Characteristics of participating health facilities
Overall, 116 health facilities across Uganda were included in the study. In terms of ownership-type, the majority 68 (58.62%) were public facilities while 27 (23.27%) were private not-for-profit (PNFP) and 21 (18.10%) were private for-profit (PFP).
With regard to level of care in the Ugandan health system , most participating facilities (34/116 or 34.46%) were Health centre IVs (sub-district facilities) followed by Health centre IIIs (sub-county) (31/116 or 31.36%) and general hospitals were 24 (24.55%).
By setting, the majority of health facilities (69 or 61.69%) were based in urban settings compared to 43 (38.39%) health facilities located in a rural setting.
In terms of HIV client loads, the majority of health facilities (75/116 or 68.81%) had at least 500 or more active ART clients compared to 34 (31.19%) health facilities which reported having less than 500 active ART patients.
Characteristics of respondents
A total of 116 ART clinic managers participated in this study. In terms of gender, 51% (59/116) were males while 49% (57/116) were females. More than a half of all respondents (54%) were in the 30-35 age range.
The overall mean work experience of respondents was 8 years (1–20).
In terms of health worker cadres, Clinical Officers at 40 (34.1%) were the most represented, followed by nurses at 37 (31.8%) and physicians at 24 (21.1%).
Year when DARTS was first implemented at participating health facilities
Figure 2 shows that most of the health facilities (63/116 or 57%) commenced implementation of DARTS models in 2018. However, our qualitative data from facilities implementing DARTS models revealed that ART clinic managers perceived differentiated ART as not an entirely new service delivery innovation and that they had been informally implementing some models (such as 3-monthly appointment spacing) prior to the release of formal national ART guidelines designating them as such in 2016.
‘We were already doing differentiated service delivery even before the term was coined as such. That term (DSD) came in around 2014 but DSD has been with us for such a long time. We were already doing three- month refills long before this became part of formal guidelines or a systematic way of implementing it was developed. DSD is not new. What is new is the terminology used.’
[ART clinic in-charge, PNFP-01]
Proportion of health facilities implementing DARTS
The majority of health facilities 100 (86%) were implementing at least one DARTS model. As Figure 3 shows, there has been a higher uptake of facility-based DARTS when compared to community-based models. The most implemented facility-based DARTS model was the Fast Track Drug Refill (FTDR) model. FTDR was reported to be in implementation in 100 (86%) of health facilities in our sample.
Community Client-Led ART Delivery (CCLAD) was the most popular community-based DARTS model implemented in more than a half of all health facilities 63 (54%). Overall, Community Drug Distribution Points (CDDP) had the lowest uptake at 33 (24.88%) of all DARTS models.
Across our in-depth interviews with ART clinic managers it was revealed that the CDDP model was widely perceived as costly to implement. While the CCLAD model entailed self-forming patient groups primarily financed by contributions from individual members such as sharing transport costs for picking ART refills, the Community Drug Distribution Points (CDDP) model depended substantially on the resource envelop of the hosting health facility. Donors supporting the CDDP model in some participating facilities were reported to have discontinued funding due to escalating operational expenses. The costs required for running CDDPs were identified. These include fuel for transporting ART refills into communities, monetary field allowances for facility-personnel during visit to communities and the costs of maintaining the physical infrastructure in the community hosting the drug pick-up points.
‘’ We have some community approaches that are expensive such as CDDPs because the hospital has to foot the transport costs for ferrying ART refills into rural, remote outreach posts. With the CCLAD model there is no added expense on the health system. It is the patients meeting the costs of transport for picking ART refills from hospitals’’ [National-level HIV program manager, Ministry of Health].
High-volume sites or those with 500 active ART clients or more, were implementing the CCLAD model more than low-volume HIV clinics or sites which had less than 500 active clients (p < 0.001).
Characteristics of health facilities implementing all five DARTS models
Overall, only 25 facilities (21.55%) were implementing all five DARTS models recommended by the Ministry of Health of Uganda. The characteristics of the 25 health facilities are represented in Table 1. This table shows that almost half (48%) of the facilities reporting implementation of all five DARTS models were private not-for-profits (PNFPs) the majority of which had a faith-based foundation. More than half of the 25 health facilities were general hospitals while a quarter of them were Regional Referral Hospitals (RRHs). This may suggest that DARTS uptake is highest at the most advanced level of care in the Ugandan health system (tertiary and secondary levels).
Uptake of DARTS by ownership-type of health facility
We found variations in uptake of DARTS by ownership-type of a health facility. Overall, Private for-profit (PFP) health facilities reported the lowest uptake of DARTS models across ownership-type of facility. Out of the 21 private for-profit facilities in our sample, only a third had implemented DARTS models at all. In contrast, 67% of all public and 52% of all Private not-for-profit (PNFPs) commenced DARTS implementation in the year 2018 alone. The year a health facility started implementing DARTS models and ownership-type of health facility are significantly related (p = 0.001).
Our qualitative data provided insights into why there was relatively low uptake of DARTS models in private for-profit facilities. Participants from for-profit facilities reported that their health workers had not been trained in differentiated ART delivery even when their counterparts in public and not-for-profit facilities were being trained by PEPFAR implementing organizations and the Uganda government at no charge as highlighted in the quote below:
‘No training in differentiated HIV services has been done for our staff in private hospitals. They always tell our staff to pay when attending training by the Ministry of Health. Private hospitals have to dig into their pockets to pay for training of health workers yet this is free for public and not-for-profits. But we don’t have a budget for that’ [ ART clinic in-charge, PFP-06].
Interviews with national-level HIV program managers at the Uganda Ministry of Health and district health team leaders revealed that donors in Uganda prioritized public and for-profit facilities because these had the highest HIV client loads and hence donors perceived their investments to have ‘a higher yield’ there. As such, it emerged from this study that for-profit (PFP) facilities were not being prioritized in the national scale-up of differentiated ART delivery across Uganda and that health workers in this category of health facilities had not been trained on how to offer DARTS models.
High-volume facilities (those with 500 or more active ART clients) implemented a more diverse mix of models across both community (mainly CCLAD) and facility-based (principally FTDR) models compared to low-volume facilities (those with less than 500 active ART clients) which exclusively implemented facility-based (principally FTDR) models (p =0.001).
Understanding the relatively low uptake of community-based models
Overall, the uptake of community-based models was less than that of facility-based models. Figure 4 shows that facility-based models have received more demand from patients when compared to community-based models according to majority of respondents (85.0%).
Our qualitative findings shed light on why this was the case. Participants attributed the relatively low demand for community-based models to HIV-related stigma. Patients were said to be filled with fear of unintentional disclosure of HIV status or plainly, being known to live with HIV, by peers living in the same neighborhoods if they joined patient groups such as CCLADs.
‘Patients do not actually like the community models because of stigma, especially CCLADs which involves forming groups of six patients living in the same community. That means all the patients in the group know each other and reside in the same neighborhood. But patients will tell you “I do not want the other person to know that I am sick and get (HIV) care from this facility’ [ART clinic manager, PF-03].
Rural-urban differences in DARTS uptake
Rural-based facilities implemented the Community Client-Led ART delivery (CCLAD) model (67.4%) more than urban-based facilities (42.0%) (p < 0.001). The latter category of facilities reporting a more even uptake of both community and facility-based models (such as Fast Track Drug Refill at 44.9%).
In-depth interviews with ART clinic managers revealed that urban patients were sophisticated and that they tended to prefer individualized facility-based care due to their need for privacy and the convenience offered there.
‘Patients in urban settings like Kampala (city) do not want to form community groups. They just want to go to the facility, get their drugs and go home, or go to a point somewhere, get their drugs and go home. Patients in towns are not interested in forming groups. They are interested in convenience and saving time. They just want to go to a nearby pharmacy, pick their drugs and in ten minutes get out’ [ART clinic manager, PUB-04].
Our qualitative findings also suggest that there is a section of patients who prefer regular face-to-face interactions with health workers and the psychosocial support and satisfaction derived in standard clinic-based care. There was a perception among some patients that enrolling in community-based models entails, to some degree, a dis-engagement from the formal health system which offers some insight into understanding patient preferences and the outcomes of uptake of varied DARTS models.
‘Community models are good because they reduce transport costs and allow us more time at work but for me, as an individual, I want more of that psychosocial service… that personal touch by a health worker. That physical interaction. Because if you are not at the facility you don’t get that (individual) attention and care and that is why I prefer to come here on a regular basis’ [ Focus group discussion with patients, PNFP,003].
On the other hand, several patients in rural-based facilities expressed satisfaction in community-based models owing to the savings in time in seeking facility-based care and the significant reduction in travel costs associated with being enrolled in ART refill pick- ups from outreach sites within the community (CDDPs). Patients enrolled in the CCLAD model appreciated the opportunity of sharing transport costs with peers for ART refill pick-ups from facilities.
Gender dimensions in uptake of community models
Interviews with ART clinic managers revealed gender dimensions in the uptake of community-based DARTS models. Across six case-study facilities it was consistently reported that the majority of the members of Community Client-Led ART delivery (CCLAD) groups were adult women and that adult men had not taken as active an interest in enrolling in CCLADS groups when compared to women.
‘We are noticing some gender dynamics in uptake of community models. Most of the CCLAD groups we have here are made up of mainly women. Men have shunned these groups. The most active within these groups are actually women who are loyal to their groups and they are consistent in them’ [ART clinic in-charge, PUB-02].
The most practical DARTS model to implement in the assessment of ART clinic managers
We asked health facility staff to mention the most practical DARTS model to implement in terms of one that required the least material resources and health worker in-put at the facility-level.
Figure 5 shows that the majority of them (64.9%) selected Fast Track refill (FTDR) as the most practical DARTS model to implement in terms of requiring the least resource in-puts (p < 0.001).
Our qualitative data, particularly through interviews with ART clinic managers, offered keen insights into understanding why FTDR was selected as the easiest DARTS model to implement.
‘The Fast Track Drug Refill (FTDR) model has had a high success rate with us. This is good because it is the easiest to implement at the facility- level in my opinion. The guidelines are very clear, you conduct two viral load tests and if the patient is stable in the two instances, you then extend their appointments by 3-6 months. It is that easy’ [ART clinic in-charge, PNFP-02].
In contrast, ART clinic managers contended that community-based DARTS models (such as community drug distribution points) were relatively expensive to implement and yet facilities had not received adequate funding and resource facilitation from leading donors such as PEPFAR and the Uganda government to implement them.