Combining GFR estimates from cystatin C and creatinine—what is the optimal mix?

Background: Combining estimated glomerular filtration rate (eGFR) equations based on creatinine and cystatin C has been shown to improve the accuracy of GFR estimation. This study aims to optimize this strategy for height-independent GFR estimation in children. Methods: Retrospective study of 408 inulin clearance tests with simultaneous International Federation of Clinical Chemistry-calibrated measurements of creatinine, cystatin C, and urea in children (mean age 12.5 years, GFR 91.2 ml/min/1.73m2) comparing the arithmetic (meanarith) and geometric means (meangeom) of a height-independent creatinine-based (full age spectrum, based on age (FASage)) and a cystatin C-based equation (FAScys), with the complex height-dependent CKiD3 equation incorporating gender, height, cystatin C, creatinine, and urea. Results: Meangeom had a P30 accuracy of 89.2% compared to meanarith 87.7% (p = 0.030) as well as lower bias and %precision error and performed almost as well as CKiD3 (P30 accuracy 90.9%). Modifying the weight of FASage and FAScys when calculating the means showed that an equal contribution was most accurate in most patients. In spina bifida patients, FAScys alone outperformed any combination. Malignancy or nephritis patients had slightly higher accuracy with weighted means favoring cystatin C or creatinine, respectively. Disagreement between FAScys and FASage was inversely correlated with the accuracy of meangeom. When disagreement exceeded 40%, application of weighted means based on diagnosis improved the performance of eGFR. Conclusions: In the absence of height data, the optimal strategy for estimating GFR in children is by using the geometric mean of FASage and FAScys. When there is large disagreement between the two, weighted means based on diagnosis improve accuracy.