Postoperative infection outbreak (PIO) is a concentrated manifestation of the harmfulness of hospital-acquired infections, and how to effectively control the occurrence of PIO promptly is currently the focus of infection control. At present, some infection prevention personnel lack emergency response experience and ability to investigate suspected sources of infection in the face of infection clusters.

To develop an investigation form for PIO, and identify the sources of infection in the early stage to prevent outbreaks.

From January to May 2023, Worldwide Database for Nosocomial Outbreaks (WDNO), PubMed, CNKI, and other databases were searched for studies related to PIO. Data extraction and annotation were performed. After cleaning the title and abstract text information of infection control literature downloaded from the PubMed database, a large-scale training set was constructed to build a word vector model based on the Skip-Gram method. Preliminary identification of risk factors for PIO was summarized based on the classification results of the word vector model constructed. Subsequently, 15 consulting experts were invited to evaluate the importance of PIO risk factors using a modified Delphi method. The Analytic Hierarchy Process (AHP) was adopted to calculate weight and combined weight for the primary and secondary indicators to determine the possible sources of infection outbreaks. Based on this, an investigation form for PIO was constructed.

A total of 203 PIO-related studies were finally included. A total of 15 experts were consulted by the modified Delphi method, and all the questionnaires for the two rounds of expert consultation were returned with a positive coefficient of 100.00%. The average authoritative coefficient (Cr) of 15 experts was (0.86±0.10), the average familiarity (Cs) was (0.81±0.10), and the average judgment (Ca) was (0.92±0.10). Kendall's W coefficient (Kendall's W) of concordance after two rounds of expert consultation was 0.351 (P<0.005), which showed that the consensus of expert opinion was good, indicating a good consensus among expert opinions. After two rounds of the modified Delphi method, 4 primary indicators and 19 secondary indicators were constructed. Within the indicator system, the weights of the 7 primary indicators ranged from 22.23% to 27.50%, and the combined weights of the secondary indicators ranged from 3.32% to 6.29%. Based on this, the PIO investigation questionnaire was developed, primarily including five aspects of patient basic information, infection status, perioperative factors, personnel, and environment.

Based on epidemiological characteristics and expert consultation, a form used for PIO investigation was constructed. The content covers the main risk factors and critical points that may lead to outbreaks, providing a reference for identifying potential sources of PIO.

术后感染暴发（PIO）是医院感染危害性的集中体现，如何及时、有效控制PIO的发生是当前医院感染管理的重点。目前，部分感控人员缺乏应急处置经验，在面对感染聚集事件时缺乏调查可疑感染来源的能力。

建立PIO个案调查表单，以期在感染聚集的早期探寻感染来源，防止感染暴发。

于2023年1—5月，选取全球医院感染暴发数据库（WDNO）、PubMed、中国知网（CNKI）等中、英文数据库检索，并纳入PIO相关研究后，进行数据提取与标注，在PubMed数据库中下载感染控制相关文献的标题、摘要文本信息，清洗后归集，并基于Skip-Gram生成大规模训练集构建词向量模型，根据资料提取人工标注后词向量模型构建的分类结果，初步总结PIO危险因素。随后邀请15名咨询专家采用改良德尔菲法评价PIO危险因素的重要性，运用层次分析法计算指标体系中各一级指标和二级指标的权重及组合权重来确定感染暴发的可能来源，基于此构建PIO个案调查表单。

最终纳入共203篇PIO相关研究。改良德尔菲法共计咨询15名专家，两轮专家咨询问卷全部回收，专家积极系数均为100.00%；15名专家咨询2轮的平均权威系数（Cr）为（0.86±0.10），平均熟悉程度（Cs）为（0.81±0.10），平均判断系数（Ca）为（0.92±0.10）；专家咨询两轮后的肯德尔和谐系数（Kendall's W）为0.351（P<0.005），显示专家意见共识性较好。经两轮改良德尔菲法修订，构建4个一级指标和19个二级指标。指标体系中，7个一级指标的权重为22.23%~27.50%，二级指标的组合权重为3.32%~6.29%，基于此构建PIO个案调查表单，主要包含患者基本信息、感染情况、围术期相关因素、人员与环境5大方面。

本研究基于流行特征与专家调研，构建了可用于手术相关感染暴发的早期调查表单，内容涵盖可能引起手术相关感染的主要危险因素与重点环节，为识别感染暴发潜在来源提供参考。