Understanding the Time Needed to Link to Care and Start ART in Seven HPTN 071 (PopART) Study Communities in Zambia and South Africa

For the HPTN 071 (PopART) trial intervention, a community was defined as the population catchment of a government health care facility providing HIV services including ART. The seven communities included in this analysis were randomly allocated to receive the full PopART intervention package, including offer of ART initiation regardless of CD4 count for PLHIV from the start of the trial, and prior to changes in guidelines. The communities included have a total population of ~ 200,000 in the four Zambian communities, and ~ 100,000 in the three South African communities. The social context and HIV services in all the communities in both countries have been described elsewhere [38]. An ancillary study on stigma was undertaken, which provides additional information on stigma experiences in health facilities and levels of stigma in different communities, also described elsewhere [35, 39, 40].

The PopART intervention was a combination HIV prevention package delivered by community HIV care providers (CHiPs) and is described in detail in Hayes et al. [18], and summarised here. CHiPs visited every household within a community annually offering HIV education and testing using routine point of care finger prick rapid testing [41] to all consenting household members. ART was provided by local government health care facilities. For all individuals newly-testing HIV-positive or reporting that they know they are HIV-positive but not yet on ART, the CHiPs through repeated household visits, encouraged and facilitated linkage to care for ART initiation at the local clinic. The ART drugs were provided through the routine health care system funded by PEPFAR-supported implementing partners. It should be noted that until the change in national guidelines in 2016 (recommending prompt treatment, regardless of CD4 count), Arm A participants had to sign a consent form if their CD4 count was > 500, prior to initiation of ART. Because they had to wait for their CD4 count results, there was a brief delay of approximately 2 weeks between linkage to care and ART start.

In Round 2, several new strategies were put in place to encourage linkage to care, to be undertaken by the CHiPs, and existing practices from Round 1 were reinforced [42]. These included assisted referrals: CHiPs escorting clients to the clinic to help them link to care and specially trained counsellors in each community to assist with linkage; working with existing community health care workers to track clients; following up with clients who missed clinic appointments and holding meetings with local clinic staff on a regular standardised basis to review which CHiP clients had, or had not, linked to care and/or started ART.

In order to offer annual HIV testing as well as other HIV prevention services, CHiPs aimed to contact all individuals (aged ≥ 15 years) at least once every round to offer participation in the intervention. All household members were asked for verbal informed consent to take part in the intervention and permission to collect data electronically. Individuals provided written consent for HIV counselling and testing. CHiPs recorded basic data on the household and more detailed information for individuals who consented to participate in the intervention. Data captured electronically included age and gender, self-reported HIV status, the outcome of HIV testing by CHiPs, self-reported information on HIV care registration and current (during the last 1 month) ART use, and whether a referral to HIV care was given. Analysis was restricted to participants aged 15 years and older, although HIV testing was offered irrespective of age.

Individuals were “known HIV-positive” to the CHiPs in Round 2 if they participated in Round 2 and either (i) had self-reported or tested HIV-positive in Round 1, and verbally confirmed their HIV-positive status in Round 2 (ii) self-reported HIV-positive for the first time in Round 2, or (iii) tested HIV-positive with CHIPs in Round 2. For individuals who self-reported they had previously registered for HIV care, CHIPs asked to see the patient ART card, and if the card was provided then the ART card number was recorded.

For all individuals known to be HIV-positive by the definition above, CHIPs undertook follow-up visits to support linkage to care and subsequent ART initiation, as well as to collect self-reported information on whether and when these outcomes were achieved. If the individual was not at home at the time of the CHiP follow-up visit, then the CHiP recorded that they had made a visit but were not able to meet the individual.

Analyses of the time from CHiP referral to linkage to care, from CHIP referral to ART initiation, and the time from linkage to care to ART initiation, were done using the Kaplan–Meier “time-to-event” method. We used Cox regression for analysis of whether these times differed by country, community, gender or age group. We used follow-up data to 30th September 2017 and for individuals who were not known to have started ART by this date, we censored their follow-up on the last date on which they met the CHiP in person. Individuals do not contribute to the analysis if they were never followed up in person after they were referred to care.

Social science enquiry encompassed qualitative research prior to the intervention period (2012–13) and during the intervention. The trial qualitative research used ethnographic and participatory research methods [43] and was designed to first document rapidly key features of study communities of relevance to HIV prior to the intervention period through broad brush surveys (BBS) [44], and then, during the intervention, focus on community responses, implementation and different trajectories around a range of HIV outcomes and decisions (labelled ‘qualitative cohort’). In this paper, we draw on findings on community characteristics from the BBS (the methods are described in detail in other papers, see [43, 44]) and on findings from a sub-sample of PLHIV in the qualitative cohort in both countries documenting the challenges in linkage to care alongside other experiences. The research was carried out by a core team of Zambian and South African men and women social scientists trained in qualitative data collection methods, who were competent in the appropriate local languages. In both countries, individual social scientists were allocated to three to four communities where they, and a field research assistant, conducted all research components as a pair, building their familiarity with particular communities and families over time. Efforts were made to have each pair composed of a man and a woman where possible.

The aim of the qualitative cohort research was to recruit participants purposively sampled to represent different HIV-related decisions and outcomes across different trial arms, geographical locations, age, gender, and key populations. Sixty-eight households (21 in Zambia and 47 in South Africa) where PLHIV resided were recruited from March 2016 to November 2017 (a time period which overlapped with the quantitative data collection in Rounds 2 and 3). Two-thirds of the PLHIV in the households included in the cohort were on ART. Of the other third, half had defaulted from treatment and half were yet to start ART. Participants in the cohort were interviewed at least three times, and households visited at least six times over a year. The interviewers asked about family, kin and social networks, movement within and outside of the household and community, household livelihood, HIV and general health service access, sex, love and romance, and future ambitions and fears. Interviews were recorded, and notes were also taken during and after interviews, with each social scientist first writing up a reflective summary of the findings and then debriefing with other team members, before translating and transcribing the interviews. ATLAS-ti qualitative analysis software was used to manage the data.

This body of data was closely reviewed and analysed by the social science team for this paper, focusing on the participants who were living with HIV from across all seven Arm A communities with the aim of explaining time between HIV diagnosis and ART initiation. Notes from field work, summaries, debriefing sessions and transcriptions were scrutinised. This was done by first holding two-day analysis workshops in both countries with the social scientists who carried out the fieldwork. This process allowed key themes to emerge across both countries (for example, `gender patterns and differences’, `when it makes sense to start ART on the same day’), which were then shared to arrive at a shared set of codes. The teams were asked to review the data to code against the themes, ensuring that coding was cross checked for consistency across coders throughout the process. Each country produced a synopsis document, which was discussed by the country teams highlighting differences and similarities, to arrive at the list of themes (shown in Tables 3 and 4) used in this paper.

The HPTN 071 (PopART) study was approved by the London School of Hygiene and Tropical Medicine, University of Zambia, and Stellenbosch University (N12/11/074), ethics committees and by other health governmental authorities. Research participants signed written informed consent. Those who consented to a finger prick rapid test signed written consent according to standard government guidelines. Data from the CHiP intervention were collected following verbal consent.

To ensure confidentially, quotations from the qualitative data are only labelled with Z = Zambia and SA = South Africa, and community number.

During Round 2, in Zambia 3435 and in South Africa 1262 HIV-positive individuals were referred to HIV care by CHiPs, among those who were not on ART on the date of first participation in Round 2. Overall, the estimated median time from CHIP referral to ART initiation was ~ 6 months in Round 2 in both Zambia and South Africa, considerably less than the median of ~ 10 months in both countries in Round 1 but still slower than had been targeted at the start of the intervention (Fig. 1a; the horizontal red line is drawn at 50% to correspond to the median time to ART initiation).

Overall, we estimated that 39, 50, and 67% initiated ART by 3, 6, and 12 months respectively after first referral to care by CHiPs in Round 2 in Zambia (Fig. 1b), with corresponding (and slightly lower) estimates of 36, 47, and 66% in South Africa (Fig. 1b).

The estimates for the percentage who were linked to care, i.e. first attended a clinic, by 3, 6 and 12 months was 45, 57 and 71% respectively after first referral to care in Zambia, and 48, 60, and 79% in South Africa (Fig. 1c); these estimates indicated slightly more rapid linkage to care in South Africa compared with Zambia.

Notwithstanding the imprecision around our estimates of both the date of linkage to care and the date of ART initiation, because both dates were based on self-reported information, we also estimated the time from linkage to care to ART initiation among individuals who linked to care. We estimated that 78, 90, and 94% started ART by 1, 3, and 6 months respectively after linkage to care in Zambia, with corresponding (and slightly lower) estimates of 64, 81, and 87% in South Africa (Fig. 1d).

Overall, for most individuals the time from linkage to care to ART initiation was short, with a median of less than 1 month (Fig. 1d), while the time from referral to linkage to care was longer, with a median of 4 months in Zambia and 3.5 months in South Africa (Fig. 1c).

Patterns by gender, community and age group for the time from CHiP referral to ART initiation were similar to those for the time from CHiP referral to linkage to care (attendance at clinic). Here we have summarized the patterns for our estimates of time to ART initiation (Tables 1 and 2), with estimates for time from CHiP referral to linkage to care provided in Supplementary Table 1.

For Zambia, findings are summarised in Table 1 and Fig. 2. The median time to ART initiation was longer for women than men (6.6 and 4.4 months respectively, p < 0.001). There was strong evidence of differences between communities (p < 0.001), with the median time to ART initiation being similar in communities 2 and 3 (5.7 and 5.6 months respectively) but shorter in community 1 (1.8 months) and longer in community 4 (9.5 months). Among men, it was longest for those aged 20–24 years, and among women it was longest for those aged ≥ 55 years.

For South Africa, findings are summarised in Table 2 and Fig. 3. The median time to ART initiation was shorter for women than men (6.0 and 9.1 months respectively, p = 0.006). There was strong evidence of differences among communities (p = 0.007) with the median time to ART initiation being 5.6, 7.9, and 3.9 in communities 1, 2 and 3 respectively. Among men, it was shorter for older men aged ≥ 45 years compared with those aged 25–29 years, and longer for women aged 15–19 years compared to those aged 25–29 years.

The estimated time from linkage to care to ART initiation was similar for men and women, and across communities, in Zambia. In South Africa there was evidence that this time was shorter in communities 1 and 3 compared with community 2, but the difference was relatively small (shown in Supplementary Table 2). In both countries, for both men and women there was no evidence of differences by age group (data not shown).

Our findings show that there are differences in time to ART initiation between communities and these can be categorised by: distance between household and clinic; levels of socio-economic status and employment; reported clinic staff workload at the health care facility; and the fear of being seen accessing services affecting time to ART initiation.

For example, the health facility in Community 1, is located quite far from the town centre, the population the facility serves is relatively educated, middle-class with some PLHIV working within the area. This one ART clinic is not particularly congested (according to users), because of a lower overall population density compared to other communities. In contrast, in Community 4, the health care facility is located close to the town centre. The population accessing care there are mostly of lower socio-economic class, engaged in the informal economy and often involved in mobile livelihoods (for example: cross country-border trade and fishing). Women traders in this community work long days in markets or selling by the roadside and are involved in trade which also involves periods away from the community, restricting their opportunities to attend the clinic. Alcohol use is reported to be high in this community. The clinic is congested (waiting time was usually > 3 h).

Similarly, in South Africa there is an observed association between clinic burden (those serving over 1000 clients a month are considered over-burdened), and a slower time to ART initiation. Community 3, where there is a shorter average time to start ART, is a relatively small community with lower HIV prevalence and fewer people initiating ART each month. In contrast, community 2 is a relatively large ‘gate-way’ community for people entering Cape Town from Eastern Cape searching for employment opportunities and has a relatively high HIV prevalence, and the clinic is busy. Even before the HPTN 071 (PopART) trial began the clinic had a much larger number of clients on ART than the other two communities. Thus, it is notable that in both countries, the populations slowest to initiate ART are those with transient populations (engaged in mobile livelihoods) and congested health facilities.

Health system, structural and social reasons for taking time to link to care are summarised in Table 3 (below). We have grouped the summary findings in the table by health care facility, community, household and individual factors, although several reasons listed are cross-cutting. We now elaborate on some of the reasons given for delaying treatment listed in Table 3, including the gender and age factors apparent in the process data, using italics to highlight reasons mentioned in the table, to ease cross-referencing.

Other factors given for not linking to care included confusion over changing and seemingly-conflicting messages from health workers about when to start ART because of shifting guidelines. Others had been advised by health workers to eat nutritious food with their treatment which they could not afford: a 47-year old woman LHIV in Zambia recalled how health workers told her to buy fruits, but she and her husband could not always pay rent for the room they lived in, so could not spend more money on food. As a result, they had not linked to care.

There were worries about the stigma of being seen once on treatment. A Zambian older man (Z10) recounted, “some people fear queuing at the clinic to collect drugs”. He attributed congestion and lack of privacy in health facilities as contributing to making accessing drugs ‘public’, and remarked, “Accessing drugs should be private or the government should find another way of providing HIV drugs”. In Zambia, respondents accessing care in the busy health facilities spoke about long waiting times (over 3 h), with staff and clients sometimes being rude to each other. A few respondents expressed concern that there was preferential treatment for some clients and some who used their authority and status to go ahead in the queue.

One woman in South Africa (S16) talked about the discomfort she felt about her nearby health care facility, a factor which contributed to other reasons she gave for not starting ART.

The clinic staff like to gossip, I mean, I don’t feel comfortable with them…. I would probably take ART somewhere else, these people from here are not right…. They know me and they like gossiping outside.

Health worker attitudes were mentioned by others with some fearing being reprimanded by health care workers for failing to take treatment. Stories of people who had been scolded at the health care facility for not adhering to treatment when they tried to access care after a gap in treatment, were recounted by people in both countries. These were all factors which made access difficult once someone had linked to care, and deterred others who were yet to start ART.

While health system factors deterred some PLHIV from accessing treatment, family and community views also influenced uptake, often in association with the stigmatising attitudes in the community about those accessing HIV-care services.

In both countries people were often juggling their livelihood needs with starting ART, fearing time lost accessing care when they needed to work [45]. Some PLHIV who were the main breadwinners and had not started ART, would rather use their strength to earn an income and continue providing for their family, while they were feeling healthy, than go for treatment. Sometimes making a living involved long days in the market or months away in fishing camps, farming, mining or the road and construction camps.

A woman sex worker living with HIV and not yet on ART in South Africa (S14) commented that she had to use her time to look for clients, and that having no clients meant she earned no money and there would therefore be no food at home. When she was pregnant she used antiretroviral drugs for the prevention of mother to child transmission of HIV (PMTCT), because her partner used to support her then, so that she did not infect the unborn baby. However, she did not continue with ART after PMTCT. She said she has not been sick ever since she found out about her status so did not see the need for treatment.

Poverty and lack of food influenced the decision of some people to delay treatment uptake, as mentioned above. A 41-year-old woman in South Africa (S14) commented:

So, you have to take treatment every day. So, can you take treatment on an empty stomach? Really? Because even if you are working, there is this maybe two weeks before payday you don’t have anything to eat. So, you’ll have to take a glass of water and pills and go to bed without anything you see. That’s why HIV people, most of the cases, they just default and go drinking [alcohol].

Maturity and age engendered responsibility and autonomy in both women and men, sometimes, but not always, making it easier for older people to decide to start ART. The responsibility of being a provider prompted one 48-year-old Zambian man (Z10) to take treatment; “if I take treatment, I will be able to look after my family. If I do not take the treatment it is the family that will suffer”.

Disclosure of HIV status to partners, between women to their male spouses (in Zambia) was cited as a reason to delay attending care and starting ART. One strategy was to hold off disclosing an HIV-positive status to a partner until they also tested for HIV. In Zambia, women were more likely to be married and living with their husbands than in South Africa. Sometimes, it was up to men in the household to decide if women should start ART or not. Many Zambian women participants spoke of experiencing gender-based violence (physical and verbal), because of the stigma of HIV-infection and assumptions on the part of the male partner about the woman’s sexual behaviour or the woman’s accusations against the man and his behaviour that could have resulted in her infection. Thus, women in Zambia appeared constrained by male hegemony. This was reflected by a younger woman on ART (Z2) who talked about women in her community being beaten and insulted by their husbands for questioning the man’s behaviour, and then hiding indoors for fear of people seeing their physical injuries. Another woman who was HIV-negative (Z7) relayed how violence and the threat of violence can affect a woman’s decision to take treatment:

a man beat up his wife because she started ART, although even he tested HIV-positive. He is one of the people who says ‘HIV and its treatment is a lie, they just cheat us about these drugs’. So right now the woman is sick, she is bedridden.

Although fewer women in our study population were married in South Africa than in Zambia, amongst those who were, participants said it was harder for a woman to start treatment if her husband had not begun ART. A 21-year-old women living with HIV whose husband was in prison said:

I disclosed my status to my in-laws, they advised that I wait for my husband to come out of jail before I could start ART. I will wait for him to be released from prison to get permission to start ART. If he refuses, then I won’t start.

In both countries, gender was an important factor in treatment access. Women had more frequent contact with the health facility than men through pregnancy and child health needs, and the health facility was seen more as women’s space. However, whereas young men’s freedom to move around the community and engage in informal livelihood activities sanctioned and facilitated their use of the clinic in Zambia, in South Africa, men were more likely to resist going to the clinic. A South African man living with HIV (S21) was initiated onto ART but quickly stopped taking it. He was a gang member and an active drug user (taking Mandrax [methaqualone] and Tik [crystal meth]), he had recently been in prison and continued to be involved in criminal activities. As in this example, past-history, gang culture, alcohol and drugs were said by participants to mean that men did not prioritise their health, as also shown in findings from other research [34, 46, 47].

Despite community engagement and messaging on the benefits of immediate ART, many people had internalised previous health messaging about treatment being appropriate only for those who were sick, and were cautious about accepting the change in advice [38]. Therefore, some considered immediate treatment appropriate for those who are ill, pregnant or in a discordant relationship. A Zambian 51-year-old man (Z7) explained: “

When I tested HIV-positive I did not see it fit to start ART because I was not feeling sick, I did not have any pain and I did not feel weak because I used to even lift heavy weights. So, I felt that even if I start taking treatment, it will not change anything.

For other PLHIV, although understanding the health benefits of antiretroviral treatment and expressing gratefulness for wider ART access, people talked of the time needed to ‘accept’ the diagnosis, to contemplate disclosure, household living and clinic options, and to consider the impact of treatment on the body, livelihood and relationships. People often talked about the need to `process’ the information that they were HIV-positive. A 21-year-old man (S13) who tested HIV-positive with one of the CHiPs teams in April 2017, said that when he heard he was HIV-positive “I couldn’t believe it”. He was previously HIV-negative in October 2016 and his disbelief prevented him from accessing care. He also said he was kept busy with other things which kept him from the clinic.

While some respondents talked about the fear of starting ART, they did not always indicate concern that ART would itself have an adverse impact on their bodies. Instead, concern often focused on the worries of being forced to interrupt ART due to drug stock-outs, a situation beyond their control.

Denial, anger, sadness, and questioning were common responses amongst adolescents living with HIV when they learnt of their status. One 21-year-old perinatally-infected man (Z2), who had stopped taking ART, said that he ‘did not believe in ART’. A 22-year-old woman living with HIV (Z8) shared her experience when she first found out she was HIV-positive:

I felt bad, I cried I won’t lie. I cried because I wasn’t expecting [it], I had a lot of thoughts that came that I should kill myself and do other things, […] then I thought about home where I stay, how they were going to receive me looking at how I am living…

For younger people, it was said to be harder to accept your status. A woman in South Africa (S18) mentioned that young people feel they are too young to be taking treatment for the rest of their life and they do not have the discipline to be taking ART every day. Another Zambian 17-year-old girl living with HIV (Z7) said when she first found out she had HIV:

I felt very bad, whereby many questions came to my head like, was I born to be an ARV taker or maybe to be addicted because my life relies on these same ARVs. It was very hard for me.

Yet, despite the many reasons given for delaying starting care, participants in both countries did recognise the importance of starting treatment straightaway, at least for some people. These reasons, summarised in Table 4, point to the value placed on health care facility access and the importance of the support of family and friends as well as being well enough to make a living to contribute to the family.

Past experiences did play a part in the decision to start treatment for some, even if they were not yet ill themselves. Seeing a family member grow weak and die because of HIV, and in some cases remembering episodes of ill health as a child, prompted access to treatment. A 21-year-old woman (Z2) commented:

I have experienced this, not myself but with my late mum when she defaulted, she then stopped taking her medication and this cost her so much such that she had meningitis and her kidney could not function, so I learned from her. So, I started, and I can’t stop because what happened to her can also happen to me.