Supplementary Material for: Risk of Reintervention or Postoperative Bleeding after Laparoscopy for Benign Gynecological Disease: a Clinical Prediction Model

Objective: To develop a clinically applicable prediction tool to early seek for postoperative major complications after laparoscopic surgery for benign pathologies. Design: Retrospective analysis of prospectively collected data Setting: Tertiary-care University Hospital Participants: Reproductive aged women undergoing laparoscopy for benign conditions Methods: Anamnestic, intraoperative, and postoperative characteristics from January 2019 to December 2021 were retrospectively reviewed. Patients with postoperative complications (reintervention or postoperative bleeding) were matched in a 1:2 ratio with women with same surgical indications without complications. Cases and controls were matched for preoperative hemoglobin, hematocrit, weight, height, body mass index (BMI), age and blood volume. A prediction model was created by inserting multiple independent modifying factors through logistic regression. The receiver operating characteristic (ROC) curve was used to evaluate the predictive accuracy of the model, and the Hosmer–Lemeshow (H–L) test was carried out to evaluate the goodness- of-fit and a calibration curve was drawn to confirm the predictive performance. A nomogram was depicted to visualize the prediction model. Results: Thirty-nine complicated procedures were matched with 78 uncomplicated controls. According to the multivariate logistic regression analysis findings, the prediction model was developed using C reactive protein (CRP), intraoperative blood loss and 24 hours postoperative urinary volume, therefore a nomogram was generated. The area under the ROC curve (AUC) of the prediction model was 0.879, depicting good accuracy, the sensitivity was 60.00%, while specificity reached 93.59%. The H–L test (χ2 = 4.45, p= 0.931) and the calibration curve indicated a good goodness-of-fit and prediction stability. Limitations: The retrospective design, moderate sensitivity and study population limit the generalization of the findings, requiring additional research. Conclusions: This prediction model based on CRP, intraoperative blood loss and 24 hours postoperative urinary volume might be a potentially useful tool for predicting reintervention and postoperative bleeding in patients undergoing planned gynecological laparoscopy.

研究目标：开发一款可临床应用的预测工具，用于早期识别良性疾病腹腔镜手术后的严重术后并发症。
研究设计：对前瞻性收集的数据进行回顾性分析。
研究地点：三级教学医院（Tertiary-care University Hospital）。
研究对象：2019年1月至2021年12月期间因良性疾病接受腹腔镜手术的育龄女性。
研究方法：对2019年1月至2021年12月收集的病史、术中及术后相关特征进行回顾性分析。将术后出现并发症（需再次手术干预或术后出血）的患者与同手术指征且无并发症的女性按照1:2比例进行匹配。匹配因素包括术前血红蛋白、血细胞比容、体重、身高、体质量指数（Body Mass Index, BMI）、年龄及血容量。通过logistic回归（logistic regression）纳入多个独立影响因素构建预测模型。采用受试者工作特征曲线（Receiver Operating Characteristic, ROC）评估模型的预测准确度，通过Hosmer-Lemeshow（H-L）检验评价模型拟合优度，并绘制校准曲线以验证预测性能。最终构建列线图（nomogram）以可视化呈现该预测模型。
研究结果：共纳入39例出现并发症的病例，匹配78例无并发症的对照病例。多因素logistic回归分析结果显示，采用C反应蛋白（C reactive protein, CRP）、术中失血量及术后24小时尿量构建预测模型，并生成对应列线图。该预测模型的受试者工作特征曲线下面积（Area Under the ROC Curve, AUC）为0.879，提示具备良好的预测准确度；其灵敏度为60.00%，特异度达93.59%。H-L检验（χ²=4.45，P=0.931）及校准曲线结果均表明模型具备良好的拟合优度与预测稳定性。
研究局限性：本研究为回顾性设计，且灵敏度中等、研究样本量有限，因此研究结果的外推性存在一定限制，仍需开展更多相关研究加以验证。
研究结论：本研究基于CRP、术中失血量及术后24小时尿量构建的预测模型，有望成为预测计划性妇科腹腔镜手术患者需再次手术干预及术后出血的实用临床工具。