KDpredict to Evaluate Risk for Kidney Failure, Death

Treatment decisions regarding chronic kidney disease (CKD) are supported by tools that solely predict kidney failure risk; however, CKD disproportionately affects older people, who are more likely to die than develop kidney failure. To address this, Pietro Ravani and others developed and evaluated a super-learner strategy (KDpredict) to predict 1- to 5-year risk of kidney failure and death in adults with incident moderate to severe CKD (newly recorded estimated glomerular filtration rate [eGFR] 15-44 mL/min/1.73 m²). They presented results in an abstract during the International Society of Nephrology World Congress of Nephrology 2024.

Super-learner is a meta-algorithm that uses cross-validation to choose the best-performing learners among many prespecified regression models and machine learning algorithms, based on their ability to minimize prediction error. Predictors included age, sex, eGFR, albuminuria, and with or without diabetes and cardiovascular disease. The team used the index of prediction accuracy (IPA) to compare KDpredict with kidney failure risk equation (KFRE).

The study included patients from Canada (67,942), Denmark (17,528), and Scotland (7740) with G3bG4-CKD (median age, 77-80 years; median eGFR, 39 mL/min/1.73 m²). Median follow-up times were 5 to 6 years in all cohorts. Rates of kidney failure and death were 0.8-1.1 and 10.0-12.0 per 100 person-years, respectively. KDpredict outperformed KFRE in kidney failure risk prediction: 5-year IPA 27.8% (95% CI, 25.2%-30.6%) versus 18.1% (15.7%-20.4%) in Denmark and 30.5% (27.8%-33.5%) versus 14.2% (12.0%-16.5%) in Scotland.

KDpredict provides accurate risk predictions of kidney failure and death for people with incident G3bG4-CKD. To support more comprehensive decision-making in this patient population, it could be included in electronic medical records or made available online. Moreover, the KDpredict learning strategy can be adapted to local needs and amended regularly over time to account for changes in health system and care processes.

Source: KI Reports