Construction and evaluation of oral infection risk warning model for patients with acute leukemia undergoing chemotherapy
10.3760/cma.j.cn211501-20240506-01115
- VernacularTitle:急性白血病化疗患者口腔感染风险预测模型的构建与评价
- Author:
Jie ZHANG
1
;
Qin WANG
;
Zongjing HU
;
Yue SUN
;
Qianqian ZHANG
;
Yueshen MA
;
Wenjun XIE
Author Information
1. 中国医学科学院血液病医院（中国医学科学院血液学研究所）血液与健康全国重点实验室 国家血液系统疾病临床医学研究中心 细胞生态海河实验室，天津 300020
- Publication Type:Journal Article
- Keywords:
Nomograms;
Risk factors;
Acute leukemia;
Oral infection;
Logistic regression model
- From:
Chinese Journal of Practical Nursing
2025;41(1):13-19
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish an early warning model of oral infection risk in patients with acute leukemia undergoing chemotherapy and to verify its predictive efficacy, so as to provide reference for formulating strategies to prevent oral infections.Methods:A retrospective study was conducted to select 288 patients with acute leukemia undergoing chemotherapy from January 2021 to January 2023 in Hematology Hospital of Chinese Academy of Medical Sciences (Institute of Hematology, Chinese Academy of Medical Sciences) as the training set. According to whether they developed oral infection after chemotherapy, they were divided into the infected group and the non-infected group. The risk factors of oral infection in patients with acute leukemia undergoing chemotherapy were investigated, and a risk warning model was established. A total of 246 acute leukemia undergoing chemotherapy patients admitted to the same hospital from February 2023 to February 2024 were selected as the validation set to conduct external verification of the model.Results:The oral infection rate was 19.44% (56/288) in the training set. There were 21 males and 35 females in the infected group (56 cases), with 49 cases<60 years old and 7 cases ≥60 years old. There were 102 males and 130 females in the non infected group (232 cases), with 196 cases<60 years old and 36 cases ≥60 years old. Multivariate analysis showed that neutrophil count <1.5×10 9/L, nutritional risk screening 2002≥3 points, high-dose of methotrexate, antibiotic types ≥3, poor oral self-cleaning habits, oral pH ≤6.5 were the risk factors for oral infection in patients with acute leukemia undergoing chemotherapy ( OR values were 2.716-10.074, all P<0.05). Based on this, the risk early warning model was as follows: Logit ( P)=-5.849+2.310× neutrophil count <1.5×10 9/L+1.363× nutritional risk screening 2002≥3 points +1.150× high-dose methotrexate +1.132× antibiotic types ≥3 + 1.044× oral pH ≤6.5 + 0.999× poor oral self-cleaning habits. The area under receiver operator characteristics curves (ROC) curve of this model Logit ( P) was 0.892, the maximum approximate entry index was 0.653, the sensitivity was 0.804, and the specificity was 0.849. Hosmer-Lemeshow test results indicated that χ2=4.91, P=0.768. For external validation, the goodness of fit test results were χ2=6.47, P=0.595. The area under ROC curve was 0.884, the sensitivity was 0.832, and the specificity was 0.825. Conclusions:The established early warning model of oral infection risk in patients with acute leukemia undergoing chemotherapy has good predictive value, which is helpful for medical staff to conduct early risk assessment of oral infection in such patients, and formulate countermeasures to reduce the incidence and improve the treatment effect of the disease.
