急性粘连性小肠梗阻手术治疗选择的预测模型

Prediction model of surgical treatment selection for acute adhesive small intestinal obstruction

目的探讨急性粘连性小肠梗阻选择手术治疗的影响因素,同时建立选择手术治疗的预测模型,为临床决策提供参考。方法回顾性收集陕西省人民医院2019年1月至2022年12月期间收治且符合本研究纳入和排除标准的急性粘连性肠梗阻患者,根据治疗方式选择分为手术治疗和保守治疗,比较手术治疗和保守治疗患者临床病理因素方面的差异,同时将其有统计学意义(P<0.05)的因素或基于专业知识判断可能有临床意义的因素纳入进行多因素logistic回归分析,以筛选影响手术治疗选择的因素,并且采用筛选出的影响因素构建logistic回归预测模型方程,采用受试者操作特征曲线下面积及其95%可信区间评估该预测模型方程的预测效能。结果共纳入231例急性粘连性肠梗阻患者,其中行手术治疗患者117例(50.6%)、行保守治疗患者114例(49.4%)。手术治疗和保守治疗患者在16个临床病理因素方面比较差异有统计学意义(P<0.05),包括体质量指数,入院前有高热、肠型及腹膜炎体征,除去年龄评分的急性生理与慢性健康评估Ⅱ(acute physiology and chronic health evaluationⅡ,APACHEⅡ)评分,腹部手术史和腹部手术史次数,入院前就医次数及术前保守治疗时间,X射线平片示气液平面,CT检查提示重度小肠梗阻和粘连带形成方面,以及白细胞计数、中性粒细胞百分比、白蛋白及尿素氮;对此进一步进行多因素logistic回归分析结果显示,伴有腹膜炎体征(β=1.778,P=0.028)、术前有腹部手术史(β=1.394,P=0.022)、CT检查提示粘连带形成(β=1.321,P=0.010)和重度小肠梗阻(β=1.183,P=0.018)、白细胞计数(β=0.524,P<0.001)、除去年龄评分的APACHEⅡ评分(β=0.291,P<0.001)及体质量指数(β=0.191,P=0.011)对急性粘连性小肠梗阻患者是否选择手术治疗产生正向影响,而术前白蛋白(β=–0.101,P=0.023)和保守治疗时间(β=–0.391,P<0.001)对急性粘连性小肠梗阻患者是否选择手术治疗产生负向影响。根据这9个影响因素构建的logistic回归预测模型方程判断选择手术治疗的准确度为84.8%、特异度为71.1%、敏感度为77.7%,该方程区分是否选择手术治疗的受试者操作特征曲线下面积(95%可信区间)为0.942(0.914,0.970)。结论从本研究初步研究结果提示,对伴有腹膜炎体征、有腹部手术史、CT检查提示有粘连带形成和重度小肠梗阻、术前白细胞计数升高、除去年龄评分的APACHEⅡ评分高、体质量指数较高、术前低白蛋白水平及术前保守治疗时间短的急性粘连性小肠梗阻患者建议选择手术治疗,并且根据这些特征构建的logistic回归预测模型方程对选择手术治疗或保守治疗患者具有较好的区分度。

Objective To explore the risk factors affecting operation treatment selection of acute adhesive small bowel obstruction(ASBO),and establish a prediction model of surgical treatment selection to provide a guidance for clinical decision-making.Methods The patients with acute ASBO admitted to this hospital and met the inclusion and exclusion criteria,from January 2019 to December 2022,were retrospectively collected,and the patients were assigned into the surgical treatment and conservative treatment according to the treatment selection.The differences in the clinicopathologic factors between the patients with surgical treatment and conservative treatment were compared.Meanwhile,the factors with statistical differences(P<0.05)or the factors with clinical significance judged based on professional knowledge were included to screen the influencing factors of surgical treatment selection using the multivariate logistic regression analysis,and the selected influencing factors were used to construct the logistic regression prediction model equation.The area under the receiver operating characteristic curve(AUC)and its 95%confidence interval(95%CI)was used to evaluate the prediction efficiency of the prediction model equation.Results A total of 231 patients with acute ASBO were included,117(50.6%)of whom underwent surgical treatment and 114(49.4%)underwent conservative treatment.In all 16 clinicopathologic factors between the patients with surgical treatment and conservative treatment had statistical differences(P<0.05)including the body mass index(BMI),preopeative high fever,intestinal type,sign of peritonitis,acute physiology and chronic health evaluationⅡ(APACHEⅡ)score excluded age scoring,abdominal surgery history and times of abdominal surgery history,times of pre-admission seek medical advice and preoperative conservative treatment time,the air-liquid level by X-ray plain film,and severe small bowel obstruction and adhesive bands by CT examination,as well as the white blood cell count(WBC),neutrophil percentage,albumin(ALB),and urea nitrogen.The multivariate logistic regression analysis showed that the acute ASBO accompanied by sign of peritonitis(β=1.778,P=0.028),history of abdominal surgery(β=1.394,P=0.022),and adhesive bands(β=1.321,P=0.010)and severe small bowel obstruction(β=1.183,P=0.018)by CT examination,WBC(β=0.524,P<0.001),APACHEⅡscore excluded age scoring(β=0.291,P<0.001),and BMI(β=0.191,P=0.011)had positive impacts on adopting surgical treatment,while preoperative ALB(β=–0.101,P=0.023)and conservative treatment time(β=–0.391,P<0.001)had negative impacts on adopting surgical treatment.The accuracy,specificity,and sensitivity of the logistic regression prediction model equation constructed according to these 9 influencing factors were 84.8%,71.1%,and 77.7%,respectively.The AUC(95%CI)of the prediction model equation to distinguish selection of surgical treatment from conservative treatment was 0.942(0.914,0.970).Conclusions According to the preliminary results of this study,surgical treatment is recommended for patients with acute ASBO accompanied by signs of peritonitis,history of abdominal surgery,adhesive bands and severe small bowel obstruction by CT,increased preoperative WBC,high APACHEⅡscore excluded age scoring,high BMI,preoperative low ALB level,and shorter preoperative conservative treatment time.And the logistic prediction model equation constructed according to these characteristics in this study has a good discrimination for patients with surgical treatment or conservative treatment selection.