Abstract:

Objective To develop a risk prediction model based on the influencing factors of chemotherapy-induced oral mucositis in patients with hematologic tumors，and to verify the effect. Methods A case-control study was conducted to retrospectively collect the data of 444 adult patients with hematological malignancies who underwent chemotherapy in the hematology department of a tertiary hospital in Wuhan from March 2021 to January 2022. Univariate analysis，collinearity diagnosis and binary logistic regression analysis were used to explore influencing factors for chemotherapy-induced oral mucositis in patients with hematologic tumors，and then a prediction model which was presented with nomogram was constructed. Receiver operating characteristic，Hosmer Lemeshow and calibration charts were used to verify the predictive effect of the model，and Bootstrap resampling method was used for internal verification of the prediction model. The clinical data of 171 patients with hematologic malignancies who underwent chemotherapy in the hematology department of 2 tertiary hospitals in Wuhan from February to October 2022 were prospectively collected for external verification. Results History of oral lesions，treatment methods，high-dose chemotherapy，vomiting，combined chemo-therapy，platelet count，and hemoglobin were independent predictive factors of chemotherapy-induced oral mucositis in patients with blood tumors. The area under the curve of the model was 0.822，and Hosmer-Lemeshow test P=0.602. After internal verification and external verification，the area under the curve of the model was 0.813 and 0.735，respectively. Hosmer-Lemeshow test were 0.115 and 0.820，respectively. These calibration charts showed that the observed results of the model were consistent with the predicted results. Conclusion The risk prediction model of chemotherapy-induced oral mucositis in hematologic tumor patients has good predictive performance and extrapolation，which is an effective screening tool to help medical staff identify the high-risk patients with chemotherapy-induced oral mucositis early.

Key words: Hematologic Neoplasms, Chemotherapy-Induced Oral Mucositis, Prediction Model, Nursing Care

摘要：

目的 探讨血液系统恶性肿瘤患者化疗相关性口腔黏膜炎发生的影响因素，构建风险预测模型并验证其效果。方法 采用病例对照研究方法，回顾性收集2021年3月—2022年1月在武汉市某三级甲等医院血液内科化疗的444例血液系统恶性肿瘤患者的资料，采用单因素分析、多重线性分析和Logistic回归分析明确血液系统恶性肿瘤患者化疗相关性口腔黏膜炎发生的影响因素，建立风险预测模型并绘制列线图。采用受试者操作特征曲线下面积、Hosmer-Lemeshow检验和校准图评估模型的预测性能。采用Bootstrap重复抽样法对模型进行内部验证。前瞻性收集2022年2月—10月于武汉市2所三级甲等医院血液内科化疗的171例血液系统恶性肿瘤患者的资料，对模型进行外部验证。结果 Logistic回归分析结果显示，有无口腔疾病史、治疗方法、是否采用大剂量化疗、化疗期间是否发生呕吐、是否采用联合化疗、血小板计数和血红蛋白浓度是血液系统恶性肿瘤患者化疗相关性口腔黏膜炎发生的影响因素。模型的受试者操作特征曲线下面积为0.822，Hosmer-Lemeshow检验P值为0.602。内部验证和外部验证的受试者操作特征曲线下面积分别为0.813、0.735，Hosmer-Lemeshow检验P值分别为0.115、0.820，校准图中模型的预测曲线和理想曲线重合度均较高。结论 该研究构建的血液系统恶性肿瘤患者化疗相关性口腔黏膜炎风险预测模型具有较强的预测能力，可作为化疗相关性口腔黏膜炎高风险患者的筛查工具。

关键词: 血液肿瘤, 化疗相关性口腔黏膜炎, 预测模型, 护理