目的 探讨脾切除术病人术后发生胰瘘的影响因素,并构建预测模型.方法 回顾性分析第九○九医院(厦门大学附属东南医院)2019年6月至2022年6月收治的89例行脾切除术病人临床资料,根据术后是否发生胰瘘分为胰瘘组(13例)和无胰瘘组(76例),采用单因素和logistic多因素回归分析术后胰瘘发生独立影响因素,并建立风险预测模型,绘制受试者操作特征(receiver operator characteristic,ROC)曲线,计算曲线下面积(area under the curve,AUC),评估该模型对胰瘘发生的预测价值,分析胰瘘病人的预后情况.结果 单因素分析结果发现体重指数(body mass index,BMI)≥27 kg/m2、急诊手术、脾破裂、使用切割闭合器、高年资主治医师主刀手术、脾脏长度≥15 cm、术中暴露不佳的病人胰瘘发生率更高(P＜0.05);多因素回归分析结果发现胰瘘独立影响因素为:BMI≥27 kg/m2[HR=17.194,95％CI(2.151,137.407),P＜0.05];脾破裂[HR=7.551,95％CI(1.891,64.017),P＜0.05];高年资主治医师主刀手术[HR=7.250,95％CI(1.888,59.188),P＜0.05];术中暴露不佳[HR=13.803,95％CI(1.515,125.749),P＜0.05].构建胰瘘风险模型为 Logit(P)=-18.155±2.845×X1±2.022×X2±1.981×X3±2.625×X4(X1 为 BMI、X2为脾破裂、X3为主刀医师资历、X4为术中暴露不佳).ROC曲线分析发现该模型预测胰瘘AUC为0.943,灵敏度为0.884,特异度为0.859,95％CI(0.875,1.000),P＜0.05.13例胰瘘病人中,8例为生化漏、5例为B级胰瘘,保守治疗后均好转.结论 肥胖、脾破裂、主刀医师资历和术中暴露不佳是脾切除术后胰瘘发生的独立影响因素,根据多因素分析结果构建风险预测模型对术后胰瘘发生具有较好的预测价值.
Analysis of influencing factors and construction of prediction model for pancreatic fistula after splenectomy
Objective To explore the influencing factors of postoperative pancreatic fistula(PF)and construct a prediction model in post-splenectomy patients.Methods The relevant clinical data were retrospectively reviewed for 89 hospitalized patients after splenectomy from June 2019 to June 2022.According to the occurrence of postoperative PF,they were assigned into two groups of PF(n=13)and pancreatic non-fistula(n=76).Univariate and Logistic multivariate regression were employed for examining the independent influencing factors of postoperative PF and a risk prediction model was established.Receiver operator characteristic(ROC)curve was plotted and area under the curve(AUC)calculated for evaluating the predictive value of the model for the occurrence of PF and analyze the prognosis of PF.Results Univariate analysis revealed that the incidence of PF was higher in patients with BMI ≥27 kg/m2,undergoing emergency surgery,splenic rupture,incision closure device,seniority of attending surgeon,splenic length ≥15 cm and poor exposure(P＜0.05).Multivariate regression analysis indicated that BMI ≥27 kg/m2(HR:17.194,95％CI:2.151,137.407,P＜0.05),splenic rupture(HR:7.551,95％CI:1.891,64.017,P＜0.05),seniority of attending surgeon(HR:7.250,95％CI:1.888,59.188,P＜0.05)and poor exposure(HR:13.803,95％CI:1.515,125.749,P＜0.05)were independent influencing factors of PF.The risk model of PF was Logit(P)=-18.155+2.845× X1+2.022×X2+1.981×X3+2.625×X4(X1 BMI,X2 splenic rupture,X3 primary surgeon's qualifications & X4 poor exposure).ROC curve analysis showed that area under the curve of PF as predicted by the model was 0.943 with a sensitivity of 0.884 and a specificity of 0.859(95％CI:0.875,1.000,P＜0.05).Among 13 PF patients,there were biochemical fistula(n=8)and grade B(n=5).All of them improved after conservative measures.Conclusion Obesity,splenic rupture,seniority of attending surgeon and poor intraoperative exposure are independent influencing factors of PF post-splenectomy.The risk prediction model based upon the results of multi-factorial analysis has an excellent value in predicting the incidence of postoperative PF.
万方数据
维普
