The Cockcroft-Gault formula should not be used in children

The Cockcroft-Gault formula should not be used in children.BackgroundAlthough designed for adults, the Cockcroft-Gault formula was recently proposed for use in children ≥13 years of age.MethodsWe compared the feasibility of the Cockcroft-Gault formula against the standard pediatric Schwartz formula and a novel cystatin C–based formula. Our patient cohort included 262 children aged 1 to 18 years with various renal pathologies, who underwent a 99-technetium diethylenetriaminepentaacetate (99Tc DTPA) glomerular filtration rate (GFR) renal scan. Calculations were performed in Système International (SI) units using published constants and recalculated constants from our patient population. Agreement was assessed using Bland and Altman analysis.ResultsPublished and recalculated constants for the Cockcroft-Gault formula were 1.23 and 0.96, respectively, for boys, and 1.05 and 0.90, respectively, for girls. The published and recalculated constants for the Schwartz formula were 48 and 49.9, respectively, for boys ≥13 years old, and 38 and 46.2, respectively, for all girls and for boys <13 years old. Using published constants, there was agreement between GFR and Cockcroft-Gault formula in boys ≥13 years old (average bias 5.0 ± 23.5%) while there was an average error of -19.0%± 36.4% for all ages. Similarly, the average bias with Schwartz for boys ≥13 years old was -6.8 ± 24.0% and for all patients was -12.8 ± 24.2%. Using recalculated constants, the average bias with Cockcroft-Gault in boys ≥13 years old was -19.8 ± 23.5% and for all patients was -38.5 ± 35.2%. Similarly, the average bias with Schwartz for boys ≤13 years old was -1.1 ± 24.3% and for all patients was 3.0 ± 24.0%. The novel cystatin C–based GFR calculations showed an average error of -4.9 ± 20.3% in the adolescent boys and 2.4 ± 20.4% for all ages.ConclusionCockcroft-Gault formula showed the worst agreement with GFR, regardless of using published or recalculated constants. The cystatin C–based approach resulted in the least error, and should be used for estimation of GFR.