Longitudinal Cognitive Outcomes in Children With HIV in Zambia: 2-Year Outcomes From the HIV-Associated Neurocognitive Disorders in Zambia (HANDZ) Study

Abstract

Objective: To describe longitudinal outcomes and predictors of cognitive outcomes in children with HIV in Zambia.

Background: Multiple studies have shown that children with HIV are at risk for impaired cognition. However, there are limited data on longitudinal cognitive outcomes in children with HIV.

Methods: We conducted a prospective cohort study of 208 perinatally infected children with HIV ages 8-17 years, all treated with antiretroviral therapy, and 208 HIV-exposed uninfected controls. Participants were followed for 2 years. Cognition was assessed with a custom NIH Toolbox Cognition Battery, and tests were combined to generate a Summary Cognition Score (SCS). The contribution of potential risk factors to outcomes was explored using regression models and group-based trajectory modeling.

Results: HIV was strongly associated with lower SCS at baseline [β-14, 95% confidence interval (CI): -20 to -7, P < 0.001]. Change scores over time were similar between groups, but poorer average performance in children with HIV persisted at the 2-year follow-up visit (adjusted β = -11, 95% CI: -22 to -0.3, P = 0.04). Other than HIV, the strongest predictors of baseline SCS included socioeconomic status index (β =3, 95% CI: 1, 5, P = 0.004), history of growth stunting (β=-14, 95% CI: -23 to -6, P = 0.001), history of CD4 count below 200 (β = -19, 95% CI: -35 to -2, P = 0.02), and history of World Health Organization stage 4 disease (β = -10, 95% CI: -19 to -0.2, P = 0.04). In the group-based trajectory model, HIV+ status predicted membership in the lowest performing trajectory group (odds ratio 2.5, 95% CI: 1.2 to 5.1, P = 0.01).

Conclusions: Children with HIV are at risk of poor cognitive outcomes, despite chronic treatment with antiretroviral therapy.

Figure 1:

Directed Acyclic Graph (DAG) demonstrating a causal model linking HIV to Cognitive Function. According to the DAG, the minimum sufficient adjustment set to isolate the causal effect of HIV on cognitive outcomes is to control for socioeconomic status. Potential mediators in the relationship between HIV status and cognitive outcomes include HIV-related disease severity, effects on education, malnutrition, antiretroviral neurotoxicity.

Figure 2:

Participants with HIV scored lower on all cognitive tests at baseline, and these differences persisted over time in two out of three cognitive tests (DCCST and PCPS). Legend: PCPS, Picture Comparison Processing Speed; DCCST, Directional Change Card Sort Test; HEU, HIV-exposed uninfected.

Figure 3:

Group-based trajectory modeling demonstrated three trajectory groups, which were primarily differentiated by Summary Cognition Score at baseline. All three groups showed improvement over time. HIV status predicted membership in the lowest performing trajectory group, even when controlling for socioeconomic status, poor health, and history of severe malnutrition (OR 2.5, 95% CI 1.3, 5.1, p=0.009.)