Subjects and Design
We conducted a cross-sectional research survey of college students from 22 comprehensive colleges in Anhui Province, China between October 2019 and January 2020. An electronic questionnaire was sent to college students using an online survey. Verbal consent was obtained from all participants and there was no intervention throughout the information filling process. A total of 2,000 e-surveys were sent to these college students via a random sampling method within each college, and total returns were 1,817. The data analysis rejected 109 questionnaires and included a total of 1,708 valid questionnaires, with a response rate of 94.0%. The effective response rate was 85.4%. The exclusion criteria were as follows: (1) questionnaire missing values greater than 10%; (2) reading and comprehension difficulties, and (3) logical inconsistency of answers.
Moreover, our study is based on a cross-section of the population of Anhui province. The sample size was calculated according to the formula N = Z²P(1-P)/d² (Z = 1.96, P = 60.1%, d = 0.05) [32]. Where N is the sample size. Z is the statistic, when the confidence level is 95%, Z = 1.96. P was calculated using the oral health knowledge rate of 60.1% of the population from a previous survey [33]. And considering 20% invalid response rate. The minimum sample size obtained was 441. We exceeded the minimum required sample size to ensure the credibility of our results.
Ethical Approval
The study was ethically approved by the Medical Ethics Committee of Bengbu Medical College, China (No. 2019-062). Prior to the start of the study, the details of the study and the processing of the results were explained to college students. Students who did not want to cparticipate in the study could refuse to fill out the questionnaire, and all participants did so voluntarily.
Hypotheses
According to our hypotheses (Figure 1), self-rated oral health and oral health behaviours have a positive effect on subjective oral conditions and OHRQoL. Oral health behaviours are also positively related to self-rated oral health. Also, subjective oral conditions are positively related to OHRQoL.
Survey Instruments
We designed our questionnaire based on reference literatures. The questionnaire included the following items: demographic characteristics, self-rated oral health, subjective oral conditions, oral health behaviours, and OHRQoL.
The first part is self-rated oral health, which consists of 5 questions. This part is divided into three categories. The first category is the oral health perception, and use the question ‘What is your judgment about your oral health?’. The answer used the 5-point Likert scale to calculate the score, ‘0 = very good’, ‘1 = good’, ‘2 = fair’, ‘3 = poor’ and ‘4 = very poor’ [11]. Use ‘What is your assessment of your tooth health’ to determine the second category of tooth condition perception. The answer uses the 5-point Likert scale to calculate the score, ‘1 = excellent’, ‘2 = very good’, ‘3 = good’, ‘4 = average’, ‘5 = poor’. The third category is oral health interventions. The questions include ‘When did you last visit the dentist (stomatologist)?’. The response options were ‘1 = never’, ‘2 = last month’, ‘3 = last three months’, ‘4 = last six months’ and ‘5 = one year ago’. ‘What was your reason for going to the dentist?’ and ‘Have you received any oral health education (health education seminars) ?’. The response options are: ‘1 = yes’ and ‘2 = no’.
The subjective oral conditions were evaluated using 9 questions such as ‘Have you ever had bleeding gums?’; ‘Have you ever had tooth pain?’; ‘Did you ever feel TMJ pain when opening your mouth or chewing food?’. The answer uses the 3-point Likert scale to calculate the score, ‘0 = never’, ‘1 = occasionally’, ‘2 = frequently’.
Oral health behaviours include 9 questions. Participants filled in the questions according to their actual situation. This part is divided into four categories. The first category is brushing. The questions include ‘Do you brush your teeth ≥ 2 times a day?’ (B1); ‘Do you brush your teeth every time ≥ 3 minutes?’ (B2); ‘Do you change a toothbrush every three months?’ (B3). The second category is tooth cleaning. The questions include ‘Do you often use dental floss (or interdental brushes) to help clean interdental spaces?’ (C1) [34]; ‘Do you often gargle after meals?’ (C2); ‘Do you clean your teeth regularly (teeth washing)?’ (C3); ‘Is your brushing method the ‘horizontal tremor brushing method’ recommended by the Chinese Stomatological Association?’ (C4). The third category is the oral health checkup. The question is ‘Do you regularly perform oral health checkups?’ [34]. The fourth category is the use of fluoride toothpaste, the question is ‘Do you often use fluoride toothpaste?’ [20]. The option scores of the questions are ‘1 = Yes’ and ‘0 = No’.
The fourth part is OHRQoL. The Chinese version of Oral Health Impact Profile (OHIP)-14 was selected to evaluate OHRQoL. OHIP scores were significantly correlated with life satisfaction [6]. Response options are ‘4 = often’, ‘3 = fairly often”, ‘2 = occasionally’, ‘1 = hardly ever’, and ‘0 = never’ [35]. The higher the OHIP-14 score, the worse the OHRQoL is influenced by the oral condition.
The overall Cronbach's α for the questionnaire is α = 0.903. The internal consistency of this questionnaire is good.
Statistical analysis
We used IBM® SPSS® Statistics 20.0 and IBM® SPSS® Amos™ 24.0 for data analysis.
We used frequency, percentage and mean ± standard deviation to describe the demographic characteristics of participants. We tested the appropriateness of the scale using the KMO measure and the Bartlett spherical test. The KMO coefficient for this questionnaire was 0.927 and the p-value of Bartlett's test was < 0.001. We used exploratory factor analysis (EFA) to examine the degree of association between questions. A project analysis was carried out to test the degree of differentiation of the questions on a 27% scale.
The associations that existed between self-rated oral health, subjective oral conditions, oral health behaviours, and OHRQoL were analyzed by constructing a SEM. Before constructing the SEM, we first performed a confirmatory factor analysis (CFA) to assess whether the factor structure chosen based on our study was also acceptable in these data or whether it should be modified [36]. When constructing the SEM, each observed variable is imported and calculated as a mean value. We use the Bootstrap method [35,37] to verify whether there are mediating effects between the variables of interest in the model. We bootstrapped 5000 samples from the raw data (n = 1,708) by putting back a random sample. The confidence interval (CI) was 95% for the direct and indirect effects of each of the suggested mediating variables beyond zero. Using the skewness–kurtosis test to check the normality of the observed variables [38].
The model fit was evaluated using maximum likelihood estimate (MLE), chi-square (χ2) values, degree of freedom (DF), goodness of fit index (GFI), comparative fit index (CFI), normed fit index (NFI), incremental fit index (IFI), adjusted goodness of fit index (AGFI), root mean square error of approximation (RMSEA) and other indicators [36,39]. For CFI, GFI, NFI, IFI and AGFI, a fit index above 0.90 (preferably above 0.95) and an RMSEA less than 0.05, indicating that the model fits well [39-41]. Regression coefficients were used with a significance level of p < 0.05 [12].