Refining absolute GFR estimation in children: development and validation of the GROW-GFR equation

Accurate estimation of glomerular filtration rate (GFR) is essential when dosing many medications. Existing paediatric GFR equations vary in accuracy, with only one (Modified Schwartz) validated in children <2 years. This study developed a new paediatric GFR equation (GROW-GFR) using a tobramycin population pharmacokinetic model and evaluated its performance against current equations and nuclear medicine (NM)-GFR scans.

Retrospective analysis of NM-GFR scans from July 2016-November 2023 in children aged 0-19 years at The Royal Children's Hospital, Melbourne. Results were included if NM-GFR was ≤150 mL/min and height, weight and serum creatinine were recorded. Accuracy of GFR estimates within 30% and 10% of NM-GFR results (p30, p10 accuracy) and bias for the GROW-GFR, Modified Schwartz, Chronic Kidney Disease in Children Under 25 years (CKiD-U25), Full Age Spectrum (FAS), and FAS-height (HT) equations were assessed.

Overall, 1560 NM-GFR results from 711 children (median age 8.5 years; range 0.1-19) were included (median NM-GFR 55 mL/min; range 0.5-150). GROW-GFR showed a p30 accuracy of 76% and bias of -1.9 mL/min, with consistent performance across all ages. The Modified Schwartz equation had a p30 accuracy of 46% and bias of 17.3 mL/min. In children ≥2 years, GROW-GFR outperformed other equations (p30: 76% vs. 49%-58%; bias -1.9 vs. 13.7-18.4 mL/min). After applying linear mixed-effects models, GROW-GFR p30 accuracy and bias were 80% and -2.5 mL/min, outperforming other equations (p < 0.001).

The GROW-GFR equation provided more accurate GFR estimates across all paediatric ages, including children <2 years, compared to existing equations. Therefore, should be considered the preferred GFR estimation method.