In Kenya, despite national gains in reducing new HIV infections among children (0–14 years) from 18,000 in 2010 to 7600 in 2018, and drastically reducing HIV-related deaths from 16,000 in 2010 to 5200 in 2018,¹ a need to identify children living with HIV remains. A national survey in 2012 found that among children 18 months to 14 years old with a parent living with HIV, only 52.9% had been tested for HIV.²

Children with undiagnosed HIV often present with severe opportunistic infections and inadequate growth and have a higher mortality risk.^2–4 A delay in antiretroviral therapy (ART) initiation also increases the risk of morbidity and mortality among children.^5–8

Although Prevention of Mother-to-Child Transmission (PMTCT) programs have been an effective setting to identify HIV-exposed and – infected infants, the success in finding undiagnosed children outside of PMTCT has been underwhelming.³ Multiple approaches have been used to improve HIV detection among children, including Provider-Initiated Testing and Counselling (PITC), community and school-based approaches, and home-based testing.³ A systematic review of HIV testing among children and adolescents in sub-Saharan Africa found PITC approaches with the highest yield (percent of children tested identified as HIV-positive) include inpatient (12%), outpatient (7%), and family testing (3%), whereas home-based, outreach, and school-based programs have lower yield at 2.3, 0.06, and 0.1%, respectively.⁹ Additionally, implementation of routine testing of pediatric inpatients has resulted in increased identification of children living with HIV and higher rates of subsequent ART initiation for those identified.^10,11

Strengthening case finding is critical to the achievement of the 90–90–90 targets.¹² Family testing, a focused PITC approach that targets a high-risk population by testing the children and partners of index patients in HIV care, may be an important approach for efficient and early case detection.^13–15 A study in Malawi found that 81% of the children of adults living with HIV had not been tested; however, 33% of these children ended up being positive once tested, illustrating the potential for case finding through family testing.¹³ Recent efforts to utilize a family testing strategy have also shown positive results for identification and linkage to care.^15,16 One study in Kenya identified 2.5 family members at risk per index patient, two-thirds of which were children, with more than 80% of children identified living with HIV successfully enrolled into care.¹⁵ Despite improvements in child HIV testing uptake in Kenya, HIV identification remains a challenge. HIV testing yield among children declined nationally from 2.7% in 2015 to 0.5% in 2017, illustrating the need for more efficient testing strategies (routine program data from the U.S. President’s Emergency Plan for AIDS Relief [PEPFAR] HIV reporting database; Data for Accountability, Transparency and Impact Monitoring [DATIM]). In this study, we examined the effectiveness of a routinely implemented family testing approach at identifying children living with HIV (0–14 years of age) in comparison to other HIV testing approaches. Onward linkage to care and ART initiation after family testing were also assessed.

A total of 1937 adult index clients were identified as newly enrolled in care at the 60 study sites resulting in identification of 3005 children as contacts. Among child contacts, 2848 (94.8%) were tested for HIV through the family testing approach and 127 (4.5%) were HIV-positive. Applying the positivity rate of 4.5% for the entire cohort at risk (including the 157 children not tested), this approach identified 94.1% of all children living with HIV in the cohort (127 children identified out of 135 children expected to be HIV-positive). Of the children with diagnosed HIV infection, 100 (78.7%) were successfully linked to HIV care. Ninety-three children were eligible to initiate ART according to the national guidelines, out of whom 85 (91.4%) were initiated on ART during the study period.

At the same 60 study sites, 27,402 children were tested through PITC in outpatient departments identifying 338 children living with HIV for a yield of 1.2%. In inpatient departments, 1459 children were tested through PITC, identifying 23 children living with HIV for a yield of 1.6% (Table 1). Family testing had higher yield (p < 0.001) compared to testing in both outpatient and inpatient departments and was consistent across all health facility levels (Table 1).

The family testing approach using index clients newly enrolled in care resulted in high HIV identification yield among children (4.5%), comparable to other studies evaluating the efficacy of utilizing index patients to increase HIV case finding in children.^19,20 In the study by Ahmed et al.¹⁹ assessing a similar model for identification and linkage to care in Malawi, 64.7% of HIV-positive index patients who agreed to testing of household members had an untested child or young person at home, of which 4.0% were found to be living with HIV. In the study by Yumo et al.,²⁰ when parents living with HIV in Cameroon were invited to have their children tested for HIV, they found a 3.5% HIV identification yield. A study in Kenya by Wagner et al.²¹ demonstrated 42% of patients living with HIV with children had an untested child, with an HIV prevalence of 7.4% among those children once tested. The higher HIV identification yield found by Wagner et al. could be explained by several factors: their data were collected two years earlier than ours, and there is a general temporal decrease in testing yield as the population of children with unknown HIV status decreases; their study population was drawn for people accessing services at the national referral hospital in Nairobi, compared to our study sites in urban, periurban, and rural areas, and their study population was younger (≤12 years, compared to our study population of ≤14 years).

This study found that the HIV case identification rate for children using the family testing strategy with newly enrolled index clients was almost four-fold greater than outpatient testing and almost three-fold greater than inpatient testing strategies among children. Yumo et al.²⁰ also found a higher testing yield among children tested through index parents compared to outpatient testing (3.5% and 1.6%, respectively), although this was only a two-fold difference compared to the almost four-fold difference in our study. These comparisons at the same health facilities during the same periods are unique and provide a more valid comparison between strategies, accounting for temporal and geographic differences in HIV trends. Although several studies found high HIV case identification rates among outpatient clinics and inpatient wards,^9,22 more recent studies corroborate our finding of decreased yield in pediatric outpatient clinics and inpatient wards. A study in Malawi found an HIV yield of only 1.0%, among children newly tested within inpatient wards,²³ and another study across six African countries identified a reduction in HIV yield among children in just a year’s time, from 2.2% in 2015 to 0.9% in 2016.²⁴ National data show a decline in HIV positivity among children tested in Kenya from 2.7% in 2015 to 0.5% in 2017 (routine program data from DATIM).

It is important to note that although the HIV case identification rate was significantly higher with the family testing strategy, the absolute number of children living with HIV identified was higher with universal outpatient testing. Family testing, outpatient testing, and inpatient testing contribution to the total case load of children living with HIV identified was 26%, 69%, and 5%, respectively. This indicates that despite a lower yield, outpatient testing is still an important strategy for identifying children living with HIV. Systems that better identify HIV exposure risk among children in outpatient settings may allow for more efficient targeted testing.

Despite many known caregiver barriers to HIV testing of children, related to fear and stigma, denial, and structural barriers, such as transportation to the clinic,^25–29 we found a very high uptake of testing children (94.8%) using the family approach. In comparison, studies in Malawi and Cameroon only achieved uptakes of 62.6% and 56.7%, respectively.^19,20 The high testing uptake we achieved could be due to several reasons: targeted and longitudinal follow-up with index parents and eligible contacts, where a healthcare worker was prompted by the FIT in the index parent’s chart, as a reminder to ask about any remaining untested children at every visit; home-based testing options, where an index parent could opt to having a healthcare worker visit their home to conduct the HIV testing for their children instead incurring the extra costs and inconvenience of bringing their children to the health facility for testing; and the patient–provider relationship established with the index client as they receive their HIV care and ART over time, allowing for the establishment of trust and for multiple opportunities to discuss the importance of child testing. No incentives were provided to parents to have their children tested.

The family testing approach resulted in identification of 94.1% of the children expected to be positive in the cohort, surpassing the first “90” in the 90–90–90 targets. However, linkage to care was only 78.7%. ART initiation was 91.4% of the linked patients who were eligible for ART. Based on other studies, the family approach may boost linkage over routine approaches; a program in Lesotho using family testing found 90% linkage among children identified with HIV whereas a retrospective review of baseline routine testing and linkage among children in Uganda reported 58% linkage.^30,31 The ART initiation findings in our study are corroborated by previous studies using the family approach, in which ART was initiated in over 85% of children living with HIV.^15,16,31

This study had several limitations to consider. This study was conducted in an HIV high-burden region of Kenya that may have different societal stressors and cultural norms related to HIV testing and treatment compared to other countries. Therefore, our results may not be generalizable to other populations. There may also be a degree of participation bias, as the clinics participating in the study were selected as a convenience sample. Data on linkage and ART initiation were collected only for the family testing strategy, so a direct comparison with linkage and ART initiation with outpatient and inpatient HIV testing strategies could not be made.

The family HIV testing approach resulted in a high HIV identification yield for children and demonstrates promise in achieving the first two “90s” of the 90–90–90 targets for children, with additional efforts required to improve linkage. Implementing family testing in populations with similar epidemics may help to increase diagnosis of children living with HIV. Despite the decreasing and relatively low yield for inpatient and outpatient testing approaches, outpatient testing still provided the highest contribution to children living with HIV identified. More efficient models of finding undiagnosed children in outpatient and inpatient settings should be explored.