头颈癌切除游离皮瓣修复患者术后谵妄预测模型的建立与验证

Development and validation of a postoperative delirium prediction model for patients with head and neck can⁃cer resection and free flap repair

目的建立并验证头颈部癌(head and neck cancer,HNC)患者游离皮瓣修复术后谵妄(postoperative delirium,POD)的风险预测模型,为临床提供参考。方法本研究已通过单位伦理委员会审查批准,并获得患者知情同意。对2016年1月1日至2022年1月1日在徐州市中心医院头颈肿瘤外科接受头颈癌切除后游离皮瓣重建术患者的相关危险因素和生命体征进行回顾性收集和评估,共纳入241例,其中171例进行预测模型的建立,另外收集70例进行内部验证,采用单因素和多因素Logistic分析,R Studio软件包进行建模统计分析。结果本研究模型最终包含4个风险因素:年龄、输血、术后睡眠障碍、术后疼痛VAS得分;该模型训练集的受试者工作特征曲线下面积(AUC)为0.869(95%CI:0.789~0.948),Youden指数为0.692,预测值为0.215,敏感性为85.3%,特异性为83.9%,Hosmer-Lemeshow拟合优度为10.336(P=0.242),该模型拟合较好;根据模型对验证集进行验证,C指数为0.827(95%CI:0.681~0.973),模型预测效果较好。结论该模型可应用于头颈部恶性肿瘤切除后游离皮瓣术后谵妄的预测,对入院时HNC-POD的风险具有很高的预测价值,使用该模型可能有助于更好地实施预防性治疗和护理措施。

Objective To establish and validate a risk prediction model for postoperative delirium(POD)in patients with head and neck cancer(HNC)resection with free flap repair and to provide a reference for clinical practice.Methods This study has been reviewed and approved by the Ethics Committee,and informed consent has been obtained from patients.This study retrospectively collected and evaluated the risk factors and vital signs of patients undergoing head and neck cancer free flap reconstruction in the Department of Oromaxillofacial Head and Neck Oncology,Xuzhou Central Hospital from January 1,2016,to January 1,2022.A total of 241 cases were included,of which 171 cases were used to establish the prediction model,and 70 cases were collected for internal verification.Univariate and multivariate logistic analyses and the R Studio software package were used for modeling and statistical analysis.Results The research model finally included five risk factors:age,blood transfusion,postoperative sleep disorder and postoperative VAS pain value.The area under the working characteristic curve(AUC)of the subjects in the training set of the model was 0.869(95%CI:0.789-0.948),the Youden index was 0.692,the predictive value was 0.215,the sensitivity was 85.3%,the specificity was 83.9%,and the goodness of fit of the Hosmer Lemeshow test was 10.336(P=0.242).The model fit well.The validation set was verified according to the model.The C index was 0.827(95%CI:0.681-0.973),and the model prediction effect was very good.Conclusion This model may be applied to predict post-operative delirium for patients with head and neck cancer resection and free flap repair,which has a high predictive value for the risk of HNC-POD at admission.The use of this model may help to better implement preventive treatment and nursing measures.