Prevalence of undiagnosed HIV among children in South Africa, Côte d'Ivoire and Zimbabwe: a model-based analysis to inform paediatric HIV screening programmes

Abstract

Introduction: To improve the diagnosis and survival of children living with HIV (CLWH), the World Health Organization recommends testing approaches beyond traditional infant HIV testing programmes. Information about undiagnosed HIV prevalence among children of varying ages in the general population is needed to guide innovative national/subnational case-finding and testing approaches.

Methods: We used the Cost-Effectiveness of Preventing AIDS Complications (CEPAC)-Pediatric model to estimate the prevalence of undiagnosed HIV in 2-, 5- and 10-year-old children in South Africa, Côte d'Ivoire and Zimbabwe in 2018. We simulated cohorts of children born in 2008 (10-year-olds), 2013 (5-year-olds) and 2016 (2-year-olds). Country-/year-specific inputs for pregnant/breastfeeding women included: HIV prevalence (4.2-32.3%), HIV incidence (0.03-0.24%/month), knowledge of HIV status (27-89%) and antiretroviral drug coverage (36-95%). Paediatric inputs included early infant testing coverage (6-95%) and breastfeeding duration (0-20 months). We projected the proportion of surviving CLWH in whom HIV remained undiagnosed and the undiagnosed HIV prevalence among surviving children of each age in the general population. For children born in 2016, we projected survival and diagnosis of all CLWH through 2026. We conducted sensitivity analyses on model parameters.

Results: In 2018, the projected proportion of surviving CLWH whose HIV remained undiagnosed in South Africa/Côte d'Ivoire/Zimbabwe was 44.2%/55.8%/52.9% among 2-year-old CLWH; 29.0%/37.8%/33.2% among 5-year-old CLWH; and 18.3%/25.4%/23.1% among 10-year-old CLWH. Projected general population undiagnosed HIV prevalence in South Africa/Côte d'Ivoire/Zimbabwe was 0.44%/0.32%/0.68% among 2-year-olds; 0.25%/0.17%/0.41% among 5-year-olds; and 0.24%/0.14%/0.38% among 10-year-olds. Among all CLWH born in 2016, 50-54% were projected to die without HIV diagnosis (and subsequently without treatment) within 10 years after birth; 80-85% of these deaths occurred in the first 2 years.

Conclusions: Projected population-level undiagnosed HIV prevalence is low and sharply decreases after age 2, with more CLWH dying than being diagnosed. Despite low undiagnosed prevalence in the general population of older children, we project that a large proportion of CLWH remain undiagnosed, suggesting that innovative strategies targeting untested children of all ages outside of health facility settings should be prioritized. Programmes could consider routine testing of the general population of children below 2 in all settings and children of all ages in high-prevalence settings.

Keywords: Africa < region; HIV care continuum; modelling; paediatrics; testing; vertical transmission.

Figure 1

Projected vertical transmission by country, year and ARV status. Total projected vertical transmission by country, year and ARV status. On the y‐axis, South Africa, Côte d'Ivoire and Zimbabwe are shown, stratified by the 2008, 2013 and 2016 birth cohorts. Each bar represents the total vertical transmission (defined as the total number of children who acquire HIV/total exposed to HIV) per year. The dark blue represents the vertical transmission attributable to pregnant/breastfeeding women off ARVs, who either did not know their HIV status or did know their status, but did not initiate ARVs. The light blue represents the vertical transmission attributable to pregnant/breastfeeding women on ARVs. For 2008, this includes those who received only short‐course zidovudine or single‐dose nevirapine during pregnancy. Abbreviation: ARV, antiretroviral drug (includes three‐drug combination ART, single‐dose nevirapine or short‐course zidovudine).

Figure 2

Projected 10‐year survival and diagnosis among children living with HIV born in 2016 in South Africa (Panel a), Côte d'Ivoire (Panel b) and Zimbabwe (Panel c). Projected survival and diagnosis among children living with HIV born in 2016 in (a) South Africa, (b) Côte d'Ivoire and (c) Zimbabwe, with proportion of children who ever had HIV (currently alive or dead) on the vertical axis and time since birth, in months, on the horizontal axis. For each time point, the proportion of children with HIV who remain alive and undiagnosed is shown in light green, the proportion who remain alive and diagnosed is shown in blue, the proportion who have died (of any cause) while undiagnosed is shown in yellow and the proportion who have died (of any cause) after being diagnosed is shown in dark green. The proportion of children in each category at 24 months is displayed on the chart in boxes. The spike in the number alive and diagnosed seen in Côte d'Ivoire and Zimbabwe, but not South Africa, between months 16 and 20 represents the end of breastfeeding (breastfeeding duration was shorter in South Africa). To avoid underestimating undiagnosed prevalence, model inputs from 2016 are held constant for the 2016–2026 projection. Before the first infant HIV test is conducted (usually birth and/or 6 or 10 weeks of age), all children in the CEPAC‐P model are considered to have undiagnosed HIV, even if they are receiving multi‐drug ARV prophylaxis (sometimes known as presumptive treatment). Abbreviation: HIV, human immunodeficiency virus.

Figure 3

One‐way sensitivity analyses: effect of model input parameters on projected undiagnosed HIV prevalence of 2‐year‐old children in the general population in 2018 in South Africa (Panel a), Côte d'Ivoire (Panel b) and Zimbabwe (Panel c). The effect of model input parameters on undiagnosed HIV prevalence of 2‐year‐old children in 2018 is shown for (a) South Africa, (b) Côte d'Ivoire and (c) Zimbabwe, with input parameters on the vertical axis and undiagnosed HIV prevalence on the horizontal axis. For each parameter, the range of variation and base‐case values are shown in parenthesis. Abbreviations: BC, base‐case value; m, months.