目的：异体输血是改善围术期贫血的主要手段。手术部位感染（Surgical Site Infection，SSI）是手术预后和整体护理质量的重要指标。输血相关的免疫调节作用可能使患者发生SSI风险增加。面对输血决策时，如何平衡纠正贫血的获益与输血的潜在风险是临床医生最为关注的问题。本文旨在探讨围术期输血/输红细胞对手术部位感染的影响及输红细胞剂量反应关系，并探讨自体输血可否减少手术部位感染的发生，为输血决策者提供基于真实世界大数据研究的证据。方法：本研究采用回顾性队列研究方法，数据来源为国家卫生健康委员会授权使用的医院质量监测系统（Hospital Quality Monitoring System，HQMS）数据库（包含全国31省病案首页数据）。研究对象为2013-2018年我国三级医院外科住院且行手术患者，共采用两种分组方式：1)将患者分为输注任意血液成分（暴露组）和未输注任何血液成分（对照组）两组；2)将患者分为输红细胞（暴露组）和未输红细胞（对照组）两组，分析其对手术部位感染（主要结局指标）的影响，并在输注任意血液成分/输红细胞人群中探讨自体输血是否会减少手术部位感染的发生。选取住院期间死亡、院内感染次数和住院时长为次要结局指标。此外，还将输红细胞人群按照输注剂量（≤1U、>1且≤2U、>2且≤4U、>4且≤8U、＞8U）分为5组，探讨是否与手术部位感染存在剂量反应关系。输血对结局的效应采用比值比（Odds ratio, OR）测量，使用多因素logistic回归模型对OR值及其95%置信区间进行估计。结果：本研究共纳入1896584例患者。患者平均年龄45.97±20.13岁，男性占57.4%，来自全国29个省份和地区。其中，输注任意血液成分有89629例（4.7%），输红细胞76078例（4.0%）。输任意血液成分/输红细胞组年龄显著大于未输红细胞组，两组手术种类分布不均，且任意血液成分/输红细胞组ASA分级更高，围术期存在贫血诊断的患者更多，围术期合并肿瘤的患者更多，手术时间更长，II、III级切口比例更多，全麻手术比例更高。在调整了上述混杂因素后，输任意血液成分是SSI的危险因素（OR 2.14; 95%CI [2.01-2.29], P<0.001），输红细胞是SSI的危险因素（OR 2.24; 95%CI [2.09-2.40], P<0.001）。自体血的使用在输任意血液成分（OR 0.29; 95%CI [0.18-0.46], P<0.001）和输红细胞人群（OR 0.52; 95%CI [0.30-0.88], P=0.016）中均是SSI的保护因素。按照输注红细胞剂量分组后，分别得到当剂量为0U，>0且≤1U、>1且≤2U、>2且≤4U、>4且≤8U、＞8U时，对SSI的OR值为1.20（95%CI [0.76-1.91]）、1.27（95%CI [1.10-1.47]）、1.70（95%CI [1.49-1.93]）、2.12（95%CI [1.83-2.45]）和3.65（95%CI [3.13-4.25]），存在剂量反应关系。结论：外科围术期输任意血成分可能增加术后SSI风险。外科围术期输红细胞也可增加术后SSI风险，且呈剂量反应关系，输注剂量越高，SSI发生率越高。在输任意血成分和输红细胞的人群中，自体输血可能减少切口感染。

Objective：Allogeneic blood transfusion is the main method to correct perioperative anemia. Surgical site infection (SSI) is an important marker of postoperative morbidity and overall quality of care. Transfusion-related immunomodulation may lead to weakened immunity in response to blood transfusion and predispose patients to SSIs. While facing with the decision whether to transfuse or not, how to balance the benefits of anemia and the potential risk of blood transfusion is one of the most concerned issue for clinicians. The aim of this study was to assess the impact of perioperative blood transfusions on SSIs and dose-response effect of red blood cells, as well as exploring whether autologous blood transfusion can reduce the incidence of surgical site infection and provide evidence for decision-makers based on real-world big data. Methods：Retrospective cohort study was used in the Hospital Quality Monitoring System (HQMS) database authorized by the National Health Commission of the People’s Republic of China (including home page medical records of 31 provinces). Patients who were hospitalized in surgery departments and underwent surgeries from 2013 to 2018 in China's tertiary hospitals were included. Two grouping method were used in the study：1) Patients were divided into perioperative any blood transfusion (exposed) group and no transfusion (control) group. 2) Patients were divided into perioperative RBCs transfusion (exposed) group and non-RBCs (control) group. SSIs (primary outcome) were compared between 2 groups in both grouping methods. SSI incidence were also compared between whether autologous blood transfusion were used or not in the population of any transfusion and RBCs transfusion to discuss the impact of autotransfuion on SSIs. Secondary outcomes included in-hospital death, hospital infections and length of hospital stay. In addition, RBCs transfusion population was divided into 6 dose groups (0U，>0 and≤1U, >1 and ≤2U, > 2 and ≤4U, > 4 and ≤8U, >8U) to investigate the dose-response relationship of RBCs transfusions and SSIs. Multiple logistic regression analysis was used to calculate the adjusted odds ratios (ORs) with 95% CIs to identify the independent associations between exposures and outcomes. Results：A total of 1896584 patients from 29 provinces and regions across the country were included. The mean age of patients was 45.97±20.13, and 57.4% of the population were men. Among them, 89629 cases (4.7%) were infused with any blood product perioperatively, and 76078 (4.0%) were infused with red blood cells. The average age of perioperative any blood component transfusion group and perioperative RBCs transfusion group were both significantly higher than that of the non-transfusion group. Surgery types were unevenly distributed between transfusion and non-transfusion groups, and ASA grade was higher in perioperative any blood component transfusion group and perioperative RBCs transfusion group. There were more patients diagnosed with anemia and tumors, and people who were transfused had longer operation time, more percentage of clean-contaminated and contaminated incisions, and a higher proportion of general anesthesia. After adjusting for the confounding factors, any blood component transfusion perioperatively statistically significantly increased the risk of SSIs(OR 2.14; 95% CI [2.01-2.29], P<0.001). RBCs transfusions perioperatively were also a risk factor for SSIs (OR 2.24; 95% CI [2.09-2.40], P<0.001). Autologous blood was a protective factor for SSI in the population with any blood component transfusion (OR 0.29; 95%CI [0.18-0.46], P<0.001) and the population with RBCs transfusion (OR 0.52; 95%CI [0.30-0.88], P=0.016) . After dividing the infusion dose of RBCs into 6 groups (0U，>0 and≤1U, >1 and ≤2U, > 2 and ≤4U, > 4 and ≤8U, >8U), the ORs of SSI were 1.20 (95%CI[0.76-1.91]), 1.27 ( 95%CI[1.10-1.47]) , 1.70 (95%CI[1.49-1.93]), 2.12 (95%CI[1.83-2.45]) and 3.65 (95%CI[3.13-4.25]) respectively, which showed a dose response relationship between RBCs transfusion and SSIs. Conclusion：Any transfusion in perioperative period significantly increased the risk of postoperative SSIs. Perioperative RBCs transfusion also increased the risk of SSIs. There were dose-response effect between RBCs transfusion and SSIs. The greater amount of RBCs transfused perioperatively, the higher the incidence of SSIs occurred. Autologous blood transfusions may reduce SSIs in the population of either any transfusion or RBCs transfusion.