Study design and population
This was a cross-sectional study based on caries examination and parental interviews using a structured questionnaire. It was conducted in 2014 among 18-year-old Lithuanian adolescents and their parents. A multistage cluster sampling method was used to draw a representative sample of 1063 adolescents and their parents. Lithuania is divided into 10 counties. In the first stage, each county was divided into smaller urban and rural administrative units (clusters). During the second stage, in each cluster, the first and last schools (sub-clusters) on the alphabetic list of all the schools were selected (based on data from the education management information system of the Centre of Information Technologies in Education). In the third stage, gymnasium class (a block) was selected. One hundred parents of chosen adolescents from each selected block were asked to complete the questionnaire. In total, 2000 adolescents from across the country were approached. Only 1063 of these met the inclusion criteria (age range 17.5–18.5 years and voluntary agreement to participate in this study). The sample size was calculated using Paniott’s formula with an error of 0.05 % based on the 18-year-old population; this amounted to 37,036 according to Statistics Lithuania. Using this formula, it was determined that at least 396 18-year-old adolescents should be included in the study.
The dental examinations were performed according to the methodology of oral status evaluation recommended by the World Health Organization , under standardized conditions using the dental chairs available in the school’s dental offices and portable dental units equipped with a halogen light source, compressed air and a suction device. The examinations were performed by two pediatric dentists who were trained and calibrated to record the parameters of oral health. Training and calibration was performed on 35 18-year-old subjects who were not included in the final sample. Cohen’s kappa statistic was used to test the reliability of observations made by two different examiners (inter-examiner) and two observations of each examiner made at two different points of time (intra-examiner) at the baseline and after 1 year, at the end of the study. Reliability of the caries diagnostic criteria was assessed at the tooth surface level. The kappa value for inter-examiner reliability was 0.92, and for intra-examiner reliability the values ranged from 0.92 to 0.94.
To identify the group of adolescents in which dental caries was polarized, the SiC index was used . Individuals who had the highest third of DMFT scores formed the SiC positive group, while all others formed the SiC negative group.
Participating parents completed an anonymous self-administrated questionnaire of 25 questions. They were asked about their children’s early childhood, including information about biological factors: the mother’s pregnancy course, preterm birth, and child birth weight. They were also asked to provide information about when their child’s first deciduous and permanent teeth erupted, at what age the child first visited the dentist, and whether the child had experienced caries in the deciduous teeth. Social determinants, such as the mother’s education level and the child’s birth order, were also included in the survey. Pilot study of the questionnaire was performed 6 months’ prior the study. The questionnaire was additionally administrated for a 50 participants who were not included in the final sample. Corrections to the questionnaire were made according to the comments presented by the parents.
To determine the extent to which all the questions in the questionnaire measured the same concept, Cronbach’s alpha coefficient was calculated. Measurement of internal consistency gave a coefficient of 0.8, which showed a high level of reliability for the questionnaire. Validation of the questionnaire was performed through the assessment of content validity. Content validity was examined in order to assess the extent to which a questionnaire measures what it is intended to measure.
Statistical data analysis was performed using SPSS 22.0 (Statistical Package for the Social Sciences for Windows). In accordance with the principles of descriptive data analysis, the mean of the quantitative variables with a standard deviation (SD) was presented. The interdependence of qualitative characteristics was evaluated with the chi-square (χ2) criterion. If variable distribution met the distribution normality assumption, Student’s (t) criterion was applied to compare the quantitative size of two independent groups. When the variables did not meet the distribution normality condition, a significance level was verified by a nonparametric method using the Mann–Whitney test.
Receiving operating characteristic (ROC) curve analysis helped to decide where to draw the line between sensitivity and specificity. It was used to assess the accuracy of predictions and determined the optimal cut-off value.
The probability of an event given a certain risk indicator was calculated using logistic regression analysis, including odds ratio (OR) and its confidence interval (95 % CI). For complex evaluation of probability, the multivariate logistic regression model was used. The threshold for statistical significance was set at P < 0.05.
Cohen’s kappa statistic and Cronbach’s alpha coefficient were calculated to check the reliability of the dental caries examination and questionnaire, respectively.