To analyze the influencing factors of the duration of stress hyperglycemia in patients with acute pancreatitis,and to construct its predictive model.

By using the convenience sampling method,a total of 206 patients with acute pancreatitis admitted to Suqian Hospital of Jiangsu Provincial People's Hospital from March 2022 to November 2024 were selected as the research subjects.The clinical data of the patients were collected.They were divided into the stress hyperglycemia≥48 hours group and the stress hyperglycemia<48 hours group according to the duration.Univariate and multivariate Logistic regression analysis were used to explore the influencing factors of the duration of stress hyperglycemia in patients.A risk prediction model for stress hyperglycemia of patients ≥48 hours was constructed based on the decision tree algorithm.The predictive efficacy of the Logistic regression model and the decision tree model was evaluated by using the receiver operating characteristic curve(ROC).

Among 206 patients with acute pancreatitis,41 cases(19.90%) had stress hyperglycemia for ≥48 hours.There were statistically significant differences in hypertension,admission blood glucose level,Acute Physiology and Chronic Health EvaluationⅡ(APACHEⅡ) scores,systemic inflammatory response syndrome(SIRS),pain,and nutritional risk between the stress hyperglycemia ≥48 hours group and the stress hyperglycemia <48 hours group(P<0.05).Multivariate Logistic regression analysis showed that the patients' admission blood glucose level,APACHEⅡ scores,SIRS,pain,and nutritional risk were the influencing factors of stress hyperglycemia ≥48 hours in patients with acute pancreatitis(P<0.05).The decision tree model screened out four influencing factors(pain,blood glucose level at admission,APACHEⅡ scores,and nutritional risk),with a prediction accuracy rate of 86.4%.The correlation between pain and stress hyperglycemia ≥48 hours in patients with acute pancreatitis was relatively stronger.When the pain score was >2 points,the admission blood glucose level was >13.07 mmol/L.The APACHEⅡ score was >7.The risk probability of stress hyperglycemia ≥48 hours in patients with acute pancreatitis was the highest(88.2%).The area under curve(AUC) of the Logistic regression model was 0.824,with a sensitivity of 82.9% and a specificity of 68.5%.The AUC predicted by the decision tree model was 0.828,with a sensitivity of 78.0% and a specificity of 77.6%.There was no statistically significant difference in terms of the predictive efficacy between the two models(Z=0.075,P=0.940).

The duration of stress hyperglycemia in patients with acute pancreatitis was influenced by multiple factors.Both the decision tree model and the Logistic regression model have good predictive value.However,the decision tree model demonstrates the interactive influence of various factors.Therefore,the two models could be used in combination to promote the targeted and specific natures of nursing intervention strategies from different levels.