Frailty in Adults with CKD and Validation of the Kidney Failure Risk Equation in Frailty Subgroups

Key Points The four-variable kidney failure risk equation (KFRE) does not accurately estimate kidney failure risk in frail individuals. Prognostication of kidney failure is improved in frail individuals with CKD by replacing eGFR creatinine with cystatin C-based eGFR in the KFRE model. Consideration of competing mortality risk is particularly important in those with frailty and higher KFRE risk predictions. Background Frailty is common among adults with CKD, and its presence can influence clinical outcomes, such as advancing CKD and mortality. Clinical guidelines recommend the use of the kidney failure risk equation (KFRE) to guide management of CKD. Our aim was to validate KFRE by frailty status and assess whether model performance could be improved by using cystatin C-based eGFR equations and to assess the effect of accounting for competing mortality risk. Methods We studied adults from the prospective research cohort UK Biobank with CKD G3–5 (eGFR <60 ml/min per 1.73 m 2 ) by any of the three CKD Epidemiology Collaboration consortium eGFR equations: eGFR creatinine (eGFRcr), eGFR cystatin C (eGFRcys), and eGFR cystatin C-cystatin. Frailty was assessed by a modified frailty phenotype and two cumulative deficit frailty indices. Kidney failure was defined as long-term dialysis or kidney transplantation. Model assessment included discrimination, calibration, and overall fit at 2 and 5 years. Results The prevalence of frailty by one or more measures was 35% ( n =8533). Those classed as frail had a higher discrepancy between eGFRcys and eGFRcr compared with the nonfrail group (−15.8 versus −6.9 ml/min per 1.73 m 2 ). Discrimination of KFRE was good (area under the receiver operating characteristic curve ≥0.88 across all frailty subgroups and eGFR equations). Kidney failure at 5 years was underestimated in individuals with frailty (observed/expected ratio, 1.70; 95% confidence interval, 1.55 to 1.85). Calibration-in-the-large improved when eGFRcr was replaced by eGFRcys (5-year observed/expected ratio, 1.20; 95% confidence interval, 1.05 to 1.35). Overestimation of kidney failure risk in analyses that do not account for competing mortality risk, compared with those that do, was most apparent in the frailty group and the higher KFRE predicted risk groups. Conclusions KFRE underestimates kidney failure risk of individuals with CKD and frailty. Risk prediction improved for those with frailty when cystatin C-based eGFR equations are used and when analyses account for competing risk of mortality. These factors should be considered when KFRE calculation is used in clinical practice in individuals with frailty.