Introduction

Refugees are frequently characterized as a mobile population [1]. The term ‘refugee’ itself implies movement: fleeing from conflict or persecution in one country and crossing an international border to find safety in another [2]. Much of the research that has been conducted on refugee mobility has focused on mobility in transit or mobility after permanent or semi-permanent relocation in the Global North [3, 4, 5]. Less is known about the mobility patterns of refugees after they have been granted asylum in the Global South.

Sub-Saharan Africa is home to 7.1 million refugees, or 20% of the global refugee population [6]. In sub-Saharan Africa, mobility is closely linked to income generation and constitutes an essential livelihood strategy. For refugee populations who often face food scarcity, have limited livelihood options available to them, and are more vulnerable to economic shocks, the economic drivers of mobility are likely amplified. Social drivers of mobility may also be heightened for refugees as a result of disrupted social networks and separated families. The close proximity of host countries to countries of origin in sub-Saharan Africa, coupled with the ease of movement across borders, may facilitate refugee mobility between countries [7].

Refugee mobility is important in the context of HIV. Research among non-refugee populations demonstrates that mobility is associated with lower uptake of HIV testing [8], an increased likelihood of loss to follow-up or disengagement from care [9, 10, 11, 12], a higher probability of antiretroviral therapy (ART) non-adherence [13, 14], and lower rates of HIV viral suppression [15, 16]. A recent publication by Thorp et al. highlighted the need to unpack the concept of mobility and gain a better understanding of the types of mobility that most impact HIV outcomes, as well as the demographic profiles associated with high-risk mobility [17]. As conflict and violence continue to rise globally and the impacts of climate change worsen, forced migration will become more prevalent. Against this backdrop, understanding how mobility impacts HIV outcomes will be vital to the optimization of HIV programs in refugee settlements.

In this study, we aimed to characterize the mobility patterns of refugees and Ugandan nationals living with HIV accessing HIV services in refugee settlements in Uganda and to evaluate the association between mobility and retention in HIV care.

Methods

Study Design and implementation

Participants

Refugees and Ugandan nationals ≥18 years of age, living with HIV (either newly testing positive for HIV or previously enrolled in HIV care), accessing HIV services at participating health centers, with access to a phone were eligible to participate. Eligible participants were recruited at the time of HIV testing, during routine HIV follow-up appointments, or by phone using contact information listed in clinic HIV registers.

Data collection

Eligible individuals interested in participating were read a consent form in their language of choice (Kiswahili, Kinyarwanda, Runyankore, Somali, Arabic, Kakwa, Dinka, Lugbara, or English) by a research assistant with the help of an interpreter when necessary. Following written informed consent, contact information, socio-demographic data (sex, age, refugee status, relationship status, education, time living in settlement), and mobility history (travel in the past year, reasons for travel, relocation frequency) were collected in an intake survey administered verbally by the research assistant, with answers entered directly into a mobile REDCap database with offline capabilities. Following the intake survey, all participants who had recently tested positive for HIV were linked to the HIV clinic to initiate ART per standard protocol.

Starting one month after enrollment, each month for five months, participants received a pre-recorded interactive voice response (IVR) phone survey in their study language of choice (Figure 1). Participants were able to respond to multiple-choice survey questions by pressing numbers on their telephone keypad. They were asked questions about mobility in the past four weeks (10 questions) and questions about HIV perceptions and their HIV care (10 questions) (complete survey in Appendix 1). Various aspects of mobility were captured in the survey, including change in place of residence, number of nights spent away from home, distance of travel (outside country > district > subcounty > parish > village), frequency of travel, and details regarding their most recent trip. Participants indicating they did not travel in the last month did not receive any subsequent survey questions. To prevent unintentional HIV disclosure to individuals other than the participant in the call, participants were asked to enter the three-digit study passcode that was given to them at enrollment before gaining access to the second half of the survey, which contained HIV-related content. Participants were compensated UGX 20,000 (~$5.49, €4.94) in mobile money, which was transferred to their phone upon completion of the survey.

Figure 1 

Data collection timeline.

Participants were followed-up for six months from enrollment. For this follow-up period, all available data on clinic attendance and viral load were extracted from participants’ HIV clinic records.

Results

Between September 6, 2021, and April 1, 2022, 629 participants (31 newly diagnosed index clients and 598 index clients diagnosed >90 days prior) were recruited to participate in the study. Intake survey data and at least one completed round of the monthly mobility survey were available for 479 participants (24 newly diagnosed index clients and 455 index clients diagnosed >90 days prior) who were included in the analysis. Participants answered an average of 3.3 out of five possible monthly mobility surveys, providing mobility follow-up data for a total of 1,565 person-months. Question completion rates were higher for questions posed earlier in the survey than those posed later in the survey (Appendix 2). Most participants were female (68%) and had a mean age of 41 years (Table 2). Participants included 273 refugees and asylum seekers (57%) and 206 Ugandan nationals (43%). The most common countries of origin among refugee and asylum seeker participants were Rwanda, Somalia, and Burundi. One in four participants reported experiencing food insecurity.

Prospectively collected mobility data

The proportions presented are of person-months for which data on each topic was available (Appendix 2). Participants reported a residence change in almost a third of the follow-up timepoints (473/1,552 person-months, 30%); similar trends were observed for spending nights away from home (465/1,482 person-months, 31%). Travel often concerned only one trip (246/513 person-months, 48%) and lasted a short period of time; in 50% of cases (264/530 person-months), total monthly travel lasted only 1–2 nights, though in 5% of cases (24/530 person-months), travel was >3 weeks. Based on the monthly mobility surveys, 60% of participants (185/307 participants) could be considered mobile having taken ≥2 trips during any follow-up month, having traveled outside the district or further during any follow-up month, or having spent >1 week to 2 weeks away from home during any follow-up month (Table 4). Some differences in mobility could be observed by sex, food security status, and refugee/Ugandan national status (Appendix 3). Namely, female sex was associated with a higher prevalence of home change in the past 4 weeks (p < 0.01), spending nights away from home was more prevalent among participants who were food insecure (p = 0.03), and refugees were more likely than Ugandan nationals to report a home change or having spent nights away from home in the past 4 weeks (p < 0.001 and < 0.01, respectively).

Table 4

Prospectively collected mobility.

	PARTICIPANT TOTAL (N = 307)1
Mobile (meet any of the criteria below), N (%)	185 (60%)
Frequency: took ≥2 trips during ANY month, N (%)	150 (49%)
Distance: travelled outside the district or further during ANY month, N (%)	68 (22%)
Duration: Spent >1 week to 2 weeks (8–14 nights) away in ANY month, N (%)	69 (23%)

¹ Of the 479 study participants, 307 (64%) answered monthly mobility survey questions about travel frequency, distance, or duration.

To gain a better understanding of mobility characteristics, participants were asked to answer questions specifically about their most recent trip. Participants provided information on their most recent trip for a total of 319 person-months. Proportions are presented of person-months for which data on each topic were available (Appendix 2). Most of these trips (143/319 person-months, 45%) were conducted to obtain food or to support farming activities (Figure 2). Trips to obtain food or support farming activities were typically short, lasting 1–2 nights (59/143 person-months, 41%) or 3–7 nights (73/143 person-months, 51%) and were mostly carried out within the subcounty (79/143 person-months, 55%) or even within the parish (34/143 person-months, 24%). These trips often concerned regular trips that were carried out weekly (62/143 person-months, 43%). In a minority of cases (6/143 person-months, 4%), however, travel to secure food lasted >3 weeks. Work and trade was the second most common reason for travel (106/319 person-months, 33%) and had a similar duration of 1–2 nights (72/106 person-months, 68%) or 3–7 nights (32/106 person-months, 30%). These trips were usually conducted closer to home within the parish (66/106 person-months, 62%), and often concerned first-time trips (59/106 person-months, 56%). Travel conducted for education (10/319 person-months, 3%) was often of a longer duration, with the majority of trips conducted for this purpose (7/10 person-months, 70%) lasting >3 weeks. Other reasons for travel included attending a function such as a wedding or a funeral (31/319 person-months, 10%), visiting family and friends (19/319 person-months, 6%), and health reasons (10/319 person-months, 3%). Travel outside of Uganda was less common (5/319 person-months, 2%) and was reported for a variety of reasons, including health, visiting family and friends, attending a function, and looking for food. One participant reported traveling outside of Uganda weekly for health reasons.

Figure 2 

Alluvial plot showing the relationship between various travel characteristics for 319 recent trips reported by participants (N = 273).

* All travel characteristics are mutually exclusive.

Discussion

Refugees and Ugandan nationals living with HIV and accessing HIV care in refugee settlements in Uganda are a mobile population. Roughly 14% of participants reported spending a continuous month or more away from their home in the past year, and 60% of participants conducted long, far, or frequent trips during the six-month study period. Men were more likely to be mobile than women, and younger participants were more likely to travel than older participants. Travel was commonly conducted to secure basic survival needs such as food and income; it typically spanned a few days to a week, and 72% of trips did not reach beyond the subcounty. A minority of participants reported travel lasting multiple weeks and travel outside of the region. The responses of a subset of participants who answered HIV-related questions suggest that ART may not be taken consistently during travel due to logistical challenges and fear of HIV-related stigma, and that HIV clinic attendance may be negatively impacted by travel. While many participants reported never or only sometimes taking ART when they were traveling, notably, a subset of participants reported better adherence to ART during travel. In bivariate regression analysis, spending a prolonged period of time away from home was associated with a lower likelihood of retention in HIV care.

Our findings on refugee mobility are consistent with what has previously been described in the literature. Others have similarly found that mobility among refugees is strongly linked to obtaining the income and food necessary for daily survival. Ethnographic fieldwork conducted in Guinea revealed that Sierra Leonean refugees established business relationships, set up local and long-distance trade, and maintained crucial social ties through multi-directional cross-border movement [24]. In Kenya, Somali refugees commuted between refugee settlements and cities to buy and sell produce [25]. Among Liberian refugees in Ghana, three types of economic mobility were observed: movement between the refugee settlement and the capital Accra to sell and purchase goods and pursue employment opportunities, less frequent travel between the refugee settlement and Liberia for cross-border trade, and mobility between the settlement and subregional countries for trade in particular items [26]. Similar to our study findings, food has previously been identified as an important travel motivator among South Sudanese refugees in northern Uganda [27]. Lack of dependable food provision and cashflow sources in the refugee settlement led South Sudanese refugees to engage in cultivation across the border, where farming was said to be cheaper and easier, and to seize money-generating opportunities throughout Uganda when they were presented to them. In contrast to our study findings, the travel observed for South Sudanese refugees in Uganda was often for extended periods of time. Reasons for travel included visiting family and friends in other refugee settlements, undertaking secondary education at boarding schools (mainly boys), health, and business reasons.

In our study, we found that men were more likely to spend prolonged periods of time away from the refugee settlement than women. Differences in mobility profiles by sex have also been described by others [17, 28, 29]. Apopulation-based study in Kenya and rural Uganda (non-refugee populations) showed that men were more likely to travel for labor reasons, while women more often traveled for other reasons such as care-giving or -seeking, visiting family or friends, and attending a funeral [28]. Differences in mobility profiles by socioeconomic class have also been described. For Liberian refugees in Ghana, it was suggested that the majority of refugees participating in cross-border movement for economic reasons were better off and had sufficient assets to utilize mobility to their financial benefit [26]. Among South Sudanese refugees in Uganda, those conducting international business frequently had dependable access to resources (transport and trade goods) to support this travel [27]. Cross-border movement, however, was seen at both ends of the socio-economic spectrum, with the majority of people moving out of desperation.

To our knowledge, no studies have been published on the association between mobility and HIV care outcomes for refugees living in refugee settlements in sub-Saharan Africa. Outside the refugee settlement context, mobility has been associated with heightened HIV acquisition risk and inferior treatment engagement, though the heterogeneity of mobility measures and study designs has precluded synthesizing findings in a meta-analysis [30]. Our findings indicate a possible negative association between mobility and HIV care outcomes that is consistent with other populations.

Given the systemic drivers of mobility for refugee populations, it is unlikely that these behavioral patterns will change soon. To improve HIV care for refugee populations, it is imperative that mobility be considered when designing interventions to optimize HIV care outcomes. The UNAIDS Gap Report published in 2014 already made the case for developing migration-aware responses to HIV [31, 32]. Ideally, interventions would ensure uninterrupted care access for refugees during travel, both inside and outside of host-countries. Potential solutions include the development of cross-border initiatives such as the cross-border forums on HIV and TB organized by the Musina Municipality in South Africa and their counterparts in neighboring Beitbridge, Zimbabwe [33]. Collaboration and communication across health centers within host countries are equally important. Biomedically-linked electronic medical records may play a facilitating role in fostering communication by making patient histories and treatment plans available to care providers everywhere. Though not ubiquitous, a 2022 report showed that mobile phones are owned by 81% of refugees in Nakivale, Bidi Bidi, Palorinya refugee settlement, and Kampala, suggesting that there may also be a role for mobile health (mHealth) interventions, for example through regular mobile check-ins between patients and providers while they are away and messaging to promote ART adherence [34]. Improving ART access for traveling refugees is also of importance, as evidenced in our study, in which participants reported bringing insufficient ART for their travels, losing ART medication, and having ART stolen during their trip. Currently, accessing services at other facilities may be challenging for mobile populations, as others have reported that official transfer letters are sometimes required in order to do so [17]. Finally, a proportion of refugees deemed lost to follow-up (LTFU) may have transferred out to other facilities [35]. Strengthening handoff and linkage of health records will be vital to correctly estimating LFTU rates and ART engagement for mobile populations.

The findings of this study should be considered in light of a number of study limitations. First, heterogeneity exists in the way mobility has been defined in the literature. In the absence of a unified understanding of this complex multi-dimensional phenomenon, we used a definition related to travel duration and, in sensitivity analysis, a composite measure of mobility that incorporated duration, distance, and frequency of travel. Study conclusions are dependent on how the exposure was defined, and different associations may have been found for alternative definitions of mobility. Second, one of the enrollment criteria for this study was access to a phone. The population with access to a phone in refugee settlements likely represents a group with higher socioeconomic status, limiting the external generalizability of study findings to the general refugee population. Third, while we initially set out to assess mobility longitudinally, high attrition and non-response among participants necessitated the aggregation of monthly surveys and conducting a cross-sectional assessment, hindering our ability to assess temporal associations between mobility and our study outcomes. Fourth, in addition to retention in HIV care, we also hoped to examine the association between mobility and viral suppression. Unfortunately, due to high missingness and only three participants with viral non-suppression in the dataset, there was insufficient data to ascertain this relationship. Fifth, due to the nature of interactive voice responses, we were able to collect only categorical data on travel duration. This, combined with the necessity of aggregating the six follow-up months, made it difficult to determine the total travel length. Sixth, entering a passcode to access the second HIV-related half of the survey proved challenging for many participants, and adding this extra layer of participant protection resulted in lower question completion rates and a smaller sample size for this subsection. Finally, it should be noted that Uganda’s model of hosting refugees offers refugees greater freedom of movement and livelihood pursuit than is granted by most other countries, which limits the generalizability of our findings.

Conclusion

Refugees and Ugandan nationals accessing HIV care in refugee settlements in Uganda frequently traveled to meet their survival needs, and more than one in ten spent a prolonged period of time away from home in the past year. ART adherence may be challenging during travel, and mobility was associated with inferior retention in HIV care. Mobility should be considered in interventions to optimize HIV care in refugee settlements.

Additional File

The additional file for this article can be found as follows: