论文部分内容阅读
目的筛选儿科多烯磷脂酰胆碱输液渗漏有统计学意义的危险因素,初步建立logistic回归模型,评价该模型的预测价值。方法选择儿科脉输注多烯磷脂酰胆碱患儿153例,其中103例作为模型训练组用于模型的创建,另外50例作为模型验证组用于模型的评判。采用单因素和多因素非条件Logistic回归分析模型训练组中患儿发生输液渗漏的危险因素。通过优化组合方式建立Logistic回归模型,对其稳定性进行验证,并将得到的预测模型代入验证数据集进行评价。结果年龄、同一血管进针次数、护士工作年限及用药时间4个危险因素在经过Logistic回归分析之后进入模型;利用该回归模型对模型验证组中50例多烯磷脂酰胆碱输注病人进行预报,其曲线下面积、灵敏性、特异性分别为0.983,92.3%,91.6%。结论 Logistic回归分析能够筛选出对儿科多烯磷脂酰胆碱输液渗漏有意义的危险因素;该Logistic回归模型对儿科多烯磷脂酰胆碱输液渗漏风险有初步预判的作用。
Objective To screen the pediatric polyene phosphatidylcholine infusion leakage statistically significant risk factors, the initial establishment of a logistic regression model to evaluate the predictive value of the model. Methods A total of 153 pediatric patients with polyene phosphatidylcholine were enrolled in this study. 103 of them were used as model training group for model creation and the other 50 were model validation groups for model evaluation. Univariate and multivariate non-conditional logistic regression models were used to analyze the risk factors of infusion leakage in the model training group. Logistic regression model is established through optimization and combination, and its stability is verified. The forecast model is substituted into the validation data set for evaluation. RESULTS: Four risk factors, including age, number of needle in the same vessel, working hours of the nurses and medication duration, entered the model after Logistic regression analysis. Fifty patients with polyene phosphatidylcholine infusion in the model validation group were predicted by this regression model The area under the curve, sensitivity and specificity were 0.983, 92.3% and 91.6% respectively. Conclusions Logistic regression analysis can screen out the risk factors of pediatric polyene phosphatidylcholine infusion leakage. The logistic regression model has preliminary predictive value for the risk of pediatric polyene phosphatidylcholine infusion leakage.