肝胆胰术后患者非计划重返SICU的危险因素及预测模型

Readmission to surgical intensive care unit after hepatobiliary-pancreatic surgery: risk factors and prediction

目的探讨肝胆胰术后患者非计划重返外科重症加强治疗病房(SICU)的病因、危险因素,并建立预测患者重返概率数学模型。方法选择2013年1月至2018年8月青岛大学附属医院SICU收治的首次入住SICU,经过治疗由手术医师和SICU医师共同评估后转出SICU,后因病情变化再次转回SICU的肝胆胰腺术后患者。分别统计3 d和7 d非计划重返SICU的情况。以7 d重返SICU患者作为重返SICU病因、危险因素及构建预测模型的研究对象,并按照1∶1筛选未重返患者作为对照。观察指标包括一般临床资料、既往史、首次转入SICU前手术情况、首次SICU住院时间及首次转出SICU时的指标(包括5大类、共43项指标)。先经Logistic回归分析筛选出重返SICU的危险因素,再采用R 3.4.1软件绘制列线图(Nomogram),对非计划重返SICU进行预测,通过自助法抽样检验评价Nomogram的分类性能。结果 SICU转出肝胆胰术后患者763例,其中3 d非计划重返率为2.10%(16/763),7 d非计划重返率为3.28%(25/763)。导致患者7 d重返SICU的病因中,感染占56.00%(14/25),心力衰竭占16.00%(4/25),梗死占12.00%(3/25),出血占12.00%(3/25),手术切口裂开占4.00%(1/25);导致患者3 d重返SICU的病因中,感染占56.25%(9/16),心力衰竭占18.75%(3/16),梗死占12.50%(2/16),出血占12.50%(2/16)。Nomogram分析显示,非计划重返SICU的危险因素包括首次SICU住院时间、高血压史以及转出SICU时的日常生活能力评分(ADL)、白细胞计数(WBC)、动脉血氧分压(PaO2)、凝血酶原时间(PT)、纤维蛋白原(FIB)。对绘制的Nomogram图进行自助法抽样检验,结果显示,一致性指数(C-index)为0.962〔95%可信区间(95%CI)=0.869~1.057〕,说明模型分类性能良好。结论感染、心力衰竭、梗死和出血是导致肝胆胰术后患者非计划重返SICU的主要病因。患者重返SICU的高危因素包括首次SICU住院时间、高血压史及转出SICU时的ADL评分、WBC、PaO2、PT、FIB。基于以上危险因素建立的预测模型能够有效预测肝胆胰术后患者从SICU转出时的重返概率。

Objective To find the pathogenies and risk factors related to surgical intensive care unit (SICU) readmission for patients who underwent hepatobiliary-pancreatic surgery, and to develop a predictive model for determining patients who are likely to be readmitted to SICU. Methods The patients who admitted to SICU of the Affiliated Hospital of Qingdao University from January 2013 to August 2018;who first stayed in SICU after hepatobiliary-pancreatic surgery;who were assessed and discharged from SICU by surgeons and SICU physicians after treatment, and then transferred to SICU again because of the change of their condition were enrolled. The unintended return to SICU within 3 days and 7 days were recorded. Patients who returned to SICU within 7 days were studied for the pathogenies, risk factors and predictive model of returning to SICU, and non-returning patients were enrolled according to 1∶1 as the controls. A total of 43 indicators were divided into five categories, including general clinical data, medical history, surgical indicators before first admission of SICU, length of first SICU stay, and other indicators on the day of first discharge from the SICU. Logistic regression was used to screen the risk factors associated with SICU readmission, then the Nomogram diagram was drawn by using the R 3.4.1 software for predicting SICU readmission, and the classification performance of Nomogram was evaluated by self-help sampling test. Results Of the 763 patients discharged from the SICU, 2.10%(16/763) of them were readmitted within 3 days and 3.28%(25/763) were readmitted within 7 days to the SICU unexpectedly. The pathogenies of SICU readmission within 7 days included infection [56.00%(14/25)], heart failure [16.00%(4/25)], infarction [12.00%(3/25)], bleeding [12.00%(3/25)], and sutures splitting [4.00%(1/25)]. The pathogenies of SICU readmission within 3 days included infection [56.25%(9/16)], heart failure [18.75%(3/16)], infarction [12.50%(2/16)], and bleeding [12.50%(2/16)]. Nomogram analysis showed that the risk factors associated with unplanned SICU readmission were length of first SICU stay, history of hypertension, and activity of daily living (ADL) score, white blood cell count (WBC), arterial partial pressure of oxygen (PaO2), prothrombin time (PT), fibrinogen (FIB) on the day of first SICU discharge. Self-help sampling test was carried out on the Nomogram map, and the results showed that the coherence index (C-index) was 0.962 [95% confidence interval (95%CI)= 0.869-1.057]. The classification performance of the model was good. Conclusions The common pathogenies of SICU readmission for patients who underwent hepatobiliary-pancreatic surgery were infection, heart failure, infarction and bleeding. Risk factors of readmission after SICU discharge included the length of first SICU stay, history of hypertension, and ADL score, WBC, PaO2, PT, FIB on the day of first SICU discharge. The model consisted of above risk factors showed a good performance in predicting the probability of readmission after SICU discharge for patients who underwent hepatobiliary-pancreatic surgery.