Construction and validation of a risk prediction model for hyponatremia caused by cervical spinal cord injury.

Abstract

To investigate the risk factors of hyponatremia in patients with cervical spinal cord injury, and to construct and validate the nomogram risk prediction model. Patients with cervical spinal cord injury who were admitted to a tertiary hospital in Xi'an from January 2019 to July 2024 were retrospectively enrolled. The outcome measure was whether hyponatremia was triggered. Binary logistic regression analysis was used to determine the independent risk variables for hyponatremia and to construct a nomogram. Bootstrap was used for internal validation to evaluate predictive performance and clinical applicability by Receiver Operating Characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). A total of 318 patients from January 2019 to December 2023 were enrolled in the modeling group, and 109 patients from January 2024 to July 2024 were enrolled in the external validation group.Univariate analysis, LASSO regression, and binary logistic regression analysis showed that White Blood Cell (WBC), concomitant craniocerebral injury, assisted respiration, and injury level are risk factors for hyponatremia after cervical spinal cord injury, while Albumin (ALB) is a protective factor.The area under the ROC curve was 0.879 (95% CI:0.839-0.919) for the modeling group and 0.875 (95% CI:0.8536-0.9141) for internal validation. and 0.793 (95% CI:0.694-0.891) in the external validation group. The model calibration curves show good agreement and the DCA indicates a high expected net benefit. The incidence of hyponatremia caused by cervical spinal cord injury is high, and the predictive factors include ALB, WBC, craniocerebral injury, assisted breathing, and extent of injury. The nomogram model is scientific and practical for predicting hyponatremia in patients with cervical spinal cord injury.