Development and validation of a predictive model for end-stage renal disease in systemic lupus erythematosus patients

Systemic lupus erythematosus (SLE) affects many populations. This study aims to develop a predictive model and create a nomogram for assessing the risk of end-stage renal disease (ESRD) in patients diagnosed with SLE. Data from electronic health records of SLE patients treated at the Affiliated Hospital of North Sichuan Medical College between 2013 and 2023 were collected. The dataset underwent thorough cleaning and variable assignment procedures. Subsequently, variables were selected using one-way logistic regression and lasso logistic regression methods, followed by multifactorial logistic regression to construct nomograms. The model's performance was assessed using calibration, receiver operating characteristic (ROC), and decision curve analysis (DCA) curves. Statistical significance was set at P < 0.05. The predictive variables for ESRD development in SLE patients included anti-GP210 antibody presence, urinary occult blood, proteinuria, white blood cell count, complement 4 levels, uric acid, creatinine, total protein, globulin, glomerular filtration rate, pH, specific gravity, very low-density lipoprotein, homocysteine, apolipoprotein B, and absolute counts of cytotoxic T cells. The nomogram exhibited a broad predictive range. The ROC area under the curve (AUC) was 0.886 (0.858-0.913) for the training set and 0.840 (0.783-0.897) for the testing set, indicating good model performance. The model demonstrated both applicability and significant clinical benefits. The developed model presents strong predictive capabilities and considerable clinical utility in estimating the risk of ESRD in patients with SLE.