Globally, an estimated 1·4 million children (younger than 15 years) were living with HIV in 2023, with only 58% receiving treatment.¹ Progress towards reaching the UNAIDS 95–95-95 targets for children living with HIV continues to lag behind that for adults.² Population HIV impact assessments (2015–17) across seven sub-Saharan African countries found that only 61% of children living with HIV were diagnosed.³ In 2023, diagnosis among children living with HIV was at 66% compared with 86% among adults.¹ There were an estimated 72 000 children living with HIV in Uganda and 58 000 in Zambia, with 23% in Uganda and 29% in Zambia not yet on treatment.¹

Children are affected by household and health systems factors that make it challenging to receive HIV testing.⁴ These factors include underestimation of HIV risk by parents and health-care workers, undiagnosed parents living with HIV, asymptomatic infection, dependence on caregivers to access health care, unstable guardianship, transport costs, and risk of parental HIV status disclosure.^4–8 At the facility level, barriers include the workload of health-care workers, restrictive age of consent regulations, distance to health facilities, overlapping school and clinic hours, and lack of training for health-care workers to discuss HIV testing with parents and children.^4,6,9,10

Starting in 2017, index testing services, which identify and test biological children of people living with HIV, expanded rapidly, identifying a large number of children living with HIV.¹¹ Despite declines in paediatric HIV testing during the COVID-19 pandemic, index testing remained an effective case identification strategy.¹² Typically, people living with HIV have their children tested at a health facility or through a home visit by a health-care worker. Studies from 2018 to 2019 in Zimbabwe and Kenya showed promise for a third option, caregiver-assisted HIV self-testing (CG-HIVST), whereby the caregiver screens the child for HIV using an oral HIVST kit.^13,14 During the COVID-19 pandemic, the US President’s Emergency Plan for AIDS Relief collaborated with ministries of health to support the use of oral HIVST kits for children, given the restricted access to health services.¹⁵

Initial studies from Kenya and Zimbabwe of CG-HIVST in children explored caregivers’ perceptions and results interpretation, oral HIVST kit validation, and uptake among people living with HIV.^13,14,16,17 Few caregivers had heard of oral HIVST, yet perceived potential benefits to include convenience, privacy, reduced travel costs, administration ease, and child comfort.¹⁴ A validation study of oral HIVST in children compared HIVST results read by health-care workers (100·0% sensitivity and 99·9% specificity) with blood-based diagnostic tests.¹⁶ Another study showed low uptake (5·6%) of CG-HIVST.¹³ Although initial studies showed caregivers can correctly interpret screening results (≥97%) with or without a demonstration,¹⁷ questions remained regarding feasibility and acceptability¹⁴ and whether unintended social harms might result. We aimed to estimate the acceptability, feasibility, effectiveness, and potential risk of social harm of oral CG-HIVST for children offered as an index-linked testing approach to people living with HIV in Uganda and Zambia.

From Feb 1 to Oct 31, 2021, 12 998 caregivers were screened in Uganda and Zambia (7504 [57·7%] in Zambia and 5494 [42·3%] in Uganda). After applying the exclusion and inclusion criteria before screening for intimate partner violence, 4268 caregivers were eligible for further evaluation (figure 1). Of these 4268 caregivers, 4045 (94·8%) were eligible after screening for intimate partner violence. 65 (3·8%) of 1721 caregivers in Zambia and 158 (6·2%) of 2547 caregivers in Uganda had a history of violence in the home and were referred to standard-of-care HIV testing and social services. Of the 4023 eligible caregivers after final age verification, 3903 (97·0%; 1609 [41·2%] in Zambia and 2294 [58·8%] in Uganda) consented to accept HIVST kits to screen their child for HIV and were enrolled in the study with their 7601 children, with a mean of 1·94 children per caregiver (range one to eight children per caregiver).

Among caregivers, median age was 36 years (range 18–76), 3094 (79·3%) of 3903 were female, and 809 (20·7%) were male (table 1). Most caregivers presented to ART clinics. Among the children, median age was 7 years (range 1–14 years), 3779 (49·7%) of 7601 were female and 3822 (50·3%) were male. In Uganda, 1386 (28·5%) children were non-biological children of caregivers. 1779 (23·4%) children had male caregivers (table 2).

Of the 7601 children who enrolled and received CG-HIVST, nearly all results were returned (4766 [97·9%] of 4866 in Uganda and 2647 [96·8%] of 2735 in Zambia). Across all demographic characteristics, result return was 95% or more in Uganda and 94% or more in Zambia. Of the 4766 HIVST results returned in Uganda, 3932 (82·5%) were returned within 3 days, 4489 (94·2%) within 1 week, and 4721 (99·1%) within 1 month. Of the 2647 HIVST results returned in Zambia, 2643 (99·8%) were returned within 3 days, 2645 (99·9%) within 1 week, and 2647 (100%) within 1 month. Overall, 119 (1·6%) of 7413 children had a reactive HIVST result and required facility-based confirmatory HIV testing. A higher proportion of reactive HIVST results was observed in Uganda (98 [2·1%]) than in Zambia (21 [0·8%]). 116 children with reactive HIV self-tests received confirmatory testing (three were lost to follow-up in Zambia).

Among the 116 children with confirmatory testing, 43 were confirmed HIV-positive, giving an HIV prevalence of 0·7% (32/4766) in Uganda and 0·4% (11/2647) in Zambia. All children with confirmed HIV-positive results were linked to HIV treatment, with 42 (97·7%) initiating same-day ART. HIV prevalence was similar among biological and non-biological children in Uganda (table 3). The 43 newly diagnosed children living with HIV shared similar demographics to the general population (table 2).

The positive predictive value of HIV self-tests was different in Uganda and Zambia (table 3). In Uganda, the proportion of total HIVST results determined to be false positive decreased from 3·3% (45/1373) in study weeks 1–8 to 0·6% (21/3393) in study weeks 9–26. Among reactive HIVST results in Uganda, the odds of a false-positive result was significantly lower after the second refresher training (odds ratio 0·36 [95% CI 0·15–0·86]; p=0·021; figure 2). The positive predictive value was 23·7% before and 46·2% after the second refresher training; HIV positivity was 1·0% (95% CI 0·6–1·7) before and 0·5% (0·3–0·8) after the second refresher training.

The caregiver survey was administered to 2251 (98·1%) of 2294 caregivers in Uganda and 469 (29·1%) of 1609 caregivers in Zambia, of whom 86 caregivers in Uganda and 14 in Zambia had a child with reactive or invalid HIV self-tests (appendix p 1). There were no reports of social harm to children screened using HIVST in Zambia or Uganda, and no reports of social harm to caregivers in Uganda. In Uganda, seven (0·3%) of 2251 caregivers reported the following types of issues: the child was initially scared but then relaxed and agreed to testing, other children or the partner were not interested in testing and then either agreed or the caregiver tested when the partner was away, or the child discarded the liquid from the test tube. 12 minor adverse events were reported in 11 (0·4%) of 2720 children, including gum pain (six cases), bleeding (five cases), and itching (one case; appendix p 1).

Among caregivers surveyed, 2650 (99·1%) of 2674 reported they would use an HIV self-test again to test a child. When asked if they would recommend HIVST to other parents, 2185 (99·1%) of 2205 caregivers in Uganda and 430 (91·7%) of 469 in Zambia, reported that they would recommend this testing approach to others. In Uganda, 2201 (97·9%) of 2249 caregivers reported their experience testing their child with an oral HIV self-test was easy or very easy, compared with 436 (93·0%) of 469 in Zambia, and 433 (92·3%) of 469 caregivers in Zambia reported HIVST kit instructions were easy or very easy to understand. In Zambia, 443 (94·5%) caregivers did not need assistance administering the HIV self-test to their child (not measured in Uganda).

To improve CG-HIVST, 1450 (53·6%) of 2704 respondents wanted more information, including community education and mobilisation to increase awareness of this testing option; 1082 (40·0%) repondents wanted increased access, including better access to HIVST for caregivers. Some caregivers recommended improved counselling (110 [4·9%] of 2238 in Uganda and 46 [9·9%] of 466 in Zambia) and improved HIVST kit instructions (80 [3·6%] in Uganda and 48 [10·3%] in Zambia). In the absence of CG-HIVST, caregivers would have chosen the following testing options: home-based testing by a health-care worker (390 [17·5%] of 2226 in Uganda and 126 [28·1%] of 448 in Zambia) or facility-based testing (1782 [79·2%] of 2251 in Uganda and 313 [66·7%] of 469 in Zambia), and 2·4% were uncertain how they would test their child.

The findings of this study suggest a high acceptability and feasibility of CG-HIVST among interested caregivers, as nearly all accepted home-based, CG-HIVST to screen their children using an oral kit and returned their results. This novel testing option was highly effective in linking children with reactive HIVST results to confirmatory testing and all newly diagnosed children living with HIV to treatment. Numerous studies have shown that HIV-related stigma and discrimination deter clients from seeking facility-based testing services,¹⁹ including concerns regarding client confidentiality with an HIV-positive diagnosis.^8,20 CG-HIVST addresses these barriers, affording parents greater autonomy, privacy, and convenience to screen their children at home. This study’s findings support previous studies of parents who reported acceptability of home-based oral HIVST among caregivers, with convenience and privacy as potential benefits.^14,21

A key concern when implementing any screening test is its performance and usability, which depend on its inherent accuracy and the ability of individuals to perform the test. Previous studies have shown the oral HIV self-test is highly sensitive in children compared with blood-based rapid tests performed in a clinical setting by providers.¹⁶ Although health-care workers and programmes have expressed concerns about the ability of caregivers to correctly administer oral HIV self-tests to children,¹⁴ one study conducted in Zimbabwe showed 96% or more of caregivers performed HIVST accurately after receiving a demonstration, including collecting the oral fluid, inserting the flat pad into the buffer solution, using a timer, and interpreting the result.¹⁷ In this study, most caregivers reported the HIVST kit was easy to use, with only 5·5% of caregivers in Zambia requesting administration assistance. Digital results reporting (ie, telephone photograph) could help to ensure result authenticity and credibility, where feasible.

CG-HIVST in children was highly feasible. Nearly all caregivers returned their child’s screening result irrespective of country, rural or urban setting, or facility type and nearly all within 3–7 days. Several studies have shown that the distribution of HIVST kits has increased the uptake of HIV testing.^22,23 Because children have a higher risk of AIDS-related morbidity and mortality than adults,²⁴ failure to follow up or report results could have more devastating consequences for children. In this CG-HIVST study, the proportion of results reported among children (97·5%) was higher than in other HIVST studies among men (83%)²⁵ or adolescents and youth (59%).²⁶ Establishing expectations for returning results at the time of test distribution (eg, within 3 days) and expected follow-up procedures (eg, three telephone calls followed by a home visit) might have influenced the high proportion of HIVST results returned, irrespective of demographic characteristics or geographical setting, which is encouraging for programmatic use.

According to the OraQuick instructions for use, the expected positive predictive value among a population with 0·4–0·7% HIV prevalence ranges from 29·2% to 42·0%.²⁷ The HIVST positive predictive value in this study was 32·7% in Uganda and 61·1% in Zambia, which is expected in this paediatric population with a low HIV prevalence. This finding underscores the importance of blood-based confirmatory testing for every child with a reactive HIV self-test in accordance with national guidelines.²⁸ Additionally, caregiver counselling is essential to ensure that caregivers understand that HIVST is a screening test. Thus, a reactive result does not necessarily mean their child is HIV-positive; it indicates the child needs confirmatory testing to establish their HIV status. Although this study provided remote counselling and support, additional qualitative research might provide insights into the experiences of parents whose children have a reactive or false-positive HIVST result to enhance counselling messages. Although health-care workers have expressed concerns regarding loss to follow-up for children requiring confirmation testing,¹⁴ nearly all children with a reactive HIV self-test reported for confirmatory testing with a transport reimbursement. Because children are a population with a low HIV prevalence, this approach could potentially reduce the volume of testing by health-care workers needed to identify one new child living with HIV, while giving autonomy and privacy to caregivers.

However, it is crucial to train health-care workers on the procedures for home-based CG-HIVST. Programmes could consider developing an operational manual to better equip health-care workers to train caregivers on appropriate HIVST kit usage. As programmes consider operational guidance, training of health-care workers to create buy-in for this novel testing approach will be essential. Training could focus on the benefits of home administration by caregivers, the importance of test kit demonstration to improve caregiver competence and confidence with HIV self-test use and result interpretation, establishing timebound expectations (eg, 3 days) for results return with a call back line, and noting the decreased workload at the facility level.

The proportion of false-positive tests returned by caregivers in Uganda decreased after research assistant refresher trainings. Although this might be due to the research team being better equipped to advise caregivers on how to use the test, given the decrease in confirmatory positivity from 1·0% to 0·5% after the second refresher training, training for health-care workers should emphasise that both a strong and faint second line is considered a positive, or reactive, HIVST result. Another study noted that 33% of reactive HIV self-tests were weakly positive.¹⁶ Programmes could consider showing video-based instructions on CG-HIVST administration in waiting areas, as one study in Kenya found this led to higher caregiver understanding than health-care worker counselling.²⁹

Nearly all children newly diagnosed with HIV in this study initiated same-day treatment. Non-biological children living in the same household of people living with HIV in Uganda had the same HIV positivity (0·7%) as biological children, which might suggest household clustering of HIV risk. Similarly, an index testing study among paediatric contacts in Zimbabwe found 43·6% of newly diagnosed children were single or double orphans.¹³ Index testing programmes could offer testing to all children (ie, biologically and non-biologically related to their caregivers) with an undocumented HIV status living in the same household as people living with HIV, because non-biological children in these households might be affected by other social determinants of health, such as orphanhood, which increase their risk of HIV. Men accounted for nearly 20% of enrolled index caregivers and 12% of newly diagnosed children, highlighting the importance of engaging male caregivers living with HIV. Because 88% of newly diagnosed children were contacts of female caregivers, it could be useful to offer index testing services to the male caregivers of newly diagnosed children to increase HIV testing and case identification among men.³⁰

Minor adverse events were reported but uncommon. Reported adverse events included gum pain, bleeding, and itching, which might be due to common administration mistakes, such as scrubbing the gums, observed in another study.¹⁷ However, similar to a study in Zimbabwe,¹⁷ there were no reports of major social harms, such as violence, which might be due to the intimate partner violence assessment carried out before HIV testing was offered and, therefore, those who were recruited might have had a lower risk of violence. Programmes could consider this operational approach to decrease any risk of social harm, as this study screened out 3·8–6·2% of caregivers with a history of violence in the home and referred them to standard-of-care testing and social services. Caregiver counselling remains important, and programmes can include talking points for caregivers to discuss HIVST with their partners and children to facilitate age-appropriate conversations and allay fears.

Study strengths included a large sample and a study population drawn from a wide range of public sector facility types in two different countries. We acknowledge a number of limitations. The study design limited our ability to measure effectiveness, and future research using a control group is recommended to evaluate CG-HIVST compared with the standard of care. This study recruited people living with HIV who were receiving services; however, research is needed on reaching more vulnerable children, whose parents might not present for care or might be deceased. The study did not document the uptake of alternative testing approaches (eg, facility-based testing or home testing by a health-care worker) and, therefore, did not capture the testing preferences of all adult index clients. The study had dedicated research staff, active staff follow-up, and reimbursements, all of which might have contributed to the high proportion of participants returning test results and receiving confirmatory testing. Further evaluation is needed to understand the sustainability and trade-offs in costs to the health system by decreasing facility workload, while supporting transport for confirmatory testing, which was necessary for 1·6% of children. We did not perform parallel blood-based HIV testing for children with non-reactive oral test results because a previous study on the accuracy of CG-HIVST result interpretation documented a negative predictive value of 100·0%¹⁷ and, therefore, we cannot ascertain whether there were any false-negative HIV self-tests. Because the caregiver survey was administered 1–4 months after HIVST and by a research assistant over the telephone, the results are subject to both recall and social desirability bias. Finally, we cannot exclude potential under-reporting of social harms.

Our findings suggest that CG-HIVST in children is acceptable, feasible, effective, and safe, with no reports of social harm. WHO recommends an optimal mix of HIV testing approaches. Incorporating CG-HIVST in national guidelines could expand equitable access to HIV testing for children, while reducing the burden on health facilities and expanding family-centred testing options.