目的: 旨在发现行根治性肿瘤切除、颈淋巴结清扫并游离皮瓣修复重建术的原发性口腔鳞状细胞癌患者术后感染的病原学特征,并构建感染风险预测模型。方法: 选取2018年1月至2020年12月在北京大学口腔医院行根治性肿瘤切除并游离皮瓣修复重建术的口腔鳞状细胞癌患者1 596例为研究对象,按照患者术后感染发生情况分为感染组(n=154)和未感染组(n=1 442), 分析感染组患者的病原菌特征。以患者是否发生术后感染作为结果变量,采用单因素和多因素Logistic回归分析来确定术后感染的相关因素,并用于构建感染风险预测模型,采用受试者工作特征(receiver operating characteristic,ROC)曲线评估模型的预测效能。结果: 1 596例口腔鳞状细胞癌患者行修复重建术后发生感染154例,感染率为9.65%,感染部位以手术切口和呼吸道为主;共培养分离病原菌268株,其中革兰阴性菌240株,占89.55%,以铜绿假单胞菌(Pseudomonas aeruginosa)和肺炎克雷伯菌(Klebsiella pneumoniae)为主;革兰阳性菌23株,占8.58%,以粪肠球菌(Enterococcus faecalis)和金黄色葡萄球菌(Staphylococcus aureus)为主;真菌5株,占1.87%。分离出的铜绿假单胞菌对亚胺培南和美罗培南的耐药率较高,对环丙沙星等比较敏感;分离出的金黄色葡萄球菌对红霉素和克林霉素耐药率较高,对万古霉素敏感。N分期≥1、美国麻醉医师协会(American Society of Anesthesiologists,ASA)分级≥Ⅱ级、气管切开、住院天数>13 d是患者术后感染的独立危险因素(P<0.05), 据此构建感染风险预测模型的表达式为:预测概率值P=1/(1+e^-a),a=-0.803+0.674×(N分期≥1)+0.518×(ASA分级≥Ⅱ级)+0.918×(气管切开)+1.581×(住院天数>13 d),Hosmer-Lemeshow χ²=10.647,P=0.223,提示模型的拟合度较好。模型预测患者术后感染的ROC曲线下面积为0.818,95%CI为0.789~0.846。结论: 口腔鳞状细胞癌修复重建术后感染发生率较高,主要致病菌为革兰阴性菌。多因素Logistic回归分析构建的感染预测模型可以有效预测口腔鳞状细胞癌修复重建术后感染的发生,使临床能够有针对性地采取监测和干预措施,合理使用抗菌药物,有利于预防和减少术后感染。

Objective: To investigate the characteristics of pathogen infection and to establish a prediction model of infections in oral squamous cell carcinoma patients undergoing surgery with free flap reconstruction. Methods: The retrospective cohort study consisted of 1 596 patients undergoing tumor resection and free flap reconstruction for oral squamous cell carcinoma from January 2018 to December 2020. According to the postoperative infection, the patients were divided into the infected group (n=154) and non-infected group (n=1 442). The characteristics of pathogens were analyzed in the infected patients. The primary outcome variable was postoperative infection, and Logistic regression was used to determine risk factors of the infection. The prediction model was established and the discriminatory accuracy of the model was evaluated using receiver operating characteristic (ROC) curve. Results: Totally 154 cases were infected in the 1 596 cases undergoing surgery with free flap reconstruction, and the infection rate was 9.65%. The most frequent sites of infection were the surgical wound and respiratory tract. A total of 268 pathogens were isolated and cultured, including 240 strains of Gram-negative bacteria, accounting for 89.55%, mainly Pseudomonas aeruginosa and Klebsiella pneumoniae; 23 strains of Gram-positive bacteria, accounting for 8.58%, mainly Enterococcus faecalis and Staphylococcus aureus; and 5 strains of fungi, accounting for 1.87%. The isolated Pseudomonas aeruginosa had high resistant rate to imipenem and meropenem, and was sensitive to antibiotics, such as ciprofloxacin. The isolated Staphylococcus aureus had high resistant rate to erythromycin and clindamycin, and was sensitive to vancomycin. According to the multivariate Logistic analysis, four independent variables were significantly associated with an increased risk of postoperative infection (P<0.05): clinical N category≥1, the American Society of Anesthesiologists (ASA) grade ≥2, tracheotomy and length of hospital stay >13 d. The prediction model was established based on these factors and the expression of the risk prediction model was as follows: predicted probability value P=1/(1+e^-a), a=-0.803+0.674×(clinical N category ≥1)+0.518×(the ASA grade ≥2)+0.918×(tracheotomy)+1.581×(length of hospital stay >13 d), Hosmer-Lemeshow χ²=10.647, P=0.223, the degree of fitting of the model was good. The area under the ROC curve was 0.818 and 95%CI of the model for predicting infection was 0.789-0.846. Conclusion: Oral squamous cell carcinoma patients undergoing surgery with free flap reconstruction are prone to have a high incidence of postoperative infection and Gram-negative bacteria are the main pathogens causing an infection. The established prediction model is of good predictive effect. Rational antimicrobial use coupled with awareness of infection control measures is paramount to reduce the incidence of postoperative infection in the oral squamous cell carcinoma patients undergoing surgery with free flap reconstruction.