Language outcomes of preschool children who are HIV-exposed uninfected: An analysis of a South African cohort

Abstract

Introduction: There are approximately 16 million children who are HIV-exposed and uninfected (CHEU) worldwide. Studies suggest that CHEU are at risk for developmental impairment in infancy, particularly in language domains. However, there is limited research examining neurocognitive function in CHEU older than 2 years, including important pre-school years. This study aimed to investigate associations between HIV exposure without infection and neurocognitive outcomes and to determine risk factors for neurodevelopment in CHEU at age 3-4 years.

Methods: The Drakenstein Child Health Study is a South African population-based birth cohort which enrolled women in pregnancy with ongoing follow up. Neurocognitive outcomes were assessed in children at 3.5 years by trained assessors blinded to HIV status including general cognitive function, language, and memory, measured using the Kaufmann Assessment Battery for Children, Second Edition (KABC-II). Data were compared between CHEU and children who were HIV-unexposed uninfected (CHUU) using multivariable logistic and linear regression, including testing for effect modification; sex-stratified risk factor analyses were performed.

Results: A total of 497 children were included (97 [20%] CHEU; 400 [80%] CHUU; 50% male), with a mean age of 3.5 years (range 3.4-3.6). Groups had similar birth and household characteristics, although mothers of CHEU were older, on average. Overall, CHEU had lower expressive language scores compared to CHUU on unadjusted and adjusted analyses (effect size: -0.23 [95% CI -0.45, -0.01]). There were no group differences in general cognitive or memory function (p>0.05). On sex-stratified analyses, male CHEU were found to have higher odds of suboptimal cognitive development compared to male CHUU (aOR 2.28 [95% CI 1.06, 4.87], p = 0.034). Several other factors including birthweight, maternal education, maternal ART duration and HIV viral load during pregnancy were associated with cognition, memory, or expressive language outcomes in CHEU, dependent on child sex.

Interpretation: The findings suggest that perinatal HIV exposure continues to be associated with impaired language development across the preschool years, highlighting the importance of targeting early interventions to optimise language outcomes. Further, the results suggest the importance of demographic, biological and HIV-related variables influencing developmental outcomes in CHEU. The greater risk of suboptimal cognitive development in male CHEU requires investigation around sex-specific mechanisms.

Fig 1. DCHS cohort enrolment and sample used in this analysis.

Fig 2. Neurocognitive outcomes at 3.5 years compared between CHEU and CHUU.

*p<0.05. † Adjusted for child sex, maternal education, maternal age and household income. *p<0.05. CHEU: Children HIV-exposed uninfected, CHUU: Children HIV-unexposed uninfected.

Fig 3. Associations between risk and protective factors with neurocognitive development of CHEU.

Red colour indicates p<0.05. * Effect sizes for binary outcomes generated using Cohen’s d. For continuous variables (birthweight and CD4 count), regression β values were generated. Factor information: Maternal education compares higher attainment (completed secondary/tertiary) versus primary/some secondary education; Household income compared higher income (>1000 rand/month) versus lower (<1000 rand /month); ART timing compared longer maternal ART (initiation pre-pregnancy versus during pregnancy); Viral load compared unsuppressed viral load (>1000 copies/ml) versus suppressed (<1000 copies/ml).