We conducted a prospective cohort study of 208 perinatally HIV-infected children with HIV ages 8-17 taking antiretroviral therapy (ART) and 208 HIV-exposed uninfected controls.  Cognition was assessed with a comprehensive neuropsychological battery, and impairment was defined using a global deficit score approach.  Socioeconomic status was assessed by standardized interviews and included domains for food security, electricity, education, and water source.  Neighborhood effects on cognition were investigated using quantitative geographic information systems, while clustering of cognition and SES variables was evaluated using SaTScan, a spatial statistics software. Risk factors for cognitive impairment were evaluated in univariable and multivariable regression models.

Subjects with HIV were more likely to have cognitive impairment compared to controls (34% vs. 19%, p=0.001).  Risk factors for cognitive impairment among subjects with HIV included self-reported poor health (OR 2.9, p=0.02) and socioeconomic status index (OR 0.80, p<0.001).  Clustering of cognitively impaired subjects with HIV was detected in Chawama, a neighborhood that also shows clustering of malnutrition and low access to running water. Residence in Chawama was significantly associated with cognitive impairment in subjects with HIV. (OR 2.9, p=0.005). Residence in Chawama remained significant in a multivariable regression model controlling for socioeconomic status and poor health (adjusted OR 2.4, p=0.03).

Neighborhood effects may contribute substantially to cognition but are rarely studied in research evaluating cognitive function.  The causal pathways between neighborhood of residence and cognitive impairment are unclear and should be evaluated in future studies.