ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: SA-PO359

Development of an Adaptive Clinical Web-Based Prediction Tool for Kidney Replacement Therapy in Children with CKD

Session Information

  • Pediatric Nephrology - III
    November 04, 2023 | Location: Exhibit Hall, Pennsylvania Convention Center
    Abstract Time: 10:00 AM - 12:00 PM

Category: Pediatric Nephrology

  • 1900 Pediatric Nephrology

Authors

  • Ng, Derek K., Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, United States
  • Matheson, Matthew, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, United States
  • Schwartz, George J., University of Rochester Medical Center, Rochester, New York, United States
  • Mendley, Susan R., National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Maryland, United States
  • Furth, Susan L., The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
  • Warady, Bradley A., Children's Mercy Kansas City, Kansas City, Missouri, United States

Group or Team Name

  • Chronic Kidney Disease in Children Study.
Background

Clinicians need improved prediction models to estimate time to kidney replacement therapy (KRT) for children with chronic kidney disease (CKD). We aimed to a) develop and validate a prediction tool using clinical variables for time to KRT from the Chronic Kidney Disease in Children (CKiD) study using statistical learning methods and b) to design an online calculator for clinical use.

Methods

We constructed an elementary model using eGFR, proteinuria and diagnosis as predictors in a parametric generalized gamma (GG) survival model and used penalized likelihood to identify optimal nonlinear combinations of these three. We then used random survival forest (RSF) to identify additional candidate predictors from a large panel of demographics and clinical markers. From the top predictors we used best subset selection, a learning method, to identify the best combination in an enriched GG model.

Results

Among 890 children, 9 candidate predictors were identified from 172 variables in RSF which were added to the elementary model. Best subset selection yielded an enriched model additionally based on blood pressure, 1 year change in eGFR, anemia, albumin, chloride and bicarbonate. Four additional partially enriched models were constructed for clinical situations with incomplete data and integrated as an adaptive tool (Figure). Discrimination was high in cross-validation (c= 0.86) and in an external European cohort (c= 0.85) with strong calibration. We designed an adaptive online tool for clinical use.

Conclusion

This clinical prediction tool for time to KRT in children was developed in a large, representative pediatric CKD cohort with an exhaustive evaluation of potential predictors and statistical learning methods. While models performed well internally and externally, further external validation of enriched models is needed.

Decision tree describing adaptive prediction model based on availability of data. AIC statistics reported for each model (lower corresponds to lower error).

Funding

  • NIDDK Support