Optimal prediction of peritoneal dialysis to hemodialysis transition: a penalized fine-gray model
5 affiliations
- 1Department of Nephrology, Daping Hospital, Army Medical University, Chongqing, China
- 2School of Mathematics and Statistic, Southwest University, Chongqing, China
- 3Department of Nephrology, Kaizhou District People’s Hospital, Chongqing, China
- 4Department of Nephrology, Yongchuan Hospital of Chongqing Medical University, Chongqing, China
- 5State Key Laboratory of Trauma, Burns and Combined Injury, Wound Trauma Medical Center, Army Medical University, Chongqing, China
Abstract
Background: Traditional Cox proportional hazards models overestimate the risk of progression from peritoneal dialysis (PD) to hemodialysis (HD). This study aims to screen high-risk conversion populations using competing risk models.
Methods: In this multicenter retrospective cohort study, we enrolled patients aged ≥16 years who were undergoing regular PD for >1 month; transition from PD to HD was the target event. When death was recognized as a competing event, we employed cause-specific proportional hazards and Fine-Gray models for multivariate analysis. A conventional LASSO-Cox model was also established for comparative purposes. Model performance was evaluated using time-dependent area under the curve (AUC) and C-index curves. The best model was identified to construct the cumulative incidence nomogram for PD to HD at the 3-, 6-, and 9-year time points.
Results: We enrolled 700 valid samples that were randomly allocated at a ratio of 7:3 to the training (n = 491) and test sets (n = 209). The penalized Fine-Gray model was identified as the optimal model (AUC: 0.716, C-index: 0.662). Body mass index, history of peritonitis hospitalization, and nutrition-related parameters were identified as optimal predictors for transitioning from PD to HD. AUC values for the penalized Fine-Gray model at 3, 6, and 9 years were 0.675, 0.715, and 0.714, respectively. We constructed a nomogram for 3, 6, and 9 years based on the penalized Fine-Gray model, which demonstrated significant clinical utility.
Conclusion: The penalized Fine-Gray model serves as the optimal choice for predicting the transition from PD to HD, and the nomogram constructed based on this model exhibits favorable clinical applicability.
Related
- Catheter-related deep vein thrombosis complicated by iatrogenic pulmonary embolism after thrombectomy leading to hemodialysis failure and successful rescue by automated peritoneal dialysis : A case report
- Beyond solute clearance: the kidney–gut–immune–nutrition axis in hemodialysis and peritoneal dialysis and a proposed framework for integrated risk stratification
- Peritoneal dialysis-associated peritonitis caused by Ureaplasma urealyticum : A case report
- Successful Treatment of an Atypical Mycobacterial Exit Site Infection in a Patient on Peritoneal Dialysis: A Case Report
- Peritoneal Dialysis-Associated Peritonitis due to Pasteurella multocida: A Rare Case of Stray-Animal Exposure and Fluoroquinolone Resistance
BibTeX
@article{wan2026optimal,
title = {{Optimal prediction of peritoneal dialysis to hemodialysis transition: a penalized fine-gray model}},
author = {Jingfang Wan and Qingyu Wei and Dong Wu and Shihui Hou and Kaizhen Yv and Rui Tang and Jian He and Xuelian Chen and Yang Li and Yulin Hu and Jie Yang and Tingting Li and Kehong Chen},
journal = {Frontiers in Medicine},
year = {2026},
month = oct,
doi = {10.3389/fmed.2026.1843381},
url = {https://doi.org/10.3389/fmed.2026.1843381},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Optimal prediction of peritoneal dialysis to hemodialysis transition: a penalized fine-gray model
AU - Jingfang Wan
AU - Qingyu Wei
AU - Dong Wu
AU - Shihui Hou
AU - Kaizhen Yv
AU - Rui Tang
AU - Jian He
AU - Xuelian Chen
AU - Yang Li
AU - Yulin Hu
AU - Jie Yang
AU - Tingting Li
AU - Kehong Chen
PY - 2026
DA - 2026/10/21
JO - Frontiers in Medicine
DO - 10.3389/fmed.2026.1843381
UR - https://doi.org/10.3389/fmed.2026.1843381
AB - Background: Traditional Cox proportional hazards models overestimate the risk of progression from peritoneal dialysis (PD) to hemodialysis (HD). This study aims to screen high-risk conversion populations using competing risk models. Methods: In this multicenter retrospective cohort study, we enrolled patients aged ≥16 years who were undergoing regular PD for >1 month; transition from PD to HD was the target event. When death was recognized as a competing event, we employed cause-specific proportional hazards and Fine-Gray models for multivariate analysis. A conventional LASSO-Cox model was also established for comparative purposes. Model performance was evaluated using time-dependent area under the curve (AUC) and C-index curves. The best model was identified to construct the cumulative incidence nomogram for PD to HD at the 3-, 6-, and 9-year time points. Results: We enrolled 700 valid samples that were randomly allocated at a ratio of 7:3 to the training (n = 491) and test sets (n = 209). The penalized Fine-Gray model was identified as the optimal model (AUC: 0.716, C-index: 0.662). Body mass index, history of peritonitis hospitalization, and nutrition-related parameters were identified as optimal predictors for transitioning from PD to HD. AUC values for the penalized Fine-Gray model at 3, 6, and 9 years were 0.675, 0.715, and 0.714, respectively. We constructed a nomogram for 3, 6, and 9 years based on the penalized Fine-Gray model, which demonstrated significant clinical utility. Conclusion: The penalized Fine-Gray model serves as the optimal choice for predicting the transition from PD to HD, and the nomogram constructed based on this model exhibits favorable clinical applicability.
SN - 2296-858X
ER -