To compare Frascati and Global Deficit Score (GDS) criteria for classifying cognitive impairment in children with and without HIV.

Frascati and Global Deficit Score (GDS) systems commonly classify HIV-Associated neurocognitive disorders (HAND) in adults; however, their utility in pediatric patients is uncertain. Establishing whether these methods show concordance, minimize false positives, and reflect real-world function is essential.

Baseline data from 614 participants in the pediatric HANDZ study (HIV+, HIV-exposed uninfected [HEU], and healthy unexposed uninfected [HUU]) were analyzed. Cognitive impairment was classified using Frascati and GDS criteria (impaired vs unimpaired). Agreement between methods was tested using Cohen’s kappa, and false positive rates were evaluated in HUU children. Real-world validity was examined using logistic regression predicting global Brief Impairment Scale (BIS) impairment and correlations with BIS school-functioning subscales.

Of 614 participants, 125 (20.4%) were impaired by Frascati and 76 (12.4%) by GDS criteria. Agreement was substantial (κ=0.71, 92%) but differed systematically (χ²=49.0, p<0.001), with Frascati classifying more as impaired. Among HUUs, Frascati identified 4 of 69 (5.8%) as false positives versus 1 (1.5%) by GDS (p=0.08). This difference was not significant (p=0.08). Both predicted roughly threefold higher odds of global BIS impairment (Frascati OR=2.94, 95% CI=1.72–5.01; GDS OR=3.40, 95% CI=1.84–6.28) and correlated with worse BIS school functioning (ρ=0.33 and ρ=0.28, both p<0.001). Impaired children scored an average of 7 points higher on BIS school-functioning subscale (higher is worse) compared to unimpaired peers (p<0.001), with similar results when adjusting for age and sex. 

Frascati and GDS criteria demonstrate substantial agreement in classifying pediatric cognitive impairment, though Frascati identifies more children as impaired including some healthy controls. Both predict functional impairment and academic difficulties, supporting their validity. However, Frascati’s higher false-positive rate may indicate a limitation. Larger control samples and longitudinal studies are needed to clarify utility.