We have shown that self-transfer is a major reason for LTFU of patients from public ART clinics in Sheema District, Southwestern Uganda, between 2017 and 2021. We were also able to successfully trace the majority (87%) of those LTFU. The prevalence of self-transfers was significantly higher among young adults aged 18–30 years, females and those with asymptomatic or mild disease (WHO clinical stage 1–2), compared to their counterparts. The most common reasons for self-transfers were relocation, lack of money for transport, and lack of awareness on the transfer process. Presence of a phone contact in the patient’s file was independently associated with successful tracing of the LTFU patients. Given the high prevalence of self-transfers, these data highlight the need to account for self-transfers among patients considered as LTFU, to accurately estimate retention in care.

The prevalence of self-transfers among patients LTFU in the current study (76%) is higher than what other previous studies conducted in Uganda have reported, which ranged from 34 to 43% [5, 10]. Similarly, other studies, including systematic reviews conducted in low-and-middle-income countries (LMICs) have reported much lower estimates of self-transfers among patients considered LTFU, ranging from 12 to 54% [15,16,17]. This study finding is not surprising, and could be explained by the fact that our study was conducted in the era where the “treat all” guidelines had been rolled out. Moreover, it was previously hypothesized that most of the losses from ART clinics after the first year of the “treat all” guidelines' implementation would be unreported transfers to new ART-providing sites [6]. The increasing prevalence of self-transfers calls for the need to reorient ART services, to adapt to the evolving challenges in the ART care. Uganda is currently implementing differentiated service delivery (DSD) models. The DSD models adopt more patient-centered approaches, including switching to home-based care settings, reducing frequency of clinic visits, co-opting non-physician workers such as “expert clients”, and considering pharmacy-only refills for stable patients [18, 19]. Accelerating the implementation of such DSD models may reduce the increasing numbers that are self-transferring between ART-providing clinics. However, further research on the most appropriate, and cost-effective DSD models for the self-transfers in the various resource-limited settings, in the era of “treat all” are required.

In this study, participants with asymptomatic or mild disease, females and young adults aged 18–30 years were more likely to self-transfer, compared to their counterparts. Since the rollout of the “treat all” guidelines, PLHIV who begin treatment are increasingly asymptomatic or have mild disease. It was hypothesized that these would present new challenges to retention in care, including self-transfers [6]. Additionally, socio-demographic characteristics, including sex and age have been found to influence mobility of patients between ART-providing clinics [14]. Younger patients are more likely to relocate if they are not in school because of conflicting work schedules, and may be more economically disadvantaged compared to their older counterparts [20, 21]. Moreover, in the current study, the most common reasons for self-transferring were relocation, and lack of money for transport. These reasons have been cited in a number of studies done in similar resource-limited settings, as major contributors to loss to follow-up [5, 22, 23]. These findings imply that multifaceted interventions are required to minimize loss to follow-up. First and foremost, patient mobility is likely to become more common as the number of ART-providing sites increases, and healthcare workers should make deliberate efforts to raise patients’ knowledge and awareness of transfer procedures [14]. Secondly, providing incentives to patients such as drug supplies for a longer time could minimize frequent clinic visits and reduce transportation costs resulting in better retention in care [14, 18, 23]. Additionally, ART-providing sites should be ‘transfer-friendly’, assessing patients for their intention to transfer to other ART- providing sites so that they are appropriately supported through transfer decisions for better treatment outcomes [14].

Patients who had phone contacts available in their files were more likely to be successfully traced compared to others. This agrees with previous findings from studies done in Malawi and Ethiopia [24, 25]. Mobile phone technology has been suggested as one of the interventions to minimise loss to follow-up in LMICs [26]. The increasing availability of mobile phones, even in resource-limited settings provides an opportunity for successful tracing patients considered LTFU, so that their true outcomes are determined. On the basis of this finding, patients’ records in ART clinics should regularly have phone contacts updated at each visit, to ease subsequent tracing. Moreover, phone tracing could reduce the proportion of tracing patients via other resource- and labor-intensive methods, including field tracing.

Overall, our findings highlight the need to account for self-transfers among patients considered as LTFU, given the increasing trend of mobility and self-transfers. Furthermore, there is a need to implement efficient tracking systems such as electronic medical records that use unique patient identifiers to identify such self-transfers across other ART-providing facilities. Our findings also highlight a need to make deliberate efforts to maintain updated patient phone contacts in their files, to ease subsequent tracing. Given the high prevalence of self-transfers in this study, we recommend that future longitudinal studies assess outcomes among this population of self-transfers in similar resource-limited settings.

The main limitation of our study is that it was conducted in selected government-owned health facilities, in a rural district, in Southwestern Uganda. Thus, they may not be generalized to other settings, that are urban and in private health facilities. More studies in diverse settings, including private health facilities and outside Southwestern Uganda should be undertaken to corroborate our findings.