危险性急性上消化道出血患者不良预后的风险预警模型构建及验证

Construction and verification of risk warning model for poor prognosis in patients with dangerous acute upper gastrointestinal bleeding

目的构建并验证危险性急性上消化道出血(AUGIB)患者不良预后的风险预警模型。方法回顾性分析2020年1月至2023年12月安徽医科大学附属安庆第一人民医院收治的危险性AUGIB患者110例的临床资料,按预后不同分为两组,良好预后组(n=86)与不良预后组(n=24)。采集患者临床资料(性别、年龄、基础病史)、临床表现及治疗、实验室检查等信息;采用单因素与多因素Logistic回归分析危险性AUGIB患者不良预后的危险因素;采用受试者操作特征(ROC)曲线验证模型效能;构建联合预测模型,绘制模型的列线图。结果与良好预后组相比,不良预后组年龄[(65.01±3.29)岁vs.(62.47±4.94)岁]更高,呕血(62.50%vs.34.88%)、消化道肿瘤(54.17%vs.18.60%)、肝硬化(50.00%vs.17.44%)、输注新鲜冰冻血浆(45.83%vs.16.28%)占比更高,入院血红蛋白[(62.55±11.39)g/L vs.(70.10±13.78)g/L]更低,差异均有统计学意义(P<0.05)。多因素Logistic回归分析结果显示,消化道肿瘤、肝硬化、输注新鲜冰冻血浆为危险性AUGIB患者发生不良预后的危险因素(P<0.05)。ROC曲线显示,联合预测模型的曲线下面积(AUC)、敏感度、特异度分别为0.858、87.50%、81.40%。利用Nomogram量化3个独立危险因素,校准曲线和决策曲线表明预测效能较好。结论危险性AUGIB患者不良预后的危险因素有消化道肿瘤、肝硬化、输注新鲜冰冻血浆,以上述因素为基础构建的列线图风险预警模型对危险性AUGIB患者不良预后具有较高预测效能及临床应用性,对危险性AUGIB患者不良预后的识别具有一定的潜在意义。

Objective To establish and verify a risk warning model for poor prognosis in patients with dangerous acute upper gastrointestinal bleeding(AUGIB).Methods The clinical data of 110 dangerous AUGIB patients admitted to Anqing First People's Hospital,Anhui Medical University from January 2020 to December 2023 were retrospectively analyzed and divided into two groups according to the different prognosis:the good prognosis group(n=86)and the poor prognosis group(n=24).The clinical data(gender,age,basic medical history),clinical manifestations and treatment,laboratory examination and other information were collected.The risk factors of poor prognosis in patients with dangerous AUGIB were analyzed by univariate and multivariate Logistic regression analysis.The effectiveness of the model was verified by receiver operating characteristic(ROC)curve.The combined prediction model was constructed,and the model's column-graph was drawn.Results Compared with the good prognosis group,the age of the poor prognosis group[(65.01±3.29)years vs.(62.47±4.94)years]was higher,proportions of hematemesis(62.50%vs.34.88%),digestive tract tumor(54.17%vs.18.60%),cirrhosis(50.00%vs.17.44%),infusion of fresh frozen plasma(45.83%vs.16.28%)were higher,on admission,hemoglobin[(62.55±11.39)g/L vs.(70.10±13.78)g/L]was lower,the differences were statistically significant(P<0.05).Logistic regression analysis showed that digestive tract tumor,liver cirrhosis and fresh frozen plasma infusion were risk factors for poor prognosis in AUGIB patients(P<0.05).ROC curves showed that the area under curve(AUC),sensitivity,and specificity of the combined prediction model were 0.858,87.50%,and 81.40%.Nomogram quantified three independent risk factors.Calibration curves and decision curves showed that the prediction efficiency was good.Conclusion The risk factors for poor prognosis in patients with risky AUGIB include digestive tract tumor,liver cirrhosis and infusion of fresh frozen plasma.Based on the above factors,the column-graph risk warning model has high predictive efficacy and clinical applicability for poor prognosis in patients with risky AUGIB,and has certain potential significance for the identification of poor prognosis in patients with risky AUGIB.