Linkage to HIV care after home-based HIV counselling and testing in sub-Saharan Africa: A systematic review

Background: Home-based HIV counselling and testing (HBHCT) has the potential to increase HIV testing uptake in sub-Saharan Africa (SSA) but data on linkage to HIV care after HBHCT are limited. We conducted a systematic review of linkage to care after HBHCT in SSA. Methods: Five databases were searched for studies published between 1 st January 2000 and 19 th August 2016 that reported on linkage to care among adults newly identified with HIV infection through HBHCT. Eligible studies were reviewed, assessed for risk of bias and findings summarised using the PRISMA guidelines. Results: Fourteen studies from six countries met the eligibility criteria; nine used specific strategies (point-of-care CD4 count testing, follow-up counselling, provision of transport funds to clinic, and counsellor facilitation of HIV clinic visit) in addition to routine referral to facilitate linkage to care. Time intervals for ascertaining linkage ranged from one week to twelve months post-HBHCT. Linkage ranged from 8.2% [95% confidence interval (CI), 6.8%-9.8%] to 99.1% (95% CI, 96.9%-99.9%). Linkage was generally lower (<33%) if HBHCT was followed by referral only, and higher (>80%) if additional strategies were used. Only one study assessed linkage by means of a randomised trial. Five studies had data on cotrimoxazole (CTX) prophylaxis and twelve on ART eligibility and initiation. CTX uptake among those eligible ranged from 0% to 100%. The proportion of persons eligible for ART


Introduction
Access to antiretroviral therapy (ART) in sub-Saharan Africa (SSA) has expanded considerably but AIDS-related mortality remains high [1]. A major cause of this mortality is the late presentation of patients for treatment [2]. Early ART initiation is dependent on early HIV diagnosis, and prompt linkage to and retention in care [3]. HIV counselling and testing (HCT) is essential in expanding HIV prevention and treatment services [4]. However, HCT uptake in SSA remains low [5]. For instance, the proportion of HIV-positive adults in SSA who are aware of their HIV status has been estimated to be only 60% [6]. In order to expand access to HIV testing in settings with generalised HIV epidemics, the World Health Organisation (WHO) recommends community-based HCT with linkage to prevention, care and treatment services, in addition to facility-based HCT [7].
In the community-based HCT model, services are delivered through mobile, workplace-, school-, and home-based approaches thus removing structural, logistical, and social barriers to HCT [8]. Community-based HCT may also be delivered as part of multi-disease campaigns that involve intensive community mobilization lasting 1-2 weeks followed by mobile HIV testing, often coupled with other preventive medical services (campaign HCT) [9].
Under the home-based HIV counselling and testing (HBHCT) approach, HCT services are conducted by trained HCT service providers in the client's home [10]. HBHCT may be provided to everyone in a community through a door-to-door approach or to household members of known tuberculosis (TB) or HIV-positive patients [4,10]. According to WHO, there were at least 39 HBHCT programmes in 10 SSA countries by early 2011 [10]. A systematic review published the following year found 21 studies that had reported on the uptake of HBHCT in SSA [11]. Since then, several studies have reported successful implementation of HBHCT in rural [12][13][14][15][16][17][18][19][20][21][22][23][24] and urban [19,23] populations in SSA. HBHCT is highly acceptable and has the potential to substantially increase HIV testing uptake in SSA [11]. It is cost-effective at reaching previously untested persons compared with other HCT models in settings with high HIV prevalence [25,26], promotes equitable access of services [27] and may help to promote HCT among couples [28] and prevention of mother-to-child HIV transmission [29]. Importantly, HBHCT facilitates early HIV diagnosis and therefore provides an opportunity for early linkage to care [29]. These attributes highlight the potential of HBHCT as an effective platform for HIV prevention and population-based test-and-treat strategies.
Despite these advantages, few data are available on linkage to care after HBHCT particularly among newly identified HIV-positive persons or on the effectiveness of strategies to increase linkage after HCT [30]. In the absence of interventions to facilitate linkage to care, individuals that test HIV positive through HBHCT may find it more challenging to enter care compared to those identified in facility-based HCT [31]. This is because HBHCT is more likely to reach socio-economically disadvantaged populations that have difficulty accessing healthcare services compared to facility-based HCT [32]. In order to identify effective linkage strategies, data are specifically required on linkage to care soon after HBHCT. The reasons for this include the current WHO recommendation to initiate ART among all HIVpositive adults regardless of WHO clinical stage and at any CD4 count [33]; increasing use of HBHCT in Africa; and growing importance of early treatment for improved clinical outcomes [34][35][36] and HIV prevention [37]. A recent systematic review on linkage to care following community-and facility-based HCT [9] included 10 HBHCT studies, but did not distinguish linkage outcomes between HBHCT and campaign HCT, between newly and previously diagnosed HIV-positive individuals, or between children/adolescents and adults.
Individuals who previously tested HIV-positive and have not yet linked to care are likely to differ from newly identified patients with regard to barriers that may prevent service uptake [12]. Similarly linkage to care among children/adolescents may be influenced by factors that are unique to this population [38,39].
The specific objectives of our review were to: estimate the proportion of individuals in SSA linking to care within 12 months among those who were newly diagnosed with HIV; the proportion initiating daily cotrimoxazole (CTX) prophylaxis (i.e. the people who initiated daily CTX prophylaxis among those who linked to care and were eligible for CTX); and the proportion initiating ART (i.e. the people who initiated ART among those who linked to care and were eligible for ART); and to summarise data on the strategies that have been used to increase linkage to care after HBHCT.

Search strategy
We searched five databases (Medline, Embase, Global Health, Web of Science, and Africa-Wide information) for studies published between 1 st January 2000 (time at which roll-out of ART programmes began in SSA [11]) and 19 th  were eligible if they were conducted in SSA, and had original data on linkage to care among adults (≥18 years) newly identified with HIV infection through HBHCT, defined as HCT services offered in an individual's home. Studies whose study populations included persons <18 years were eligible but only data for participants aged ≥18 years were utilised to estimate linkage and other outcomes. Studies for which the required information was not published but might have been collected were identified and the corresponding authors approached with requests for additional data. Where two or more eligible articles reported on similar or overlapping populations, the article with the most complete data was included. Review articles were excluded but their bibliographies as well as those of the identified articles were manually checked to identify any additional studies. Conference abstracts were excluded. All potentially eligible papers were then subjected to full text screening.

Data extraction and synthesis
A data extraction form was used to collect the following information from each eligible article: first author's name, publication year, country and setting where study was conducted, study population, sample size, study design, definition of linkage to care, strategies used to promote and time for evaluation of linkage to care. We also obtained the number of HIVpositive adults who were newly diagnosed, and, among those, the numbers who linked to care (as defined in the respective studies), and were eligible for, and initiated, CTX prophylaxis and ART (based on national guidelines that were in use at the time of the respective studies).
Risk of bias in the included studies was assessed using a component approach, similar to the Cochrane Collaboration's [40], and based on three items: selection bias, outcome ascertainment, and attrition. The results were summarised using the PRISMA guidelines [41].
We used the reported data to calculate the proportions [and their 95% confidence intervals (CI), using the Clopper-Pearson method] who linked to care, initiated CTX prophylaxis, were eligible for and initiated ART. The denominator for linkage was all newly diagnosed HIV-positive adults (≥18 years) who had a potential minimum follow-up period corresponding to the time point when linkage was assessed (i.e. including those who outmigrated, died, or were lost to follow-up, but excluding those who entered the study at a later date so had a shorter potential follow-up period). We did not stratify our linkage estimates by individual-level factors that may potentially influence referral uptake, such as HIV disease stage or CD4 count at the time of HIV diagnosis, as these data were not available for most of the studies. However, we compared linkage estimates between studies conducted under different CD4 ART eligibility thresholds i.e. ≤250 cells/μL versus ≤350 cells/μL. The denominator for ART eligibility was all individuals who linked within the specified time period, and those for initiation of CTX prophylaxis and ART were all individuals who linked and were eligible for CTX and ART respectively. We did not perform a meta-analysis because the identified studies varied widely with regard to design, setting, definition of linkage to care, the time points of and method for ascertaining linkage, and with regard to the strategies used to facilitate linkage.

Summary of study objectives and populations
The included studies aimed to assess acceptability of HBHCT [18, 19, 21-24, 46, 47], and HIV prevalence [19], linkage to care [12-14, 18, 19, 21-24, 43-45], uptake of ART [13,14,18,24,44,47], and viral suppression [13,14,18], in the context of HBHCT. Two studies compared HBHCT and mobile HCT with regard to uptake and cost of HCT, HIV positivity rates and linkage to care [21,22]. Three observational studies [13,18,44] and one randomised trial [14] were designed to evaluate the effect of specific interventions on linkage to HIV care, uptake of ART, and other outcomes. Except for one study in which HBHCT was provided to household contacts of TB index patients [23], all the other studies used the doorto-door HBHCT approach. Study participants mostly comprised individuals aged ≥13 years.
However, two studies also recruited children (<13 years) [19,21]. With the exception of two studies that recruited newly identified HIV-positive individuals [22,23], participants in other studies were a mixture of previously and newly identified HIV-positive patients.

Risk of bias
Only two [13,18] studies had a low risk of bias for all assessed items ( Table 2). Risk of selection bias was low (≥80% HBHCT coverage) in four studies [13,18,19,44]. Risk of attrition bias was low (≥80% participant retention) in eight [13,14,18,23,24,[44][45][46] of ten studies in which participants were followed. Self-reported linkage to care was confirmed by tracking referrals and review of records at the referral clinic in only two studies [43,45]. In the first study [43], no information was reported on the proportion of participants for whom clinic records were not found. In the second study [45], clinic records were found for only 71% of the tracked referrals and self-reported data was used to ascertain linkage for the rest of the participants. Self-reported linkage to care was confirmed by review of documentation issued to patients by HIV clinics (e.g. clinic cards) in three studies [13,14,18]; linkage was not verified with the HIV clinics. In five studies [12,21,22,24,47], ascertainment of linkage to care was based on data from HIV clinics in the areas where the studies were conducted; participants who may have linked to HIV clinics outside of the study areas were not tracked.
Only self-reported data was used in the rest of the studies [19,23,44,46].

Uptake of CTX prophylaxis and ART
Five studies conducted in Kenya [19], Uganda [43,44], and Uganda and South Africa [13,14] had data on initiation of CTX prophylaxis among eligible individuals who linked to care.
CTX prophylaxis is only recommended for patients with CD4 count ≤200 cells/μL, WHO stage 3 or 4 or HIV/TB co-infection in South Africa [48] but is routinely provided to all HIVpositive persons irrespective of CD4 count or WHO disease stage in Uganda [49] and Kenya [50]. Of the studies conducted in Uganda and Kenya, additional interventions to facilitate referral uptake were offered in all except one. CTX uptake in these studies ranged from 78.2% (95% CI, 69.3-85.5%) [13] to 100% [14,43,44] (Table 1). CTX uptake was also high [90.6% (95% CI, 87.3%-93.2%)] in the study that offered referral only [19]. Interventions to facilitate linkage were offered in the two studies that were conducted in South Africa. Uptake of CTX among patients with CD4 count ≤200 cells/μL in these studies ranged from 0% [14] to 33.3% (95% CI, 4.3-77.7) [14].
ART initiation rates were highest (≥90%) in the two studies in which HIV care services were provided through community-based research clinics [43,47].

Discussion
HBHCT is increasingly being used in SSA, and an effectively conducted HBHCT strategy would be a key precondition for HIV control programmes that propagate a test-and-treat approach for HIV prevention [45]. The success of such programmes will partly depend on their capacity to achieve high levels of linkage to care following HIV diagnosis [16]. Hence, it is necessary to identify and set up strategies that will effectively link persons identified with HIV through HBHCT to care and treatment. As observed in this review however, only a few studies have investigated linkage to care among adults newly identified with HIV through HBHCT in SSA. Linkage to care was below 33% in five of six studies where participants were only referred for care with no further interventions to facilitate referral [12,19,21,23,47], a figure that is lower than those for client-initiated facility-based HCT (61%) and provider-initiated facility-based HCT (55%) [9]. With the exception of two studies [22,24], studies that used additional linkage strategies recorded moderate (>50% to <80%) [43,45,46] to high (≥80%) [13,14,18,44] levels of linkage. These trends remained irrespective of the CD4 ART eligibility threshold at the time of the studies. In general, linkage to care was highest when participants were offered POC CD4 count testing and follow-up counselling [13,14,18]. These findings suggest that HBHCT coupled with interventions to facilitate referral uptake may achieve similar or even higher linkage compared to facility-based HCT.
WHO recommends CTX prophylaxis for all adults with WHO stage 3 and 4 HIV disease and/or with a CD4 count ≤350 cells/mm 3 , and regardless of CD4 count in settings where malaria and/or severe bacterial infections are highly prevalent [33]. For this reason, CTX prophylaxis is an essential component of HIV care in many settings in SSA, and its uptake may be used as an indicator of access to HIV care [19,51]. However, only a small number of studies included in this review had data on the uptake of CTX. In Kenya and Uganda where routine CTX prophylaxis is recommended irrespective of CD4 count or clinical stage [49,50], uptake was high (>70%) irrespective of whether or not additional interventions were applied. This may be because CTX is widely available, inexpensive, and simple to use [52].
In contrast and in spite of facilitated linkage, uptake of CTX prophylaxis among those eligible (based on guidelines that were in use at the time of the studies) in South Africa was low (≤33%). The reasons for this are not clear. However previous studies have found irregular supply and lack of stocks of CTX, lack of awareness among health care workers, and perceived low priority of CTX prophylaxis due to the absence of a reporting requirement, to be some of the barriers to implementation of CTX prophylaxis policies [53].
Consistent with previous findings [11], significant proportions of HIV-positive persons identified through HBHCT were still ineligible for ART (based on national guidelines that were in use at the time of the respective studies). The finding that ART uptake was highest in the studies where services were provided through community-based research clinics may be attributed to such clinics being more accessible and less prone to limitations that characterise many public sector ART care programmes in SSA, including the requirement for several visits to prepare patients for ART [14], crowded, busy and unwelcoming clinics [54], nonfunctioning laboratories [54], and inadequate supply or lack of antiretroviral drugs [55].
Additionally, some research clinics are likely to have close and long standing relationships with communities in which they are located.
HBHCT studies with facilitated linkage have been shown to achieve higher ART initiation rates among participants who link to care compared to those without facilitated linkage [9].
However, ART initiation rates were high in some but not all studies with facilitated linkage described in this review. Additionally, some studies without facilitated linkage achieved higher ART initiation rates [47] than those with facilitated linkage [13,14,18,22,24,44,45]. It is likely that clinic level factors such as those mentioned above as well as individuallevel confounding factors such as HIV disease stage may be more important in influencing events after linkage to care. Also, people who link to care in the absence of facilitated linkage may be more motivated to receive care. This review has some limitations. The review was limited to SSA. Even so, the number of relevant studies found is small and represents only six countries, four of which are in Southern Africa. This may limit the generalizability of the findings. The methodologies used in the identified studies varied widely, making it impractical to combine findings from individual studies into a pooled analysis. Outcome assessments in most of the included studies were based solely on self-reports or records in HIV clinics within the respective study areas; hence linkage to care may have been overestimated in the case of self-reports or underestimated if some individuals linked to clinics outside of their communities. Importantly, most of the studies that included interventions to facilitate linkage to HIV care were observational, determining intervention effects without control groups. A major limitation of observational studies is that it is difficult to account for the effects of confounding factors such as HIV disease stage, fear of stigma, healthcare seeking behaviour, and familiarity with the health care services. Moreover, in some of the studies, two or more interventions were delivered concurrently, making it difficult to distinguish the effects of each. Randomised trials represent the gold-standard methodology in the evaluation of an intervention including its likely effect size [56]. It is therefore desirable to confirm the impact and establish the cost effectiveness of these interventions in randomised controlled trials.
Indeed, a number of trials have been designed to investigate the effect of different interventions on linkage to care among persons that test HIV positive through HBHCT [57-61]. Findings from these trials are expected in the near future and may provide more robust data on the impact of interventions on linkage to care after HBHCT.
Except for one study in which some of the newly identified HIV-positive individuals were offered immediate ART irrespective of CD4 count [24], all the other studies were conducted in the context of old ART eligibility criteria i.e. CD4 counts of ≤200 cells/μL, ≤250 cells/μL, ≤350 cells/μL or ≤500 cells/μL and by implication, prolonged pre-ART care periods. Therefore, the extent to which the findings of this review are relevant to settings in which the new WHO recommendation of immediate ART initiation irrespective of CD4 count [33] has been adopted is not clear. A future review of the evidence on linkage to care after HBHCT under the new treatment guidelines will be needed.
In conclusion, we found that only few published studies investigated linkage to HIV care among adults newly diagnosed with HIV through HBHCT in SSA. In general, HBHCT without additional intervention strategies to increase service uptake achieved inadequate linkage while HBHCT combined with some kind of additional strategy seemed to achieve higher linkage. There is a need to confirm the impact of the most promising linkage strategies through randomised controlled trials before they can be recommended for large scale adoption. Moreover, it will be important to demonstrate the effectiveness of linkage strategies under the new WHO treat-all policy.    HBHCT coverage could not be estimated because the total number of persons in the target population was not reported.

Unclear risk
Clinic-verified data. No information on persons who may have linked to clinics outside the study area. Unclear risk

No participant followup
Becker, 2014 [46] HBHCT was offered to all eligible participants in one village and approximately one-third in two other villages (no information on how participants in these two villages were selected). High risk Self-report for all participants. High risk 2% (1/46) loss to follow-up. Low risk Nakigozi, 2011 [43] Annual HIV testing coverage in the cohort is >90%. However, 21% of the persons who tested HIV-positive were excluded from the analysis because they had either refused to learn their HIV results (3%) or received their HIV result less than six months before data-set closure (18%). High risk HBHCT coverage could not be estimated because the size of the population targeted for testing in the participating communities was not reported. Unclear risk Self-report & review of documentation issued to patient by the HIV clinic. Low risk 3% (40/1325) loss to follow-up. Low risk a In population-based studies, there was low risk of bias if HBHCT coverage (defined as the number of persons accessing HBHCT out of the total resident population) was ≥80%, high risk if HBHCT coverage was <80% and unclear risk if there was no information on coverage. In non-population based cohort studies, there was low risk if participants were randomly selected, high risk if the selection was non-random and unclear risk if there was insufficient information on participant selection. b There was low risk of bias if ascertainment of linkage to care was by both self-report & examination of HIV clinic records or documentation issued to patients by the HIV clinic for ≥80% of the participants, high risk if ascertainment was by self-report only and unclear risk if there was insufficient information on ascertainment of linkage outcomes for some study participants. c There was low risk if retention of HIV-positive persons identified through HBHCT was ≥80%, high risk if retention was <80% and unclear risk if information was not available. d Only applicable to HIV-positive individuals who were newly identified through HBHCT. e Only applicable to newly identified HIV-positive persons who were referred to care at least 12 months before the end of the first phase of the trial i.e. May 2014. f Only applicable to HIV-positive individuals who were not engaged in care at baseline.