This cross-sectional study uses a part of the 4th National Oral Health Survey conducted in China. All parents gave their informed consent for inclusion before their children participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the Chinese Stomatological Association (NO. 2014-003).
The sample size was calculated based on the data of the 3rd National Oral Health Survey in 2005, in which the prevalence of dental caries for those aged 12 years was 28.9%. The design effect (deff = 4.5), significance level (α = 5%), margin of error (δ = 10%) and non-response rate (20%) were also included in the following formula:
In total, 28,365 12-15-year-old adolescents should be recruited in 31 provinces across the country[15]. Hence, at least 3,660 individuals should be enrolled in this study. A multistage cluster sampling method was adopted to select the sample population. The probability proportion to size sampling determined that two urban and two rural samples should be used. Cluster and quota sampling determined the three primary high schools in each region. Three hundred and twenty students (80 for each age group) from each school were included.
After giving their informed consent, the students completed the questionnaire survey and underwent an oral examination. The questionnaire collected information on the students' personal and family demographics, including age, gender, region, father/mother's education level, whether they were an only child or had siblings, and the following:
1) Oral hygiene knowledge, including the impact of brushing, bacteria, sugar, fluoride, pit and fissure sealing, and other factors that affect the teeth and gingiva.
2) Oral hygiene attitude, mainly to evaluate whether they believe that oral health is important.
3) Oral hygiene behaviours, including brushing habits, frequency of snacking, smoking, dentist visits, and trauma.
4) Troubles caused by oral problems, including eating, talking, brushing, working, schooling, sleeping, smiling, easily troubled, and communicating.
All clinical examinations were implemented at schools with external equipment (portable dental chair, disposable dental mirror, ball-ended community periodontal index probe, and intraoral light-emitting diode light). Three examiners were selected and trained. The clinical practice training would be terminated when the Kappa value was greater than 0.8 to ensure the consistency of the inter- and intra-examiners. The following indices were used according to the criteria recommended by the World Health Organisation[16]. The DMFT, calculus (CI) and gingival bleeding (GB) indexes assessed decay, missing teeth caused by decay, filled teeth and the gingival health status. All teeth present were gently probed with a CPI probe at six sites, including mesial, mid, and distal on both buccal and lingual surfaces. Calculus and gingival bleeding were scored as present (1) or absent (0) and the number of teeth where calculus/bleeding was present were recorded.
Data were entered and statistically analysed using IBM SPSS Statistics version 21.0. The prevalence and average of dental caries (DMFT, DT, MT, FT), calculus and gingival bleeding were calculated. Since the measured count data were not normally distributed, their prevalence was used to assess the association of these indices and other variables. The indicators were all binary variables, and the other independent variables were also categorical. All the variables were independent of each other. Hence, Chi-square, Fisher's exact and z tests for post hoc comparisons were conducted to explore the relationship between these oral indicators and the sociodemographic and questionnaire variables. To evaluate the independent risk factors associated with caries and gingivitis in adolescents, a multivariate logistic regression model (method: backward logistic regression) was used. Those variables with P ≤ 0.10 obtained through bivariate analyses were examined if they should be included in the final models. The tolerance and variance inflation factors obtained through the linear regression model were used to diagnose multicollinearity between independent variables. A P-value of < 0.05 was considered statistically significant.