摘要： 目的 筛查急性脑出血术后发生应激性溃疡的危险因素，并基于该危险因素构建风险预测列线图（Nomogram）模型。方法 纳入2015年1月至2020年11月中国科学技术大学附属第一医院收治的300例急性脑出血术后患者，根据是否发生应激性溃疡分为溃疡组（53例）和无溃疡组（247例），单因素和多因素逐步法Logistic回归分析筛查急性脑出血术后发生应激性溃疡的危险因素，并基于危险因素构建Nomogram模型，绘制该模型的受试者工作特征（ROC）曲线和校准曲线并行Hosmer-Lemeshow拟合优度检验。结果 Logistic回归分析显示，年龄偏大（OR=1.043，95% CI：1.059~1.086；P=0.042）、Glasgow昏迷量表（GCS）评分≤ 8分（OR=2.809，95% CI：1.018~7.740；P=0.044）、出血量≥ 30 ml（OR=3.830，95% CI：1.342~10.933；P=0.012）、中性粒细胞/淋巴细胞比值增加（OR=1.090，95% CI：1.032~1.151；P=0.002）、术后收缩压升高（OR=1.031，95% CI：1.008~1.055；P=0.008）和并发颅内感染（OR=2.809，95% CI：1.006~7.847；P=0.049）是脑出血术后发生应激性溃疡的危险因素。根据这6项危险因素建立Nomogram模型，ROC曲线下面积为0.881（95% CI：0.821~0.942，P=0.001），该模型预测急性脑出血术后发生应激性溃疡的最佳截断值为137；校准曲线显示预测概率与实际概率之间具有良好的一致性，Hosmer-Lemeshow拟合优度检验显示差异无统计学意义（χ²=7.891，P=0.445），表明Nomogram模型具有良好的区分度、校准度和稳定性。结论 年龄偏大、GCS评分≤ 8分、出血量≥ 30 ml、中性粒细胞/淋巴细胞比值增加、术后收缩压升高和并发颅内感染的急性脑出血患者术后易发生应激性溃疡。据此构建的Nomogram模型可直观、简便地为急性脑出血患者提供个体化的术后应激性溃疡风险预测。

关键词: 脑出血, 溃疡, Logistic模型, 列线图

Abstract: Objective To analysis of risk factors for stress ulcers after the surgery of acute cerebral hemorrhage, and use these factors to construct a Nomogram model. Methods In this study, 300 patients with acute cerebral hemorrhage treated in The First Affiliated Hospital of University of Science and Technology of China between January 2015 and November 2020 were recruited and divided into stress ulcers group (53 patients) and non-ulcers group (247 patients) according to whether the postoperative period was complicated by stress ulcers. Univariate and multivariate stepwise Logistic regression were used to assess the risk factors for stress ulcers after the surgery of acute cerebral hemorrhage and the Nomogram model was established. The receiver operating characteristic (ROC) curve and calibration curve of the model were plotted, and Hosmer-Lemeshow goodness of fit test was performed. Results Logistic regression showed that age increased (OR=1.043, 95%CI:1.059-1.086; P=0.042), Glasgow Coma Scale (GCS) score ≤ 8 (OR=2.809, 95%CI:1.018-7.740; P=0.044), the cerebral hemorrhage volume ≥ 30 ml (OR=3.830, 95%CI:1.342-10.933; P=0.012), neutrophil to lymphocyte ratio (NLR) increased (OR=1.090, 95%CI:1.032-1.151; P=0.002), the increase of postoperative systolic blood pressure (OR=1.031, 95%CI:1.008-1.055; P=0.008) and postoperative intracranial infection (OR=2.809, 95%CI:1.006-7.847; P=0.049) were the risk factors for stress ulcers after the surgery of acute cerebral hemorrhage. The Nomogram model was established based on the 6 risk factors, and the area under curve (AUC) of ROC was 0.881 (95%CI:0.821-0.942, P=0.001); the best cut-off value for predicting the risk of acute postoperative cerebral hemorrhage complicated by stress ulcers was 137. The calibration curves showed good agreement between the predicted and actual probabilities, and the Hosmer-Lemeshow goodness of fit test showed there was no statistical difference (χ²=7.891, P=0.445), indicating that the Nomogram model has good discrimination, calibration and stability. Conclusions Over-age, GCS score ≤ 8, cerebral hemorrhage volume ≥ 30 ml, NLR increase, postoperative systolic blood pressure increase and intracranial infection were the risk factors for stress ulcers after the surgery of acute cerebral hemorrhage. In conclusion, this Nomogram model is able to individually, visually and briefly predict the risk of postoperative complications of stress ulcers in patients after the surgery of acute cerebral hemorrhage.

Key words: Cerebral hemorrhage, Ulcer, Logistic models, Nomograms