Analysis of influencing factors of dental fluorosis in children based on logistic regression and classification tree model
10.3760/cma.j.cn231583-20220302-00057
- VernacularTitle:基于logistic回归和分类树模型的儿童氟斑牙患病影响因素分析
- Author:
Wenfeng LI
1
;
Yang WANG
;
Fang LI
;
Yushan CUI
;
Dandan ZHANG
;
Yani DUAN
Author Information
1. 天津市疾病预防控制中心环境健康与学校卫生研究所，天津 300011
- Keywords:
Fluorosis, dental;
Child;
Influencing factors;
Classification tree model
- From:
Chinese Journal of Endemiology
2023;42(2):127-133
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To analyze the influencing factors of dental fluorosis of children in the drinking-water-borne endemic fluorosis (referred to as drinking-water-borne fluorosis) areas with qualified drinking water.Methods:In 2020 and 2021, the cluster sampling method was used to select the children aged 8 to 12 years old from the drinking-water-borne fluorisis areas with qualified drinking water in Tianjin City for water and urine fluoride detection, dental fluorosis examination and questionnaire survey, and logistic regression and classification tree model were used to analyze the influencing factors of dental fluorosis in children.Results:A total of 3 795 cases children aged 8 to 12 years old were investigated, and 1 001 cases of dental fluorosis were detected, and the detection rate of dental fluorosis was 26.38% (1 001/3 795). The results of logistic analysis showed that age [odds ratio ( OR) = 1.193, 95% confidence interval ( CI): 1.115 - 1.277], high urinary fluoride (1.84 - 19.40 mg/L, OR = 1.510, 95% CI: 1.169 - 1.952) and the number of permanent residents at home ≥6 ( OR = 1.377, 95% CI: 1.090 - 1.739) were risk factors of dental fluorosis in children; and the mother's with higher education level (college degree or above, OR = 0.664, 95% CI: 0.441 - 0.999), the years of water improvement ≥5 years (5 - < 10 years, OR = 0.193, 95% CI: 0.157 - 0.238; ≥10 years, OR = 0.254, 95% CI: 0.193 - 0.333) were protective factors of dental fluorosis in children. The results of classification tree model analysis showed that the years of water improvement contributed the most to the prevalence of dental fluorosis among children in the drinking-water-borne fluorisis areas with qualified drinking water, followed by age, number of permanent residents at home and urinary fluoride. The area under the receiver operating characteristic curve (AUC) of logistic regression model and classification tree model were 0.730 (95% CI: 0.711 - 0.748) and 0.721 (95% CI: 0.702 - 0.739), respectively, with good fitting effect. Conclusion:The detection rate of children's dental fluorosis in the drinking-water-borne fluorosis areas with qualified drinking water is mainly related to the years of water improvement, age, the number of permanent residents at home and urinary fluoride.
