Derek K Ng1, Anthony A Portale2, Susan L Furth3, Bradley A Warady4, Alvaro Muñoz5. 1. Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. Electronic address: dng@jhu.edu. 2. Division of Nephrology, Department of Pediatrics, University of California, San Francisco, CA. 3. Division of Nephrology, Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA; Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, Philadelphia, PA. 4. Division of Nephrology, Department of Pediatrics, Children's Mercy Kansas City, Kansas City, MO. 5. Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
Abstract
PURPOSE: Coefficients of determination (R2) for continuous longitudinal data are typically reported as time constant, if they are reported at all. The widely used mixed model with random intercepts and slopes yields the total outcome variance as a time-varying function. We propose a generalized and intuitive approach based on this variance function to estimate the time-varying predictive power (R2) of a variable on outcome levels and changes. METHODS: Using longitudinal estimated glomerular filtration rate (eGFR) from the Chronic Kidney Disease in Children Study, linear mixed models characterized the R2 for two chronic kidney disease (CKD) risk factors measured at baseline: a traditional marker (proteinuria) and a novel marker (fibroblast growth factor 23 [FGF23]). RESULTS: Time-varying R2 divulged different disease processes by risk factor and diagnoses. Among children with glomerular CKD, time-varying R2 for proteinuria had significant upward trends, suggesting increasing power to predict eGFR change, but crossed with FGF23, which was higher up to 2.5 years from baseline. In contrast, among those with nonglomerular CKD, proteinuria explained more than FGF23 at all times, and time-varying R2 for each risk factor was not substantially different from time-constant estimates. CONCLUSIONS: Proteinuria and FGF23 explained substantial eGFR variability over time. Time-varying R2 can characterize predictive roles of risk factors on disease progression, overcome limitations of time-constant estimates, and are easily derived from mixed effects models.
PURPOSE: Coefficients of determination (R2) for continuous longitudinal data are typically reported as time constant, if they are reported at all. The widely used mixed model with random intercepts and slopes yields the total outcome variance as a time-varying function. We propose a generalized and intuitive approach based on this variance function to estimate the time-varying predictive power (R2) of a variable on outcome levels and changes. METHODS: Using longitudinal estimated glomerular filtration rate (eGFR) from the Chronic Kidney Disease in Children Study, linear mixed models characterized the R2 for two chronic kidney disease (CKD) risk factors measured at baseline: a traditional marker (proteinuria) and a novel marker (fibroblast growth factor 23 [FGF23]). RESULTS: Time-varying R2 divulged different disease processes by risk factor and diagnoses. Among children with glomerular CKD, time-varying R2 for proteinuria had significant upward trends, suggesting increasing power to predict eGFR change, but crossed with FGF23, which was higher up to 2.5 years from baseline. In contrast, among those with nonglomerular CKD, proteinuria explained more than FGF23 at all times, and time-varying R2 for each risk factor was not substantially different from time-constant estimates. CONCLUSIONS:Proteinuria and FGF23 explained substantial eGFR variability over time. Time-varying R2 can characterize predictive roles of risk factors on disease progression, overcome limitations of time-constant estimates, and are easily derived from mixed effects models.
Authors: Bradley A Warady; Alison G Abraham; George J Schwartz; Craig S Wong; Alvaro Muñoz; Aisha Betoko; Mark Mitsnefes; Frederick Kaskel; Larry A Greenbaum; Robert H Mak; Joseph Flynn; Marva M Moxey-Mims; Susan Furth Journal: Am J Kidney Dis Date: 2015-03-19 Impact factor: 8.860
Authors: Craig S Wong; Christopher B Pierce; Stephen R Cole; Bradley A Warady; Robert H K Mak; Nadine M Benador; Fredrick Kaskel; Susan L Furth; George J Schwartz Journal: Clin J Am Soc Nephrol Date: 2009-03-18 Impact factor: 8.237
Authors: Arun S Karlamangla; Dana Miller-Martinez; Carol S Aneshensel; Teresa E Seeman; Richard G Wight; Joshua Chodosh Journal: Am J Epidemiol Date: 2009-07-15 Impact factor: 4.897
Authors: Bradley M Appelhans; Eisuke Segawa; Imke Janssen; Rasa Kazlauskaite; Rebecca C Thurston; Tené T Lewis; Howard M Kravitz Journal: Ann Epidemiol Date: 2013-12-28 Impact factor: 3.797
Authors: Guillermo Hidalgo; Derek K Ng; Marva Moxey-Mims; Maria Lourdes Minnick; Tom Blydt-Hansen; Bradley A Warady; Susan L Furth Journal: Am J Kidney Dis Date: 2013-08-07 Impact factor: 8.860
Authors: Susan L Furth; Stephen R Cole; Marva Moxey-Mims; Frederick Kaskel; Robert Mak; George Schwartz; Craig Wong; Alvaro Muñoz; Bradley A Warady Journal: Clin J Am Soc Nephrol Date: 2006-07-19 Impact factor: 8.237
Authors: Joseph T Flynn; Mark Mitsnefes; Christopher Pierce; Steven R Cole; Rulan S Parekh; Susan L Furth; Bradley A Warady Journal: Hypertension Date: 2008-08-25 Impact factor: 10.190
Authors: George J Schwartz; Jennifer L Roem; Stephen R Hooper; Susan L Furth; Donald J Weaver; Bradley A Warady; Michael F Schneider Journal: Pediatr Nephrol Date: 2022-06-01 Impact factor: 3.714
Authors: Amy J Kogon; Jennifer Roem; Michael F Schneider; Mark M Mitsnefes; Babette S Zemel; Bradley A Warady; Susan L Furth; Nancy M Rodig Journal: Pediatr Nephrol Date: 2022-08-26 Impact factor: 3.651