Evaluation of eGFR methods in a Sub-Saharan African community-based pediatric population

Abstract

Backgrounds Accurate eGFR assessment is crucial for early CKD detection, management and optimal medication dosing, but its validation, especially for Cystatin C-based equations, are limited in African children and adolescents with CKD. This study evaluates eGFR equations utilising Cystatin C and creatinine in this population. Methods: This prospective community-based cohort study assessed CKD in children aged 1 to 18 using Cystatin C and serum creatinine. We evaluated eGFR performance with the Bland-Altman plot and ROC curve. Lin's concordance assessed agreement between various equations and the reference standard (Schwartz CKiD equations). Pairwise ROC curve comparisons determined statistical differences in estimation equation performance. Results: The mean age of 688 children and adolescents evaluated for CKD was 7.8 (± 3.9) years, with 341 (49.6%) being male. The Pottel equation had the lowest mean bias (-3.41, 95% CI: -2.57 to -4.25), highest discriminatory power (0.989), highest accuracy across all age groups, and highest CKD detection. Bouvet's formula had the highest mean bias [-48.66 (95% CI: -50.75 to -46.57)] across all age groups, but an average ROC discriminatory power of 0.889. Zappitelli had a lower ROC discriminatory value (0.547) but a modest bias (-20.22; 95% CI -22.21 to -18.23). Lin's concordance with Schwartz CKiD was highest with Pottel (0.99) and lowest with Bouvet (0.08). Zappetti and Chehade disagreed most, but Pottel and Chehade agreed closely on equation pairwise comparisons. Conclusion: In African children with CKD, Pottel's height-independent equation outperformed Chehade, Zappitelli, and Bouvet in terms of ROC discriminatory value, accuracy, and Schwartz CKiD concordance.