Background: Formulas do not estimate renal function with acceptable precision and accuracy.
Methods: We compared 51 creatinine-based and/or cystatin c-based formulas with a gold standard (iohexol plasma clearance) in 193 renal transplant recipients using concordance correlation coefficient, total deviation index, coverage probability and the error in chronic kidney disease (CKD) stage classification.
Results: No formula showed a concordance correlation coefficient greater than 0.90 (average for creatinine-based formulas: ∼0.70 and for cystatin c-based formulas: ∼0.85). A wide total deviation index was observed: approximately 70% (creatinine-based) and approximately 50% (cystatin c-based), indicating that 90% of the estimations showed bounds of error of ±70% or ±50%, respectively, compared with the gold standard. No formula included 90% of the estimations within a coverage probability of ±10%. Half the CKD stages classified by creatinine-based formulas were incorrect, mainly due to overestimation of renal function. One of 3 CKD stages diagnosed by cystatin c-based formulas was incorrect, with both overestimation and underestimation. Overall, the formulas showed very low precision and accuracy and a high degree of error in reflecting real renal function.
Conclusions: In conclusion, formulas do not properly reflect renal function in kidney transplantation, which makes their use in clinical practice unreliable. Moreover, their use in clinical trials should be avoided.