Impact of "test and treat" recommendations on eligibility for antiretroviral treatment: Cross sectional population survey data from three high HIV prevalence countries

Background Latest WHO guidelines recommend starting HIV-positive individuals on antiretroviral therapy treatment (ART) regardless of CD4 count. We assessed additional impact of adopting new WHO guidelines. Methods We used data of individuals aged 15–59 years from three HIV population surveys conducted in 2012 (Kenya) and 2013 (Malawi and South Africa). Individuals were interviewed at home followed by rapid HIV and CD4 testing if tested HIV-positive. HIV-positive individuals were classified as “eligible for ART” if (i) had ever been initiated on ART or (ii) were not yet on ART but met the criteria for starting ART based on country’s guidelines at the time of the survey (Kenya–CD4< = 350 cells/μl and WHO Stage 3 or 4 disease, Malawi as for Kenya plus lifelong ART for all pregnant and breastfeeding women, South Africa as for Kenya plus ART for pregnant and breastfeeding women until cessation of breastfeeding). Findings Of 18,991 individuals who tested, 4,113 (21.7%) were HIV-positive. Using country’s ART eligibility guidelines at the time of the survey, the proportion of HIV-infected individuals eligible for ART was 60.0% (95% CI: 57.2–62.7) (Kenya), 73.4% (70.8–75.8) (South Africa) and 80.1% (77.3–82.6) (Malawi). Applying WHO 2013 guidelines (eligibility at CD4< = 500 and Option B+ for pregnant and breastfeeding women), the proportions eligible were 82.0% (79.8–84.1) (Kenya), 83.7% (81.5–85.6) (South Africa) and 87.6% (85.0–89.8) (Malawi). Adopting “test and treat” would mean a further 18.0% HIV-positive individuals (Kenya), 16.3% (South Africa) and 12.4% (Malawi) would become eligible. In all countries, about 20% of adolescents (aged 15–19 years), became eligible for ART moving from WHO 2013 to “test and treat” while no differences by sex were observed. Conclusion Countries that have already implemented 2013 WHO recommendations, the burden of implementing “test and treat” would be small. Youth friendly programmes to help adolescents access and adhere to treatment will be needed.

Introduction HIV remains one of the biggest contributors to mortality and morbidity in the world with most deaths occurring in the Sub-Saharan Africa (SSA) [1]. Despite freely available treatment for HIV/AIDS over the past decade, only (66.0%) of people living with HIV in eastern and southern part of SSA were on treatment in 2017 [2]. The World Health Organisation (WHO) 2013 treatment guidelines for starting HIV-positive people on antiretroviral therapy (ART) were CD4< = 500 cells/μl and for pregnant women to commence ART regardless of CD4 cell count [3]. In 2015 the WHO guidelines changed to starting every HIV-positive person on ART regardless of CD4 cell count [4] (the so-called "test and treat" approach) although some countries have not yet implemented these recommendations.
With only half the population of HIV-positive individuals on treatment, containing the spread of HIV remains a challenge with only small declines in incidence [5]. However, more evidence is becoming available on the benefits of undetectable viral load and early ART initiation on mortality and morbidity [6][7][8] and on lowering the risk of transmission [9][10][11][12][13], bringing hope on how further spreading of the disease can be contained through a "test and treat" approach.
However, adopting the new WHO guidelines may have challenges such as costs associated with more people on ART, infrastructure, human resources and how to monitor everyone started on ART to ensure that they adhere to medication [4]. This makes it difficult for countries to transition if they do not know what to expect if they move to test and treat.
To plan properly for transitioning to the new WHO guidelines, countries need to know the number, proportion, age and sex distribution of the additional HIV-positive individuals that will need to start ART. Most studies on the impact of change in ART guidelines on eligibility have been based on mathematical modeling [14,15] which can easily over or underestimate results depending on the model assumptions. Other studies however, have used population data, for example in Kenya, a study that estimated the impact of change in treatment guidelines using nationally representative Kenya AIDS indicator survey data fell short of measuring the differential impact of the new WHO guidelines on age and sex [16]. Our aim therefore, was to measure the impact of a "test and treat" policy on eligibility, stratified by sex and age, using population data from three countries (Kenya, Malawi and South Africa) at different stages of implementing previous WHO guidelines.

Study design
We used data from three HIV Impact in Population Surveys; Chiradzulu (Malawi), Ndhiwa (Kenya) and Eshowe (South Africa). These were population representative household based surveys and used largely similar sampling design and data collection methods, which are explained in detail elsewhere [17][18][19]. Briefly, a two stage sampling method was used to randomly select households for interviews. First, clusters were selected with a probability of clusters to be selected proportional to the number of households in the cluster (Kenya and Malawi) while for South Africa, the clusters were selected with probability proportion to population aged 15-59 making it a self-weighting sample. All three surveys used the respective country National census enumeration areas to define clusters. After first stage systematic sampling of clusters, from each cluster, households were selected using simple random selection (25 households per cluster for Malawi and South Africa, and 20 per cluster for Kenya). Households where no individual was found to provide information on the day of household visit or on two subsequent visits were replaced by another household on reserve selected using the same method. Adults aged between 15 and 59 years old residents and the visitors who had spent the previous night in the selected household were eligible for inclusion in the study.

Data collection methods
The studies were conducted between September and November 2012 (Kenya), February and May 2013 (Malawi) and July and October 2013 (South Africa). For eligible members who agreed to participate in the survey, face to face individual interviews were conducted at a private place of the household using a structured survey questionnaire which included questions on knowledge of HIV/AIDS, ART use for HIV-positive individuals only, (self-reported in Kenya and Malawi, blood tested in South Africa) and pregnancy and reproduction for women.
For those who consented for HIV testing, (parental/guardian consent was sought first from minor's guardian younger than 18 in Kenya and South Africa but parental consent was not required in Malawi where minors between 14-17 are considered mature to give own consent for HIV test) rapid HIV test and counseling was administered. Determine Rapid HIV-1/2 Antibody test (Abbott Laboratories, Abbott Park, IL, USA) was used for screening with Unigold Rapid HIV Test (Trinity Biotech PLC, Bray, Ireland) to confirm positive screening results. In case of discordant results, an ELISA test (Genetic Systems HIV-1/HIV-2 Plus O EIA, Bio-Rad, Redmond, WA, USA) was used to confirm infection. A venous blood sample was collected from all individuals who tested HIV-positive for CD4 measurement using home-based (PIMA CD4 counter, Alere PIMA, Jena, Germany) (Malawi and Kenya) or laboratory-based based FACSCalibur device (Becton, Dickinson and Company (BD)) (for South Africa). Further laboratory tests were Viral load testing (G2 real-time PCR, Biocentric, Bandol, France) (Kenya and Malawi); (NucliSens EasyQ HIV-1 v2.0 assay from Biomerieux) (South Africa) and presence of antiretroviral drugs in the blood.

Statistical methods and analysis
Data were analyzed using Stata 13 (Stata Corporation, College Station, Texas). We used Chisquared tests of association to check the distribution of demographic characteristics by sex in the population. We compared the proportion of individuals eligible for ART according to guidelines that were being used at the time of the survey in the relevant country with the proportion of individuals who would become eligible if the country were to switch to WHO 2015 guidelines. Individuals in need of ART were defined based on the relevant National ART Guidelines implemented at the time of the survey. For all the three countries, HIV-positive adults were eligible for ART if their CD4 cell count was < = 350 cells/μl, however prevention of mother to child transmission (PMTCT) guidelines differed by country. Kenya was using "Option A" guidelines where HIV-positive pregnant and breastfeeding women were started on ART if CD4 count was < = 350 cells/μl [20], Malawi was using "Option B+" i.e. lifelong ART for all HIV-positive pregnant and breastfeeding women irrespective of CD4 count [21] while South Africa was using "Option B" i.e. ART for all HIV-positive pregnant and breastfeeding women until cessation of breastfeeding [22]. In this cross-sectional analysis, among patients already on ART, we were unable to assess their historical eligibility at the time of ART initiation as we did not have access to previous laboratory or clinical records. For the purpose of this analysis, we classified HIV-positive individuals as "eligible for ART" if they either (i) had ever been initiated on ART before (according to self-report (Kenya and Malawi) and laboratory tested (South Africa)) or (ii) were not yet on ART but met the criteria for starting ART based on the guidelines for their country at the time. Malawi implemented WHO 2013 guidelines which we shall refer to as "CD4< = 500" in 2014 and implemented WHO 2015 guidelines which were shall call "test and treat" in April 2016; South Africa implemented "CD4< = 500" in January 2015 and "test and treat" in September 2016. Kenya implemented "test and treat" in July 2016.
The analysis was stratified by sex and age (grouped into 4 different categories, 15-19, 20-34, 35-44 and 45-59 years old). All study participants in all the three surveys were asked to sign a written consent if they accepted to be included in the study.
In total, 4,113 (21.7%) participants were HIV-positive. HIV prevalence was highest in South Africa 25.2% (95% Confidence Interval [CI]: 23.6-26.9) followed by Kenya 24.1% (22.7-25.6) and Malawi 17.0% (16.0-18.1). In all three countries, females had a much higher prevalence than males; this difference was greatest in South Africa with prevalence among women almost twice as high as men 30.9% (29.0-32.9) compared to 15.9% (14.0-18.0). Table 1 shows the distribution of baseline characteristics among all participants by sex. In all three countries, the proportion of adolescents (aged 15-19 years) was higher among men than among women. In Kenya and Malawi, more than 50% of the women were still breastfeeding while only 31.2% were still breastfeeding in South Africa and only less than 10% in all three countries were pregnant at the time of the survey. Among HIV-positive individuals, CD4 test results were available for 3,957 (96.2%) individuals. In all three countries, HIV positivity rate was higher among older men than women and there were more women with higher CD4 cell count of > = 500 cells/μl than men Table 2. Table 3 shows changes in proportion of people who will become eligible for ART with changes in eligibility criteria. Overall 2,816 (70.8%) were classified as "eligible for ART" based Impact of "test and treat" on ART eligibility  Given the stricter criteria for ART initiation and lower coverage, Kenya had the highest proportion of HIV-positive individuals who were not on ART; 808 (58.0%), followed by South Africa 655 (46.9%) and Malawi 429 (35.7%). Of these, the proportion of individuals with CD4 of < = 350 was no more than 36% and similar across countries. Similarly, the proportion of those not on ART with CD4 between 350 and 500 was similar across countries and was no more than 25% for each country. Overall, of the HIV-positive individuals not on ART, the proportion who would be started on ART only because of test and treat (i.e. not pregnant or breast-feeding and have CD4> 500) ranged from 28. 4

Discussion
Our results have shown that at least 30% more HIV-positive individuals (CD4>500 cells/μl) not on ART would become eligible for ART as a result of "test and treat" policy. The biggest proportionate change occurred in those 15-19 years of age. The transition to "test and treat" would, however, be much easier for countries that have already implemented the CD4< = 500 guidelines and have high ART coverage. In addition to the benefits of ART in lowering mortality and morbidity [6][7][8] and of starting treatment early [12,23,24], achieving low viral load among HIV-positive individuals has potential benefits in reducing HIV transmission [9][10][11][12]. One way of achieving lower viral load in a population is to increase ART coverage. Although some people would argue that increasing eligibility will only have a big impact in countries where ART coverage is already high, increasing eligibility especially by test and treat will itself help improve coverage especially in countries where failure to complete ART eligibility assessment is one of the factors  however could be partly explained by differences in age groups for which estimates were calculated. KAIS used 15-64 years while we used 15-59 years. While our findings show no major differences in the change in proportions eligible by sex across all the three countries if eligibility criteria are changed to treating all, there was a noticeable trend by age group. Our results show that a consistently higher proportion of adolescents (in comparison to adults) in all three countries would be in need of ART by adopting the "test and treat" guidelines. This is not surprising as adolescents are more likely to be recently infected and thus have high CD4 values, and are less likely to be aware of their diagnosis. Adolescents and adults face different challenges in accessing ART [26]. Many adolescents face stigma and discrimination from their peers [27,28]. They may find it difficult to access treatment if hospital visits coincide with school or if they are in boarding schools, or if the clinic is far and they cannot easily manage to pay for transport costs [27]. In some countries, the need for parental consent for HIV testing and treatment in those <18 years of age may be a barrier to accessing HIV care [29,30]. Adolescents are also undergoing rapid developmental changes especially neurocognitive changes that may make it particularly difficult to remain in care and adhere to treatment [31-33]. This age group therefore, will need different social support structures and, programs that are youth friendly to help them access and adhere to the treatment [4,34,35]. Without proper support structure among adolescents, there is a high risk of poor adherence with development of drug resistant HIV strains. However, the benefits of increased numbers of people on ART from a young age such as reducing number of new horizontal and vertical infections and deaths greatly outweighs the increased drug resistance that will be generated [36].
Although most countries have adopted the "test and treat" policy, for some countries, implementation has not moved at a fast pace as only 55 (40%) of low and middle income countries (LMIC) have put the policy for Treat All fully into practice and 12 (9%) LMIC have implemented treat all in a majority of treatment sites [37] perhaps due to the cost associated with increased number of individuals on ART. However, the costs and benefit analyses that have been done on increasing eligibility from CD4 350 to 500 or to all HIV-positive individuals using different models found encouraging results. Those studies found relatively low cost of expanding ART to individuals with CD4 < = 500 or everyone who is positive compared to the benefits that would be accrued when many people are started on ART [38][39][40]. However, they concluded that for the test and treat policy to have bearing positive effects on populations, it has to be accompanied by increased coverage of ART although this does not guarantee increased number of people on ART as a certain proportion of those eligible will either remain undiagnosed unless testing is expanded, or will choose not to initiate ART. As all the three countries included in this analysis have now implemented test and treat, it is encouraging to see significant increases in ART coverage from the year just before implementation of test and treat to the year after, 61.0% (2015) vs 75.0% (2017) in Kenya, 60.0% (2015) vs 71.0% (2017) in Malawi and 52.0% (2015) vs 61.0% (2017) in South Africa [41] although more research need to be done to attribute this increase to implementation of new guidelines.
Our study may be biased due to lack of objective measures on certain important variables used to determine ART eligibility, for example whether individuals were on ART or not relied on self-reported information (Kenya and Malawi), as well as self-report of pregnancy and breastfeeding (all three countries). However, risk of individuals giving inaccurate information regarding ART use was reduced as participants were asked to show their health books or pills if they said they were on ART and the agreement between ART Laboratory tested and selfreport in the South African study was found to be generally good [42]. Also, the lack of clinical information to classify eligibility by WHO clinical staging, TB or hepatitis B virus co-infection with AIDS and being in a serodiscordant couple may have underestimated the number of ART eligible individuals at the time of the survey. In addition, among the people who participated, only a small proportion had missing information in one or more of the following variables: HIV awareness, viral load, education, marital status, occupation, residency status, pregnancy and breastfeeding information. Fewer than 4.0% of HIV-positive individuals were missing information on CD4, hence assessment of eligibility in terms of CD4 was near complete.
One of the main advantages of this study was the ability to have three large studies put together which allows for comparison of different country experiences at different stages of implementing the WHO guidelines and different stages of the HIV epidemic which gave us the opportunity to compare how much impact moving from one guideline to another would have on countries.
Since all the three countries in this study have now moved to test and treat. This decision is supported by our study results-for those countries that had already implemented option B, or B+, the proportion of additional individuals to be started on ART would be relatively small, with potential benefits of reduced morbidity, mortality and transmission of moving to test and treat. For Kenya where eligibility and coverage were low, test and treat offers the opportunity to promote increased coverage. To achieve the benefits of test and treat, specific support for the groups where "test and treat" has the most impact on eligibility, particularly adolescents, is needed. We therefore, need to continue to evaluate programs to assess the fidelity of implementation of "test and treat" guidelines and whether the promised benefits are being achieved.