Multi-Level Analysis On Oral Health And Its Influencing Factors For 12-Year-Old Children In Chongqing Based On Bayes

Objective: To analyse the character of the hiararchical structure in oral health influencing factors among 12-year-old children in Chongqing.To explore the use of multilevel bayesion model analyzing the influencing factors of oral health,and evaluate the effect..Then explain the influencing factors by individual level and background level respectly to provide reference basis for the children's oral health intervention policy.Method: Data were came from the assessment on the comprehensive intervention effect for oral health among children in 2015.According to the multi-stage stratified random sampling method,a total of 2025 12-year old school children were included.Two-level logistic regression model using bayesion statistical inference were used to identify oral health influencing factors among children at the age of 12.Then compared with 1-level bayesion logistic regression model to verify the advantages of multi-level model on the analysis of children's oral health factors.Results:?1?In the 2025 children at the age of 12,the number of urban children were 1022?55.8%?,the rural children were 1003?49.5%?.The number of boy were 1077?53.2?,while the girl were 948?46.8%?.75.70% of children were looked after by their parents.In diet survey,the number of children who never or seldom drink carbonated drinks,sugary drinks,snacks,and sweets/chocolates were respectly 1592?78.62%?,1260?62.22%?,831?41.04%?and 1320?65.19%?.And the number of children who never eat dessert or drink sweet drinks before going to bed were 781?38.57%?,and the number of ocassionly to eat or drink were 781?38.57%?.In oral health behavior survey,the number of brush your teeth twice a day and more than were at 1067,accounting for 52.69%.There were 445 children heard of fluoride toothpaste in the fluoride toothpaste survey,accounting for 22.0%.In oral health attitude survey,91.2% of children agreed with that oral health is important to their lives,while 91.9% of them agreed to prevent dental disease first on your own.In oral health check,according to the results of 12 years old children in chongqing,the first permanent molar caries rate was 23.85%,permanent teeth caries rate was 31.85%,and the whole teeth caries rate was 45.53%.?2?According to the single factor analysis,the region,the frequency of candy/chocolate intake,the frequency of dessert or sweet drinks intake before bed,if the number of brushing teeth is correct,if heard of fluoride toothpaste,approval rates of oral health is very important to own life,approval rates of prevention of dental disease first on your own,the risk of caries rate was significantly different among the seven variables such as his approval ratings.?2? =187.260,9.091,8.647,20.626,4.467,15.467?12.730,P<0.05??3?Two-level bayesian empty model suggested high levels of effect is not obvious,but according to professional judgment,many aspects of children's behavior at school level presented gathering effect,thus affecting the health of children.In order to analysis on the context information of the level 2 variable,explained variables were included in th two-level logistic model.?4?The convergence of tow-level bayesion logistic regression modle was ok,as the length of MCMC was 3000.According to the results,the region,gender,at least one times a week of sweets/chocolates intake,agree with oral health is very important to our own life were the influence factors of dental caries in children?OR = 3.414,1.256,1.336 and 0.623?.Due to the introduction of the independent variable of region to the level 2,there was greatly reduction of variation in the level 2,namely the difference between school mainly caused by the difference between urban and rural areas.?5?According to the 1-level bayesion model,the gender,the frequency of sweets/chocolates₁ intake?at least one times a week?,if the numeber of brushing teeth is correct,agree with oral health is very important to our own life were 12 years old children's oral health independent factors.?OR=1.270,1.269,0.678 and 0.644??7?From Deviation Information Criterion?DIC?,the DIC value of two-level logistic?2558.75?was 184.88 lower than the value of a single level logistc regression model?2743.63?,and difference is more than 10.Therefore,the two-level logistic regression model fitting effect is superior to the single level logistic regression model.Conclusion:The status of 12-year-old children's oral health in Chongqing gathers in school level,and the data has the character of hierachical structrure.Multilevel model can be used to effectively handle the data with hierarchical structure,and can accurately analyze the influence factors of children's oral health.Region,gender,the frequency of eating sweets/chocolates₁?1 times a week or more?,agree with oral health is very important to their lives have a significant impact to children's caries rate.Prevention and control of caries of children can by establishing oral intervention model of school children that have the characteristics of difference between urban and rural.Multi-level bayesion model can be applied to the children's oral health evaluation and control when there are little groups in the high level.