Estimated GFR Slope in Kidney Transplant Patients: When the Error Is Random

Background. The evaluation of renal function changes over time is crucial in day-to-day renal transplant care, and the slope of renal function is a major outcome in clinical trials. Little is known about the reliability of estimated glomerular filtration rate (eGFR) in reflecting real glomerular filtration rate (GFR) changes. Methods. We analyzed the variability of eGFR slope by 63 equations in estimating measured GFR (mGFR) changes in 110 renal transplant patients. The agreement between eGFR and mGFR slopes was evaluated by the concordance correlation coefficient and the limits of agreement. Patients were grouped based on mGFR slope in rapid GFR loss: faster than -3 mL/min/y; stable renal function: -3 to +3 mL/min/y; and improvement in GFR: higher than +3 mL/min/y. Results. Concordance correlation coefficient averaged 0.36 and limits of agreement +/- 10 mL/min/y, indicating very poor agreement between eGFR and mGFR slopes. The eGFR slope classified patients into the same group of mGFR slope only in 25% of the cases. In about two-thirds of patients, the eGFR slope was either markedly faster or slower than the mGFR slope. In half of these cases, the discrepancy between mGFR and eGFR slopes was >= 50%. Conclusions. Formulas are neither accurate nor precise in reflecting real GFR decline in renal transplant patients, making them unreliable for clinical practice and trials.