Analysis of risk factors for chronic kidney disease complicated by renal failure and constructing a nomogram model

Objective  To explore the risk factors associated with the occurrence of renal failure and column line graph model for predicting concurrent renal failure in patients with chronic kidney disease（CKD）.

Methods  The raw data were collected from the Japanese CKD-ROUTE study for prediction model development and internal validation while the external validation set obtained data from a sampling of patients from January 2013 to December 2018. Cox proportional risk regression was employed for column line graph modeling in R software. Finally model discrimination, calibration and clinical value were tested with R software.

Results  The development and internal validation datasets included 797 patients191 progressive cases（23.96%） and 341 patients89 progressive cases（26.10%） while the external validation dataset included 297 patients108 progressive cases（36.36%）. Columnar line graph model were developed with age, estimated glomerular filtration rate（eGFR）, hemoglobin, blood albumin and test paper proteinuria for predicting the 3-year probability of no adverse outcome. C statistic for column line plot was 0.90（95%CI:0.89-0.92） for development dataset,0.91（95%CI:0.89-0.94） for internal validation dataset and 0.83（95%CI:0.78-0.88）.The model was well-calibrated and examined for decision curves.

Conclusion  This visual predictive line graph model may predict accurately 3-year kidney failure outcomes in CKD patients. Thus clinical practitioners have a convenient and widely applicable tool.