Analysis of the risk factors and construction of prediction model of hyponatremia in acute exacerbation of chronic obstructive pulmonary disease

Objective: To analyze the risk factors of hyponatremia in patients with acute exacerbation of chronic obstructive pulmonary disease(AECOPD) and construct a dynamic nomogram for predicting the risk of hyponatremia. Methods: A total of 224 patients with AECOPD were analyzed, and the Boruta algorithm was used to screen for important risk factors of hyponatremia.The area under the curve(AUC) of the variable was analyzed by using the receiver operating characteristic(ROC) curve.The nomogram of the risk of hyponatremia in AECOPD patients was constructed based on risk factors.The prediction performance of the nomogram was evaluated using internal datasets.The DynNom package was used to publish the nomogram to the network to construct a dynamic nomogram of the risk of hyponatremia in AECOPD patients. Results: Among the 224 cases, 50 cases developed hyponatremia, with an incidence rate of 22.3%.The Boruta algorithm found that the albumin(ALB), body mass index(BMI), erythrocyte sedimentation rate(ESR), type 2 diabetes mellitus(T2DM), C-reactive protein(CRP), blood glucose(FBG) and creatinine(Cr) were the independent risk factors of hyponatremia in patients with AECOPD.The AUC values of ALB, BMI, ESR, T2DM, CRP, FBG and Cr for diagnosing hyponatremia were 0.759(0.698-0.814), 0.772(0.712-0.826), 0.689(0.624-0.749), 0.625(0.558-0.689), and 0.503(0.435-0.570), 0.574(0.506-0.639), 0.516(0.448-0.583), respectively.A nomogram model for hyponatremia in patients with AECOPD was constructed.The predicted values of the correction curve and actual values showed good consistency.When the threshold probability of the decision curve was 2.7%-100%, the net benefit rate of nomogram in predicting hyponatremia in patients with AECOPD was relatively high. Conclusions: The ALB, BMI, ESR, T2DM, CPR, FBG, and Cr are the independent risk factors of hyponatremia in patients with AECOPD.Based on these factors, the constructed dynamic nomogram prediction model for hyponatremia in patients with AECOPD has good predictive and clinical application value.