国际妇产科学杂志 ›› 2023, Vol. 50 ›› Issue (4): 421-427.doi: 10.12280/gjfckx.20220806

• 产科生理及产科疾病:论著 • 上一篇    下一篇

剖宫产患者术后感染列线图预测模型的构建及评价

张燕飞, 王艺璇, 张韶华, 褚莹, 王赞宏()   

  1. 030032 太原,山西白求恩医院妇产科
  • 收稿日期:2022-10-05 出版日期:2023-08-15 发布日期:2023-08-15
  • 通讯作者: 王赞宏,E-mail:wangzanhong@126.com

Construction and Validation of the Nomogram Prediction Model for Postoperative Infection after Cesarean Section

ZHANG Yan-fei, WANG Yi-xuan, ZHANG Shao-hua, CHU Ying, WANG Zan-hong()   

  1. Department of Obstetrics and Gynecology, Shanxi Bethune Hospital, Taiyuan 030032, China
  • Received:2022-10-05 Published:2023-08-15 Online:2023-08-15
  • Contact: WANG Zan-hong, E-mail: wangzanhong@126.com

摘要:

目的:探讨剖宫产患者术后感染的影响因素,构建其列线图预测模型。方法:回顾性分析2021年1—12月于山西白求恩医院行剖宫产手术患者的临床资料,根据术后是否发生感染分为感染组(n=60)和未感染组(n=1 086),筛选剖宫产患者术后感染的影响因素纳入列线图风险预测模型,评价列线图预测模型的区分度、校准度及预测效能。结果:多因素Logistic回归分析结果显示,有稳定的工作是剖宫产患者术后感染的保护因素(OR=0.570,95%CI:0.331~0.983,P=0.043),妊娠期高血压疾病(OR=2.356,95%CI:1.324~4.192,P=0.004)、B族链球菌(group B Streptococcus,GBS)定植(OR=3.154,95%CI:1.118~8.897,P=0.030)、阴道检查次数>5次(OR=2.470,95%CI:1.146~5.324,P=0.021)以及缩宫素引产(OR=2.457,95%CI:1.230~4.907,P=0.011)是剖宫产患者术后感染的危险因素。基于以上因素建立列线图预测模型,一致性指数(consistency index,C-index)为0.721(95%CI:0.651~0.791),以0.051为截断值,敏感度为66.7%,特异度为72.5%。Homer-Lemeshow拟合优度检验提示预测模型有较好的校准能力(χ2=2.169,P=0.825)。结论:基于工作、妊娠期高血压疾病、GBS定植、阴道检查次数及缩宫素引产构建的列线图预测模型可根据预测风险值早期识别剖宫产术后感染高危人群,采取对应的预防措施,减少感染发生。但仍需进一步行外部验证及前瞻性比较试验证实模型预测能力的可靠性。

关键词: 剖宫产术, 手术后并发症, 感染, 影响因素分析, 列线图, 预测

Abstract:

Objective: To investigate the influencing factors of postoperative infection in cesarean section patients and construct a nomogram prediction model. Methods: The clinical data of patients undergoing cesarean section in Shanxi Bethune Hospital from January 2021 to December 2021 were retroanalyzed and devided into infection group (n=60) and no infection group (n=1 086). The influencing factors of postoperative infection in cesarean section patients were selected into the nomogram risk prediction model. The discrimination, calibration, and prediction efficacy of the nomograms prediction model were evaluated. Results: The results of the multivariate Logistic regression analysis showed that having stable work was a protective factor for postoperative infection in cesarean section patients (OR=0.570, 95%CI: 0.331-0.983, P=0.043). Hypertensive disorders in pregnancy (OR=2.356, 95%CI: 1.324-4.192, P=0.004), GBS colonization (OR=3.154, 95%CI: 1.118-8.897, P=0.030), times of vaginal examinations >5 (OR=2.470, 95%CI: 1.146-5.324, P=0.021) and oxytocin induction (OR=2.457, 95%CI: 1.230-4.907, P=0.011) were independent risk factors of postoperative infection in cesarean section patients. The nomogram prediction model was established based on the above factors, whose consistency index (C-index) was 0.721 (95%CI: 0.651-0.791), and 0.051 was selected as the cut-off value, with a sensitivity of 66.7% and a specificity of 72.5%. The Homer-Lemeshow goodness-of-fit test suggested that the prediction model had a better calibration ability ( χ2=2.169, P=0.825). Conclusions: The nomogram prediction model constructed based on work, gestational hypertensive diseases, GBS colonization, times of vaginal examinations, and oxytocin labor induction has good accuracy and differentiation. It can identify the high-risk group after cesarean section to be infected in the early stage according to the predicted risk value, and corresponding preventive measures should be taken to reduce the occurrence of infection. However, further external verification and prospective comparative trials are needed to confirm the reliability of the predictive ability of the model.

Key words: Cesarean section, Postoperative complications, Infections, Root cause analysis, Nomograms, Forecasting