Kidney Failure Risk Equation shows poor calibration predicting ESKD in children with CKD
Click Here to Manage Email Alerts
The Kidney Failure Risk Equation showed poor calibration when predicting end-stage kidney disease in children with chronic kidney disease, according to data published in the American Journal of Kidney Diseases.
“As epidemiologists, we try to support pediatric nephrologists by developing and evaluating the utility and validity of clinical tools,” Gayathri Menon, MHS, and Derek K. Ng, PhD, from the department of epidemiology at Johns Hopkins Bloomberg School of Public Health, told Healio. “The Kidney Failure Risk Equation (KFRE) is the most popular clinical prognostic tool for nephrologists; it was developed in an older population (mean age of 70 years) but its application to children was less clear. The structure of the KFRE suggested that it may not be appropriate for pediatric practice. In previous work, the KFRE had showed strong discrimination meaning that those who ultimately experienced ESKD had higher KFRE-estimated probabilities of kidney failure collectively than those who did not experience ESKD; however, the calibration of the instrument wasn’t well characterized. Calibration refers to whether the predicted risk corresponds to real-life risk observed in data. The ongoing multicenter Chronic Kidney Disease in Children (CKiD) cohort was well-suited to answer this question of calibration because of the long-term follow-up through end-stage kidney disease.”
Researchers replicated the analytic structure of CKiD with for 817 children with CKD. Their analysis extended the follow-up period and included an estimation of albuminuria using pediatric-based equations.
Additionally, researchers used a formal statistical test to quantify calibration of the KFRE in the cohort, which had not been performed in CKiD. Patients provided data to calculate four-variable KFRE. A total of 799 patients completed data for eight-variable calculations.
Researchers calculated observed 2- and 5-year risk on KFRE for the four-variable and eight-variable equations.
Ultimately, analyses revealed the predicted KFRE risk was different than the observed risk, which the researchers noted is indicative of poor calibration. The KFRE overestimated the observed risk among those with higher KRFE scores for the 2-year (four- and eight-variables) KFRE. Similarly, the KFRE underestimated risk at lower predicted probabilities and overestimated risk at higher probabilities in the 5-year equation. Although calibration was poor, researchers wrote that discrimination was high.
“The KFRE did not demonstrate strong calibration in the CKiD cohort,” Menon and Ng told Healio. “However, there are other pediatric-specific risk calculators that pediatric nephrologists can use. A joint collaboration between the North American CKiD and European ESCAPE pediatric cohorts developed a risk calculator using CKD diagnosis, GFR and proteinuria which was published in the American Journal of Kidney Diseases in 2018. The CKiD study also published an enriched calculator using more predictors in American Journal of Kidney Diseases in 2015. Both equations are available online (https://statepi.jhsph.edu/ckid/investigator-resources/ckid-calculators/), and the latter is available on the National Kidney Foundation website (https://www.kidney.org/professionals/kdoqi/gfr_calculatorPedRisk). As with any clinical tool, each calculator has strengths and limitations, but these risk calculators were developed in and for children with CKD.”
Collapse
Read more about
Click Here to Manage Email Alerts