study were above the age of 30. Previous studies in our setting have focused on street-connected youth, [12, 53, 54] and the high number of adults who enrolled was unexpected, indicating a need for more research on HIV among adults who self-identify as street-connected in Eldoret.

The total cost of EAT over 19 months was $57,448.64. Fixed costs were $3623.04 USD. Variable costs were $2677.62 per month and $63.75 per participant per month, estimated using the median number of participants who visited at least once per month. Published cost estimates for non-food interventions that support ART adherence in low-resource settings range widely, from $33.00 USD for a 6-week mDOT intervention to $14.75 USD per visit for a peer-based intervention [52]. Cost estimates for nutrition interventions also vary widely, from $15-$50 per participant per month for monthly ration baskets or food dispersion [55‐58]. Costs in our study may be comparatively high because EAT combined behavioral, biomedical, and structural adherence supports. Our operating costs also include some food waste, because food orders were fixed at 25 servings per day, while the number of attendees was sometimes below 25. Operating costs could be reduced by reducing food waste, achieving economies of scale, or outside of a research context where some staffing costs would be eliminated. New literature suggests that nutrition supplementation or DOT may only be cost-effective in unique populations at high risk of severe malnutrition or nonadherence to ART [59‐61], which may support the use of EAT among SCI. Future trials of EAT powered to detect statistically significant changes in HIV outcomes should be paired with cost-effectiveness evaluations that account for a more complete scope of costs (including overhead costs, drug costs, and downstream healthcare costs and savings) and compare the costs of EAT to other interventions or treatment as usual for SCI.

This pilot study of the EAT intervention has a number of limitations. First, this pilot study was not powered to detect significant differences in outcomes pre- and post- intervention, and not designed assess the effectiveness of EAT compared to a control group. We did not include a direct measure of ART adherence and regular viral load measures were not required. SCI are highly transient, so our ability to follow up with participants and maintain consistent attendance was limited, although improved by the use of PN. Participants also self-identified as street-connected, so it is possible some participants did not genuinely meet our eligibility criteria for duration of street-involvement, however, ethically and practically we were limited in our ability to verify the exact duration of street involvement, and this limitation was mitigated by use of PN who know the street community well. Interpretation of our inferential statistical analyses are limited because we did not adjust for confounding, and furthermore, interpretation of differences in median time to initiation ART and to the first episode of viral suppression are limited by the fact that guidelines around initiating ART have changed over the years. Monthly operating cost estimation per participant may underestimate the true cost per participant because the actual number of participants who attended EAT every day is lower than the median number with at least one visit per month. Furthermore, we do not compare the costs of EAT to costs of any other intervention, so this study in isolation cannot guide decision making around EAT’s relative cost-effectiveness. The pilot study also has a number of strengths. Investigators were able to draw on long-standing relationships with the community of SCI in Eldoret, which likely supported engagement. It showed that the EAT intervention was feasible within the existing infrastructure, which suggests possibilities for further study and scale. Potential sources of funding for future trials of EAT might include NGOs, local governments and healthcare agencies, and research funding bodies. EAT also offers opportunities for novel partnerships, for example, between healthcare and local food or agricultural businesses who could supply meals. Anecdotally, when the pilot ended, the restaurant which operated beside the Rafiki Center voluntarily continued to provide free meals to SCI attending the clinic for ART because they observed how much it helped to stabilize and support the participating SCI. (Unfortunately, about one month later, in March 2020, the restaurant closed because of the COVID-19 pandemic.) Ultimately, EAT, if scaled, may help support SCI’s human right to an adequate standard of living, including adequate food and medical care, as set out by the Universal Declaration of Human Rights [62].

Our pilot study provided proof of concept that EAT, a combination intervention providing mDOT, food, and peer navigation services, was feasible to implement and may support engagement in HIV care and achievement of viral suppression among SCI living with HIV in an urban setting in Kenya. Future EAT trials should consider a adopting a non-daily schedule for mDOT and funding participants’ travel to reduce barriers to attendance. Future research should include controlled trials powered to assess whether EAT can improve adherence to ART and viral suppression, how EAT performs relative to other interventions, and whether EAT can be feasibly implemented with SCI in other settings.