广西医学

目的探讨5~10 mm结直肠无蒂息肉切除术后迟发性出血的影响因素并构建列线图预测模型。方法纳入184例5~10 mm结直肠无蒂息肉患者，根据术后是否发生迟发性出血分为出血组（n=16）与未出血组（n=168）。比较两组患者的一般资料（性别、年龄、高血压病史、冠心病病史、糖尿病病史）、治疗情况（术中出血情况、抗凝药物服用史、切除方式）及息肉特征（位置、直径、病理类型）。采用多因素Logistic回归模型分析5~10 mm结直肠无蒂息肉切除术后迟发性出血的影响因素。根据影响因素构建列线图预测模型并评估模型的预测效能。结果出血组与未出血组患者的高血压病史、息肉位置、术中出血、息肉直径、切除方式比较，差异有统计学意义（P<0.05）。高血压病史、息肉位置、息肉直径、切除方式是5~10 mm结直肠无蒂息肉切除术后迟发性出血的影响因素（P<0.05）。受试者工作特征曲线分析显示，基于上述影响因素构建的列线图预测模型的曲线下面积为0.964（95% CI：0.936，0.993）。校准曲线分析显示，通过1 000次Bootstrap重抽样验证后得到平均绝对误差为0.029，该模型预测风险与实际风险高度一致。结论高血压病史、息肉位置、息肉直径及切除方式是5~10 mm结直肠无蒂息肉切除术后迟发性出血的影响因素；基于上述影响因素构建的列线图预测模型具有良好的区分能力和预测精度。

Objective To investigate the influencing factors for delayed hemorrhage after resection of 5-10 mm sessile colorectal polyps, and to construct a nomogram predictive model. Methods A total of 184 patients with 5-10 mm sessile colorectal polyps were enrolled, and they were divided into hemorrhage group (n=16) or non⁃hemorrhage group (n=168) according to the presence of delayed hemorrhage occurred after surgery. General data (sex, age, history of hypertension, history of coronary heart disease, history of diabetes mellitus), therapeutic conditions (intraoperative bleeding, history of anticoagulant use, resection method), and characteristics of polyps (location, diameter, pathological type) were compared between the two groups. Multivariate Logistic regression model was used to identify influencing factors for delayed hemorrhage after resection of 5-10 mm sessile colorectal polyps, based on which a nomogram predictive model was constructed and its predictive performance was evaluated. Results There were statistically significant differences in history of hypertension, polyp location, intraoperative bleeding, polyp diameter, and resection method between the hemorrhage group and the non⁃hemorrhage group (P<0.05). History of hypertension, polyp location, polyp diameter, and resection method were influencing factors for delayed hemorrhage after resection of 5-10 mm sessile colorectal polyps (P<0.05). Receiver operating characteristic curve analysis revealed that the area under the curve of the nomogram predictive model based on these influencing factors was 0.964 (95% CI: 0.936, 0.993). Calibration curve analysis, validated by 1000 bootstrap resampling iterations, yielded a mean absolute error of 0.029, demonstrating high consistency between predicted and actual risks of the model. Conclusion History of hypertension, polyp location, polyp diameter, and resection method are influencing factors for delayed hemorrhage after resection of 5-10 mm sessile colorectal polyps. The nomogram predictive model constructed based on these influencing factors exhibits good discriminative ability and predictive accuracy.