A competing risk assessment of KFRE and Grams: Which is best?

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background: Competing risks, such as mortality, can alter chronic kidney disease (CKD) progression estimates, potentially leading to kidney replacement therapy (KRT) overestimation. While the Kidney Failure Risk Equation (KFRE) and Grams models predict CKD progression, their accuracy in Advanced CKD (ACKD) under competing risks remains unclear. This study validates and compares both models in a Spanish ACKD cohort, assessing their predictive performance while accounting for competing risks. Methods: In this retrospective observational study, data from 271 patients aged 65–85 years with stage 4 CKD attending ACKD consultations. Patients were followed for 5 years or until KRT initiation or death. KFRE at 2- and 5-years and Grams at 2- and 4-years scores were calculated. Model discrimination was assessed using ROC analysis. Cox proportional hazards and logistic regression models were used for time-to-event and binary outcomes, respectively. Competing risks were evaluated with the Fine-Gray model. Results: At 2, 4, and 5 years, KRT initiation was 20%, 28%, and 31%, and mortality reached 25%, 33%, and 37%, respectively. KFRE demonstrated strong KRT prediction (AUC; 2-year: 0.829; 5-year: 0.808), while Grams showed excellent discrimination for KRT (AUC; 2-year: 0.843; 4-year: 0.818), moderate for cardiovascular events (AUC; 2-year: 0.719; 4-year: 0.736), and limited for mortality prediction (AUC; 2-year: 0.634; 4-year: 0.649). In Cox multivariate analysis, KFRE ≥20.77 significantly predicted KRT initiation (HR:2.71, p=0.002), whereas Grams ≥17.85 was not significant (HR:1.05, p=0.9). Fine-Gray analysis confirmed KFRE’s superior predictive ability for 2-year KRT (HR:5.3, p<0.001) compared to Grams (HR:2.67, p=0.02). Conclusions: KFRE outperforms Grams in predicting KRT initiation in ACKD, maintaining superiority after accounting for competing risks. While Grams moderately predicts cardiovascular events, its utility for mortality prediction is limited. A 2-year KFRE ≥20.77 should prompt early dialysis or transplant preparation in this population. Our findings highlight the need for population-specific thresholds to optimize ACKD care.

Article activity feed