论文部分内容阅读
目的:尝试建立中间型综合征的logistic回归模型,以便较为客观地预测中间型综合征(IMS)的早期发生。方法:连续观察2000年4月~2009年2月广西医科大学附属三院收治的急性有机磷农药中毒患者136例,按成组设计分为IMS组27例与非IMS组109例。观察指标为入院24 h内的农药种类,服用剂量,复能剂量,洗胃次数,血糖,血钾,胆碱脂酶(AchE),红细胞,肝功白蛋白,年龄,是否出现烟碱样症状,呼吸骤停,阿托品化及导泻。采用logistic逐步回归方法筛选变量建立回归模型。结果:建立预测模型为:[LnR/(1-R)]=-1.844±2.196烟碱样症状-0.053胆碱酯酶。ROC曲线下面积为0.884;Hosmer-Lemeshow拟合优度检验P值为0.469,表明模型拟合较好。结论:烟碱样症状和AchE作为IMS的两个主要因素建立回归模型后可应用于IMS的早期预测。
Objective: To establish a logistic regression model of intermediate syndrome in order to predict the occurrence of intermediate syndrome (IMS) more objectively. Methods: A total of 136 patients with acute organophosphorus pesticide poisoning admitted to the Third Affiliated Hospital of Guangxi Medical University from April 2000 to February 2009 were consecutively divided into two groups according to the group design: 27 cases in IMS group and 109 cases in non-IMS group. The observation index included the types of pesticides, dosage, energy dose, number of gastric lavage, blood glucose, serum potassium, cholinesterase (AchE), erythrocyte, hepatic albumin, age, nicotine-like symptoms , Respiratory arrest, atropine and catharsis. Logistic stepwise regression was used to select the variables to establish the regression model. RESULTS: The prediction model was established as [LnR / (1-R)] = - 1.844 ± 2.196 Nicotine-like symptoms -0.053 Cholinesterase. The area under the ROC curve was 0.884; Hosmer-Lemeshow goodness of fit test P value of 0.469, indicating that the model fit better. Conclusion: Nicotine-like symptoms and AchE, as the two main factors of IMS, can be applied to the early prediction of IMS after establishing regression model.