In this retrospective observation study of a primary healthcare clinic practicing test and treat, we observed a high cumulative incidence of loss to follow up. Four years after starting ART, one in every five patients that had started ART was lost to follow up. The proportion of LTFU was higher in patients that started ART within seven days of an HIV positive diagnosis than those who delayed ART. It is possible that patients had not yet received enough counselling and consequently not yet appreciated the benefit of starting ART when they were not yet “feeling sick”. Previously, patients were taken through a minimum of three counselling sessions, were required to bring a treatment supporter, and had to demonstrate understanding of long term treatment by answering questions after counselling [22]. We therefore speculate that probable lack of the perceived benefit to start ART on the same day of a positive diagnosis, coupled with inadequacies in the preparatory counseling could have led to the higher incidence of LTFU in the T&T group. In this cohort for example, of the 3606 patients that started ART within seven days, 52.1% initiated ART the same day of a positive HIV test. Addressing structural bottle necks including counseling after a positive HIV test and before ART initiation were identified as strategies for improving ART adherence and retention [23]. Inadequacies in pre-ART initiation counseling might compromise the patient’s perceived benefit for initiating ART instantly; similarly, during such a short while, key patients’ concerns that might affect long-term retention have not been exhaustively addressed.

The absolute differences in the proportions of patients LTFU in the subsequent time points and between treatment groups increased and peaked at 24 months. During 2012 and before, ART was not commonly provided in health centers at level III (Health Centers at level three). In the Ugandan setting, these provide outpatient services, maternity, general ward and laboratory. However, at the beginning of 2013 and going forward, most health centers at level III were ART accredited. This is suggestive that a large number of patients formerly at Masaka might have opted to receive ART services at these centers without formally seeking transfers/referrals. Self-transfers across ART programs have been illustrated to conceal the actual proportion of patients categorized as LTFU in a study in another similar setting[24]. In a similar realm, Masaka clinic serves as the main HIV OPD clinic for the regional referral hospital. We therefore speculate a possibility that patients diagnosed on wards and started ART instantly or within 7 days opted to receiving ART at health facilities nearest to their usual dwellings once they got better. Furthermore, these patients could have died after starting ART but were never reported given the passive nature of surveillance in our setting. Same day HIV diagnosis and ART initiation has widely been practiced under PMTCT (specifically under the Option B+). It has however been noted that retention in such settings has remained sub optimal. Moreover, initiation of ART on the same day of testing positive was independently associated with an elevated risk of loss to follow up in the initial months of starting ART [25–28]. A higher proportion of patients retained has been reported in a study in rural Uganda [29] and an almost comparable proportion in another study in Malawi [30] both at one year. Differences in proportions reported in these studies to ours could result out of methodological variations in determining loss to follow up as well as differences in the array of service delivery across the study populations. For example, Brown et al [30] evaluated retention in a streamlined care and universal test and treat model, a model designed to reduce patient barriers to care, unlike the typical and clinical setting in our study, while Jain et al [29] reported retention in asymptomatic patients with CD4 cell count restricted to ≥350 cells/ml..

Similar to another study in the same setting [31], retention rates in all other age groups were better compared to adolescence or being a young adult (18-24 years). Retention in adolescents and young adults should be an important subject given the rising rates of infection in this particular sub population [32] and high rates of viral un-suppression [33]. If not at school, adolescents and young adults are usually at conflict with work schedules and most times fail to make routine monthly schedules, a requirement in most ART clinics. Similarly, clumping ART services of adults together with those of adolescents and young adults might blur individualized adolescent and young adults’ needs. We anticipate that adolescents and young adults’ groups might benefit from differentiated care models that address individualized patients’ needs.

We observed that clients initiated on ART with WHO clinical stage categorized as either 3 or 4 were more likely to get LTFU compared to those staged 1 or 2. In asymptomatic patients initiated on ART in South Africa, the proportion of LTFU was low. Additionally, a reduced risk of death and improved retention rates were observed in a study assessing effectiveness of a streamlined model of care [29,30]. Patients categorized as being in stage 3 or 4 manifest with a higher likelihood of getting opportunistic infections, and so, will be bed ridden most of the time. This might make their continued engagement with the HIV clinic hard. Such patients are further at an increased risk of death especially in the first 6 months of ART due to severe immune reconstitution inflammatory syndrome and Cryptococcal meningitis [34,35]. It is therefore possible that they could have died shortly after starting ART and were never reported to the ART clinic.

We observed that patients with access to a telephone (mostly mobile) were less likely to get lost to follow up. In our setting, patients are sent short reminder text messages (SMS) before the clinic day and those who miss a clinic day are immediately called for a re-appointment. This cannot happen if one has no phone and so may lead to loss to follow up. Mobile phone technologies (mHealth), specifically SMS reminders have improved patient outcomes in other health service delivery settings [36–38] but were comparable to the standard of care for HIV retention in other settings [21,30]. It is however important to note that the level of interaction between provider and patient, and subset of activities under mHealth greatly determine the effectiveness of particular interventions. Therefore, interactive SMS reminders alone might not improve patient outcomes when compared to a combined strategy of SMS reminders, home visits and direct phone calls to patients.

We did not find a statistically significant relationship between CD4 cell count and LTFU. This finding was also observed by Jain et al [26] but contrasts results of studies in other similar settings [40–42]. One possibility for the contrast could be differences in the determination of LTFU as well as the differing CD4 cell count thresholds used across the studies. A patient was classified LTFU if they had spent at least 180 days without picking their ART from the HIV clinic after a scheduled visit [41]. Berheto et al and Honge et al classified patients as LTFU if they failed to pick their ART after 90 days from the last scheduled date [40,42], a similar definition to that used in our study. In the papers written by Berheto et al and Honge et al, an elevated risk of LTFU was noticeable in patients that had a baseline CD4 cell count of <200 cells/ml [37,39] while a similar risk was observed in patients with a CD4 cell count of >200 cells/ml by Mberi et al [41]. In comparison to our cohort, we categorized CD4 cell count as <350 cells/ml, 350-500 cells/ml and >500 cells/ml. Nevertheless, changes in WHO and in-country treatment guidelines over the course of study period might have resulted into almost similar immunological patients starting ART in the T&T and deferred groups.