Abstract: Objective  To analyze the influencing factors for in-hospital gastrointestinal hemorrhage in patients with aneurysmal subarachnoid hemorrhage (aSAH) and to construct a nomogram prediction model. 
Methods  Consecutive patients with aSAH admitted to the Department of Intensive Care Unit, Beijing Tiantan Hospital, Capital Medical University, between January 2021 and January 2023 were retrospectively enrolled. Demographic data, previous medical history, assessment data at intensive care unit admission, complications, and treatment-related data were collected. Patients were divided into a gastrointestinal bleeding group and a non-gastrointestinal bleeding group according to the occurrence of in-hospital gastrointestinal bleeding. Baseline data were compared between the two groups. Least absolute shrinkage and selection operator regression was used for variable selection. Binary logistic regression was used to identify independent risk factors for in-hospital gastrointestinal bleeding, and a nomogram prediction model was constructed. The bootstrap method (1000 resamples) was used for internal validation. The predictive performance of the model was evaluated using the AUC of ROC, the Hosmer-Lemeshow goodness-of-fit test, calibration curve and decision curve analysis. 
Results  A total of 314 patients with aSAH were enrolled, including 123 males and 191 females, with a median age of 57 (47-65) years. Among them, 62 patients (19.7%) were in the gastrointestinal bleeding group, and 252 patients (80.3%) were in the non-gastrointestinal bleeding group. History of peptic ulcer, Hunt-Hess grade Ⅲ-Ⅴ, delayed cerebral ischemia, pulmonary infection, hypochloremia, and diarrhea were identified as independent risk factors for in-hospital gastrointestinal bleeding in patients with aSAH. The prediction model was logit (P)=-3.318+1.818×history of peptic ulcer+0.973×Hunt-Hess grade Ⅲ-Ⅴ+0.969×delayed cerebral ischemia+1.362×pulmonary infection+
0.889×hypochloremia+1.152×diarrhea. The AUC of the model was 0.794 (95%CI 0.736-0.853). The Hosmer-Lemeshow goodness-of-fit test yielded a χ2 of 2.849 (P=0.723). The maximum Youden index was 0.481, with an optimal cutoff value of 0.244, at which the sensitivity was 77.42% and the specificity was 70.63%. The calibration curve showed that the model curve was in good agreement with the ideal calibration curve. Decision curve analysis indicated that the model provided a high net benefit when the risk threshold ranged from 0.04 to 0.87. 
Conclusions  History of peptic ulcer, Hunt-Hess grade Ⅲ-Ⅴ, delayed cerebral ischemia, pulmonary infection, hypochloremia, and diarrhea are independent risk factors for in-hospital gastrointestinal bleeding in patients with aSAH. The nomogram prediction model constructed in this study demonstrates good discrimination, calibration, and clinical utility, and may serve as a reference for risk screening and prevention of in-hospital gastrointestinal bleeding in patients with aSAH.

Key words: Aneurysmal subarachnoid hemorrhage; In-hospital gastrointestinal bleeding; Nomogram; Nursing care

摘要： 目的 分析动脉瘤性蛛网膜下腔出血（aneurysmal subarachnoid hemorrhage，aSAH）患者院内消化道出血的影响因素，并构建列线图预测模型。 
方法 回顾性连续纳入2021年1月—2023年1月收治于首都医科大学附属北京天坛医院重症监护室的aSAH患者。收集患者的人口学资料及既往病史、入重症监护室时评估资料、并发症及治疗相关资料。根据是否发生院内消化道出血，将患者分为消化道出血组和无消化道出血组，比较两组患者的基线资料差异。采用最小绝对收缩和选择算子回归进行变量筛选，采用二元logistic回归确定院内消化道出血的独立危险因素，并构建列线图预测模型。采用Bootstrap法（重复抽样1000次）对模型进行内部验证，采用ROC的AUC、Hosmer-Lemeshow拟合优度检验、校准曲线及决策曲线分析评估模型的预测效能。
结果 共纳入314例aSAH患者，男性123例，女性191例，中位年龄为57（47～65）岁。其中，消化道出血组62例（19.7%），无消化道出血组252例（80.3%）。消化道溃疡病史、Hunt-Hess分级为Ⅲ～Ⅴ级、迟发性脑缺血、肺部感染、低氯血症、腹泻是aSAH患者院内消化道出血的独立危险因素。构建的预测模型为logit（P）=-3.318+1.818×消化道溃疡病史+0.973×Hunt-Hess分级为Ⅲ～Ⅴ级+0.969×迟发性脑缺血+1.362×肺部感染+0.889×低氯血症+1.152×腹泻。该模型的AUC值为0.794（95%CI 0.736～0.853），Hosmer-Lemeshow拟合优度检验结果显示χ2为2.849（P=0.723），最大约登指数为0.481，最佳截断值为0.244，此时模型的敏感度为77.42%，特异度为70.63%。校准曲线显示，模型曲线与校准曲线基本吻合。决策曲线分析显示，当风险阈值在0.04～0.87时，模型的净获益较高。
结论 消化道溃疡病史、Hunt-Hess分级为Ⅲ～Ⅴ级、迟发性脑缺血、肺部感染、低氯血症及腹泻是aSAH患者发生院内消化道出血的独立危险因素。本研究构建的列线图预测模型具有良好的区分度、校准度及临床实用性，可为aSAH患者院内消化道出血的风险筛查及预防提供参考。

关键词: 动脉瘤性蛛网膜下腔出血; 院内消化道出血; 列线图; 护理