基于随机森林-Logistic回归分析脊柱外科住院病人下肢深静脉血栓预警因素及护理对策OA
Analysis of early warning factors and nursing strategies for deep vein thrombosis(DVT)in spinal surgery hospitalized patients based on random forest-Logistic regression
目的:基于随机森林-Logistic回归分析脊柱外科住院病人下肢深静脉血栓(DVT)预警因素及护理对策.方法:选取2021年8月—2023年8月医院脊柱外科住院的42例发生DVT病人(DVT组)与127例未发生DVT病人(无DVT组)为研究对象,比较两组临床资料,以随机森林算法对两组比较有统计学意义的DVT特征变量进行筛选与降维,Logistic回归分析脊柱外科住院病人DVT预警因素.结果:发生组年龄、体质指数、手术时间、术中出血及糖尿病、高脂血症、吸烟史、多发骨折、合并脊髓损伤、俯卧位手术、术中输血病人占比高于未发生组,卧床时间长于未发生组,接受间歇充气加压装置(IPC)、泵踝练习病人占比及RHDS评分低于未发生组(P＜0.05);随机森林算法根据平均基尼指数对15个变量进行重要性排序,当变量数为10时,袋外数据错误率最低,故将重要性排序前10个变量作为脊柱外科住院病人DVT的预测因素纳入多因素Logistic回归分析;逐步回归分析显示,糖尿病(OR=3.207)、高脂血症(OR=3.846)、多发骨折(OR=4.652)、合并脊髓损伤(OR=7.164)、手术体位(OR=4.741)、术中输血(OR=5.168)、卧床时间(OR=3.835)是脊柱外科住院病人DVT相关危险因素,接受IPC(OR=0.481)、泵踝练习(OR=0.314)、RHDS评分(OR=0.371)是脊柱外科住院病人DVT相关保护因素(P＜0.05).结论:合并脊髓损伤、术中输血、卧床时间、多发骨折、泵踝练习、接受IPC、糖尿病、出院准备度、高脂血症、手术体位是脊柱外科住院病人发生DVT重要性排序前10位预警因素,针对以上因素制定护理对策,有助于针对性预防DVT发生,促进病情良好转归.
Objective:To analyze the early warning factors and nursing strategies for deep vein thrombosis(DVT)in lower extremities of hospitalized patients in spinal surgery based on random forest-Logistic regression.Methods:A total of 42 patients with DVT(DVT group)and 127 patients without DVT(non-DVT group)hospitalized in the spinal surgery department of our hospital from August 2021 to August 2023 were selected.Clinical data of the two groups were compared.Random forest algorithm was used to screen and reduce the statistically significant DVT characteristic variables between the two groups.Logistic regression analysis was then performed on the early warning factors for DVT in hospitalized patients in spinal surgery.Results:The DVT group had higher rates of age,body mass index,operation time,intraoperative blood loss,diabetes,hyperlipidemia,smoking history,multiple fractures,spinal cord injury,prone position surgery,and intraoperative blood transfusion compared to the non-incidence group.DVT also had longer bed rest period,a lower rate of patients receiving intermittent pneumatic compression devices(IPC)and ankle pump exercises,and lower RHDS scores compared to the non-incidence group(P＜0.05).The random forest algorithm ranked the 15 variables by importance based on the average Gini index.The out-of-bag error rate was lowest when the number of variables was 10.Therefore,the top 10 variables in importance were included as predictors of DVT in hospitalized spinal surgery patients in the multivariate Logistic regression analysis.TIC regression analysis and stepwise regression analysis showed that diabetes mellitus(OR=3.207),hyperlipidemia(OR=3.846),multiple fractures(OR=4.652),concomitant spinal cord injury(OR=7.164),surgical position(OR=4.741),intraoperative blood transfusion(OR=5.168),and bed rest time(OR=3.835)were risk factors for deep vein thrombosis(DVT)in hospitalized spinal surgery patients.IPC(OR=0.481),ankle pump exercises(OR=0.314),and RHDS score(OR=0.371)were protective factors against DVT in hospitalized spinal surgery patients(P＜0.05).Conclusion:Concomitant spinal cord injury,intraoperative blood transfusion,bed rest time,multiple fractures,ankle pump exercises,IPC,diabetes mellitus,discharge readiness,hyperlipidemia,and surgical position were the top 10 risk factors for DVT in hospitalized spinal surgery patients.Developing nursing strategies targeting these factors can help prevent DVT and promote better patient outcomes.
杜艾林;侯文秀;张世君
250012,山东大学齐鲁医院250012,山东大学齐鲁医院250012,山东大学齐鲁医院
随机森林Logistic回归脊柱外科下肢深静脉血栓预警因素护理对策
random forestLogistic regressionspinal surgerydeep vein thrombosis of the lower extremitiesearly warning factorsnursing strategies
《全科护理》 2026 (11)
2051-2055,5
评论