To the Editor:
Park et al.1 addressed a clinically relevant question by testing the pediatric chronic kidney disease in children kidney replacement therapy risk-prediction tool in the KNOW-Ped chronic kidney disease cohort, providing what appears to be the first cross-population validation of this calculator in an East Asian pediatric population. Our concern is narrower than the overall value of the study. We think the data support promising transportability, but not the stronger inference of routine clinical use as yet.
The first issue is reporting consistency. In the original article by Park et al, the abstract gives group 1 as n = 433 and reports C-statistics of 0.911–0.972, whereas the Results section and Table 2 list group 1 as n = 436, and Tables 3 and 4 present lower C-statistics for groups 1 and 2.1 The number of participants excluded because of missing essential predictors was not reported, which makes the final analytic samples harder to reconstruct.1 For an external validation study, a brief clarification of the definitive analysis set and the corrected performance estimates would materially improve interpretability.2,3
The second issue concerns calibration in relation to possible clinical use. Although the calibration slopes were all above 0.9, every Greenwood–Nam–D’Agostino test was significant, and the authors themselves note that simple recalibration may be useful before widespread implementation.1 Because calibration is central to individualized risk prediction, these findings seem more consistent with a model that travels reasonably well but still needs setting-specific updating.3,4
The statement that the calculator may improve clinical decision-making or outcomes may therefore go somewhat beyond the present validation data, particularly in the absence of formal clinical utility analyses, such as decision-curve or related net-benefit assessments.3,5 This seems especially relevant because the validation cohort had a steeper decline in estimated glomerular filtration rate than the development cohort. Clarifying the final analytic sample and, if feasible, reporting recalibrated performance would help readers judge more precisely where this tool is ready to inform care in Korean children with chronic kidney disease.
Disclosure
All the authors declared no competing interests.
Declaration of AI and AI-Assisted Technologies in the Writing Process
Generative AI tools were used solely for language refinement and formatting assistance. All scientific interpretation, critique, and scholarly content were independently developed by the authors.
References
- 1.Park P.G., Kim J., Choi N., et al. Cross-population validation of the pediatric CKD risk-prediction tool. Kidney Int Rep. 2026;11 doi: 10.1016/j.ekir.2026.106373. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Collins G.S., Reitsma J.B., Altman D.G., Moons K.G.M. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD Statement. BMJ. 2015;350 doi: 10.1136/bmj.g7594. [DOI] [PubMed] [Google Scholar]
- 3.Riley R.D., Archer L., Snell K.I.E., et al. Evaluation of clinical prediction models (part 2): How to undertake an external validation study. BMJ. 2024;384 doi: 10.1136/bmj-2023-074820. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Van Calster B., Nieboer D., Vergouwe Y., De Cock B., Pencina M.J., Steyerberg E.W. A calibration hierarchy for risk models was defined: from utopia to empirical data. J Clin Epidemiol. 2016;74:167–176. doi: 10.1016/j.jclinepi.2015.12.005. [DOI] [PubMed] [Google Scholar]
- 5.Kerr K.F., Brown M.D., Zhu K., Janes H. Assessing the clinical impact of risk prediction models with decision curves: guidance for correct interpretation and appropriate use. J Clin Oncol. 2016;34:2534–2540. doi: 10.1200/JCO.2015.65.5654. [DOI] [PMC free article] [PubMed] [Google Scholar]