Résumé
Background The Kidney Failure Risk Equation (KFRE) is a prognostic score for predicting kidney replacement therapy (KRT) at 5 years in patients with chronic kidney disease (CKD). Some studies show that the score performs poorly for certain etiologies of CKD but not all have been evaluated. The aim of this study was to evaluate the performance of the KFRE score according to the etiology of the CKD.Methods The CKD-CAREMEAU cohort, which included all patients who consulted a nephrologist for CKD between 2008 and 2017, was used. Patients were monitored for 5 years and the observed event was KRT, completed by the French REIN registry (Reseau Epidemiologique et Information en N & eacute;phrologie). Performance was evaluated by calibration (individual approach) and discrimination (populational approach), using observed vs predicted risk curves and the area under curve, respectively, according to each etiology.Results A total of 3191 patients were included in the study, median age 71 (interquartile range 61-80) years, 1921 (60%) of whom were men, and the median estimated glomerular filtration rate was 41 (28-80) mL/min/1.73 m(2). The main etiologies of CKD were vascular; 1164 patients (37%), diabetic; 667 (21%), glomerular; 512 (16%), tubulointerstitial; 459 (14%), polycystic; 121 (4%) and unclassified; and 268(8%). Discrimination was satisfactory for all etiologies, but calibration was unsatisfactory for polycystic, tubulointerstitial and unclassified (multiple or unknown) etiologies, without correlation with age.Conclusions Predictive performance of the KFRE score at 5 years varies according to the etiology of CKD, without impact on discrimination, but with a significant impact on calibration, and poor performance for polycystic, tubulointerstitial and multiple/unknown etiologies. These limitations should be known in order to develop new prognostic tools.