Development and Internal Validation of a Prediction Model for Intradialytic Hypotension in Patients with Uremia Undergoing Maintenance Hemodialysis

<title>Abstract</title> <p> <bold>Objective</bold> This study aimed to develop and internally validate a prediction model for intradialytic hypotension (IDH) in patients with uremia undergoing maintenance hemodialysis (MHD), thereby enabling early identification of high-risk patients and guiding individualized interventions. <bold>Methods</bold> Clinical data from 360 patients with uremia receiving regular MHD at the Blood Purification Center of the First Affiliated Hospital of Wannan Medical University (Yijishan Hospital of Wannan Medical University) between January 2022 and December 2024 were retrospectively reviewed. Patients were randomly divided into a training set (n = 252) and a validation set (n = 108) at a ratio of 7:3. IDH was defined as the primary outcome. Candidate predictors were first screened by univariable analysis and then entered into a multivariable logistic regression model. A nomogram was generated based on the final model. Model discrimination was assessed using receiver operating characteristic (ROC) curves; calibration was evaluated using the Hosmer-Lemeshow test and calibration curves; and clinical utility was assessed using decision curve analysis (DCA). Internal validation was performed in the randomly selected validation set. <bold>Results</bold> Among the 360 patients, 79 developed IDH, yielding an overall incidence of 21.94%; of these, 55 cases occurred in the training set and 24 in the validation set. Multivariable logistic regression identified age ≥65 years, diabetes, predialysis systolic blood pressure &lt;110 mmHg, ultrafiltration rate ≥10 mL·kg⁻¹·h⁻¹, and albumin &lt;35 g/L as independent risk factors. The final regression equation for the training set was: Logit(p) = −3.840 + 0.782 × age ≥65 years + 1.304 × diabetes + 1.532 × predialysis systolic blood pressure &lt;110 mmHg + 1.091 × ultrafiltration rate ≥10 mL·kg⁻¹·h⁻¹ + 2.040 × albumin &lt;35 g/L. The area under the receiver operating characteristic curve (AUC) for the training set was 0.873 (95% CI, 0.810–0.926), with a sensitivity of 83.64% and a specificity of 77.16%. The AUC for the validation set was 0.858 (95% CI, 0.757–0.939), with a sensitivity of 79.17% and a specificity of 85.71%. The Hosmer-Lemeshow test indicated good calibration in the training set (χ² = 1.472, p = 0.916). <bold>Conclusion</bold> Based on age, diabetes, predialysis systolic blood pressure, ultrafiltration rate, and albumin, the IDH risk prediction model for patients with uremia undergoing MHD demonstrated good discrimination and calibration, as well as potential clinical utility. This model may support early identification and stratified management of MHD patients at risk for IDH. </p>