Objectives

The primary objectives were as follows:

1. Evaluate HTS delivered through each community-based testing strategy by measuring the proportion of children (0–14 years) tested, the number of children tested, the number of children identified as HIV-positive, and the percentage testing positive (yield).
2. Compare the number of children (0–14 years) identified as HIV-positive in each M-HIT strategy to the number of children identified as HIV-positive in standard practice (i.e. facility-based testing).
3. Determine if children testing HIV-positive in a community-based setting successfully linked to care and initiated ART within three months.

Study design

M-HIT was a non-experimental prospective project conducted in two districts in Lesotho from October 2015 to March 2018. The project implemented and evaluated seven CBT strategies that aimed to identify, enroll, and link children aged 0 to 14 years who tested HIV-positive and were not already on ART. All enrolled children were then followed-up for a minimum of three months to track linkage to care, treatment, and retention outcomes.

Study population and sampling method

The districts of Maseru and Leribe were selected as they had the highest estimated volume of unidentified HIV-positive children at the start of the project. The testing strategies covered the catchment areas of all government funded health facilities. The overall evaluation population included all individuals residing or seeking care in both districts. No sampling procedure was implemented as all individuals who received M-HIT services were included in the analysis to maximize statistical power to detect effects.

Intervention: M-HIT project overview

Community-based testing strategies.

M-HIT was a collaborative effort among five partners in Lesotho with two components: testing, and linkage to care with active follow-up for those identified as HIV-positive. The project was designed to be implementation research with testing and follow-up activities conducted by implementing partners, but also integrated into the national health system. M-HIT’s evaluation was conducted for a fixed time period to monitor activities and determine effectiveness. Two implementing partners conducted and supported the testing strategies and linkage to care with follow-up: Baylor College of Medicine Children's Foundation–Lesotho (BCMCF-L) and Population Services International (PSI). HTS was provided by nurses and counselors trained in pediatric HTS. Riders for Health (R4H) provided logistical support for the mobile outreach clinics, and Clinton Health Access Initiative (CHAI) conducted program planning, management, linkage to care documentation, and the evaluation activities. Finally, the Lesotho MoH–at central, district, and health facility level–were key implementing partners, advisors, and stakeholders.

The seven M-HIT strategies are described in Table 1. Every strategy focused on providing HTS for children 0–14 years and pregnant and lactating women at a community level; however, HTS were offered to individuals of all ages in all strategies. Testing strategies were either targeted—specific attempts to find populations of children with higher HIV prevalence, or blanket—providing HTS to all children regardless of risk factors.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of community-based testing strategies.

https://doi.org/10.1371/journal.pone.0236985.t001

Testing and follow-up procedures

HTS eligibility for all age cohorts was based on the MoH guidelines [33]. Eligible individuals included anyone with an unknown status, >18 months. Eligibility also included all adolescents (in this case the MoH defined this cohort as those 10–19 years) and adults who reported being sexually active that had not been tested in the past year. Additionally, it included those who reported recent or ongoing risk of exposure who had not been tested in the past three months.

Within the M-HIT project, both index testing strategies were the exception to the aforementioned HTS eligibility requirements as they tested all household members regardless of testing history. As stated in Table 1 in the description for Facility Index testing, all household members were offered HTS once consent from the ART patient was received. Similarly, during D2D Index testing, all household members were immediately offered HTS after someone self-reported as HIV-positive. The reason for expanding HTS eligibility in this context is due to the belief that these individuals were at higher risk of infection.

Eligibility for infants (0–18 months) was based on the HIV status of the mother. If the mother was not present and her status was unknown, the infant was administered a rapid test to determine exposure status. Infants testing positive were referred to a health facility to receive confirmatory DNA PCR testing and were captured in the project’s mHealth system for client follow-up. For HIV-exposed infants, EID algorithms were followed, although DNA PCR testing was not available for all strategies and is therefore not reported in this paper. During follow-up it was discovered that some children had a previous HIV-positive diagnosis, but had not yet initiated on treatment; therefore, they remained in the follow-up cohort.

Procedures for consent were also consistent with MoH guidelines and was obtained by implementing partners at the time of testing [33]. This required verbal consent for anyone 12 years and older (the legal age of consent in Lesotho) and written consent from the parent or guardian present during HTS for children under 12 years.

Prior to Lesotho’s adoption of the Treat All policy (all HIV-positive individuals to begin treatment immediately after diagnosis), children who tested positive at an M-HIT testing event were referred to their preferred health facility for CD4 testing or treatment initiation. Per national treatment guidelines, all HIV-positive children under 5 years were eligible for ART initiation, as were those over 5 years with a CD4 count <500 and those coinfected with tuberculosis or hepatitis B [33]. After the national implementation of Treat All in June 2016 some MOCs were able to offer immediate ART initiation to children on-site. Children were to return to the facility two weeks after initiating ART and subsequently every month to receive drug refills.

Initial client follow-up was conducted by lay counselors and commenced when a newly identified HIV-positive client did not present at the facility within two weeks of a positive test. Counselors attempted to contact the parent or guardian through the mobile phone number provided. If they were not reached within two weeks or refused treatment for their child, a counselor or village health worker would visit their home for further HIV education and counseling.

mHealth enrollment and mobile money system (M-pesa)

The project utilized a customized mHealth system to enroll the newly identified HIV-positive children at the point-of-diagnosis into a follow-up cohort. Verbal consent for enrollment was required from all clients or the parent or guardian of children under 12 years of age prior to enrollment. M-Pesa, a mobile money system operated by Vodacom Lesotho, was utilized to distribute four unconditional e-vouchers to enrolled clients to serve as reimbursement for their transportation costs from their home to a preferred health facility for treatment. These e-vouchers ranged between 12 LSL—50 LSL each (~USD$1–$5) and their electronic delivery aligned with the client’s ART initiation appointments. Due to delays in the development of the mHealth system, electronic enrollment began in December 2015 and concluded in November 2017, allowing for a further three months of follow-up data collection. Therefore, all follow-up cohort analysis only includes the HIV-positive children identified in this time period.

Data collection

Monthly service delivery data on the number of rapid tests conducted and the corresponding results, disaggregated by age and sex when available, as well as other key indicators were collected from implementing partners for all strategies. BCMCF-L submitted data on MOCs in aggregate while PSI provided individual-level data for all other testing strategies. Lesotho’s Ministry of Health data warehouse, DHIS2, provided aggregated testing numbers and results for all facility-based testing that occurred in the same period.

General client information and testing history was collected by project healthcare workers into the mHealth system on all HIV-positive children at the time of testing. Subsequently, data collectors routinely visited health facilities to document facility attendance, ART initiation, and refill dates from facility ART registers.

Linkage to care and ART initiation were measured as a single documented pre-ART or ART initiation visit at an MOC or health facility within three months of the HIV-positive test date. Similarly, retention in care was also documented by ART refill visits and was defined as being engaged in care by a refill date at least three months after initiation.

Ethics approval was obtained from Lesotho’s MoH Research and Ethics Committee (ID68-2015), as well as Chesapeake Institutional Review Board (MOD00240045) for the duration of the M-HIT project and evaluation. Informed consent for the evaluation was waived by both IRBs as consent for services was already received by the implementing partners during the testing and follow-up process. While identifiable information was collected on patients in order to conduct follow-up on linkage to care, once completed the dataset was de-identified for analysis.

Data analysis

The number tested and the number identified as HIV-positive are presented as monthly averages for each testing strategy. Yields are presented as proportions (total tested positive/total tested). Descriptive statistics are used to report on linkage and ART initiation within three months of HIV-positive test date, as well as retention in care, defined by an ART visit at least three months after initiation. Time-to-event methods are used to examine ART initiation with Kaplan–Meier curves graphically displaying time from HIV-positive test date to ART initiation. Cox proportional hazards models calculating hazard ratios (HRs) and 95% confidence intervals (CIs) are shown, adjusting for age, sex, and district and stratified by the pre/post Treat All guideline change. All analyses are conducted in Stata/SE 15.1.

Pediatric testing, positive tests, and yields by strategy

During M-HIT implementation, an average of 7,351 tests were conducted across all community-based strategies for all age groups, among which 3,399 (46%) tests were on children 0–14 years. An average of 141.65 individuals tested positive each month; 9% were children. Yields were 3.73%, 0.98%, and 0.38% for adults, adolescents (15–19 years), and children, respectively (Table 2). MOCs tested the most children and identified the largest number of HIV-positive children, followed by D2D, Roadside testing, and then D2D Index strategies. The Facility Index and Targeted testing strategies contributed the smallest numbers in terms of average number of children tested and identified as HIV-positive. Semi-Static did not have any HIV-positive results among the children tested. The pediatric yield varied by strategy from 0.00% to 0.66%. During the same time period, MoH facility-based pediatric yields were 0.72% with an average of 8,526 monthly tests conducted in the two districts.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Average monthly number tested, proportion tested, number tested positive, and yields; stratified by age group for all M-HIT strategies.

https://doi.org/10.1371/journal.pone.0236985.t002

The majority of all children tested in M-HIT were between 5–14 years (55%) and female (54%). Table 3 shows the breakdown for each strategy. All strategies except MOCs were successful in finding older children to test, with 60–94% of the total children tested being over the age of 5. Yields were similar between age groups, at 0.35% and 0.41% for under and over 5 years, respectively. Differences in yield by sex were also negligible, with 0.40% of males testing HIV-positive compared to 0.37% of females.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Average monthly number tested, proportion tested, number tested positive, and yields among children; stratified by age and sex.

https://doi.org/10.1371/journal.pone.0236985.t003

Previous test history was known for all strategies except MOC. Among those with known HIV testing history, 75% were first-time testers and 86% of children testing HIV-positive had never previously been tested for HIV. When observing test history by age, 81% and 72% of children under and over 5 years were first-time testers, respectively. For those testing positive, 93% of children under 5 had never been tested before, as well as 83% of children over 5.

Linkage to care

A total of 227 HIV-positive children were identified through all M-HIT strategies in the time period examined, among which 173 (76%) were enrolled in the follow-up cohort. Among those enrolled, 20% were 0–1 years; 24% 2–4 years; 35% 5–9 years; and 21% 10–14 years. Seventy-one percent (123/173) linked to care within three months, with all children except one initiating ART. Eighty-two percent of those initiating were still engaged in care at three months. An additional 14 children linked to care within six months and five children linked to care before the study concluded, for an overall linkage outcome of 82%. All 19 of these children also initiated ART.

When the ART guidelines changed nine months into the project to adopt the Treat All policy, there was an expected increase in ART initiations as all children became eligible (eligibility was not documented prior to policy change and it cannot be determined whether eligible initiation rates did increase). Fifty-seven percent of children initiated before Treat All compared to 70% after (p = 0.002). However, children engaged in care at three months was comparable before and after Treat All (85% vs 81%, p = 0.602). The median (Interquartile Range) time to initiation was 25 (6–51) days before Treat All and 2 (0–19) days after (p<0.001).

By design all enrolled children should have received four e-vouchers; however, only 74% did due to technical challenges. This allowed for a natural experiment to assess whether receipt of e-vouchers affected linkage to care or ART initiation. As shown in Fig 1, prior to Treat All implementation there was no statistically significant difference in the time to ART initiation between those that did and did not receive a transportation voucher (p = 0.429). However, after Treat All those that received a voucher were more likely to both initiate ART and initiate more quickly (p = 0.007). The adjusted HR for children after Treat All was implemented was statistically significant for those receiving a voucher (aHR: 2.00, 95% CI 1.18–3.38). This means that children who received a voucher had a rate of ART initiation two times faster than children who did not receive the voucher.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Kaplan-Meier curves for time to ART initiation by voucher status: Pre and post-Treat All.

https://doi.org/10.1371/journal.pone.0236985.g001

Challenges

There were several limitations to this evaluation. First, we were assessing strategies that were part of ongoing implementation work through the MoH and their partners. As such, this could be considered implementation research whose advantage of the real-world setting effectively identifies feasibility challenges and seeks to improve programs [60]. However, the disadvantage is working outside of a controlled study setting. For example, 24% of all children who tested positive through an M-HIT strategy were ‘lost’, meaning they were not enrolled into the project’s mobile application and therefore also excluded from the follow-up cohort. The lack of enrollment occurred due to healthcare worker error and a shortage of mobile phones among staff as the project rapidly scaled-up during certain time periods. Additionally, some children did not receive an e-voucher due to network failures in remote locations, client errors in M-Pesa registration, or general system errors. Distance to clinic may have been a confounding factor with linkage to care as well, as these same children may have had a more difficult time getting to the nearest health facility for services.

MOCs operated through government funded health facilities who also supplied most commodities and medicines. Challenges were faced in the consistent provision of these materials. Additionally, the local communities of each MOC were responsible for providing a venue to host the clinics. This often resulted in inadequate spaces that lacked privacy for clients to receive sensitive services. Many also lacked an indoor waiting area that proved problematic in the winter months, with some MOCs being canceled due to inclement weather.

Another challenge was that wide variations in national pediatric HIV estimates were experienced throughout the life of the evaluation (19,000 in 2015; 13,000 in 2016; and 16,000 in 2017). This affected project targets and the general understanding of the pediatric HIV landscape in Lesotho [31].

One limitation on the analysis was the spatial and time overlap among the community and facility-based testing, which made it impossible to compare between strategies; nor could it be concluded whether a child tested in the community would have eventually been tested at the facility. Given that the testing strategies were being implemented in overlapping areas, disentangling the effects of each one alone and from facility-based testing would require more variation in the intensity of the implementation of the strategies over space and time. The small number of children testing HIV-positive also limited our power to detect differences.

An important consideration in determining the feasibility of CBT is to understand that resources and costs associated with such work. This paper does not address these areas in any detail; however, it is highly recommended these be thoroughly explored for each potential strategy as costs can vary greatly.