Social inequalities in overweight and obesity, and dental caries among adolescents in Northern Norway: a cross-sectional study from the Tromsø Study Fit Futures cohort.

Background. Non-communicable general and oral health conditions share common social determinants. We aimed to investigate how social inequalities (based on study program, parent’s education and employment status) manifest in general health (BMI or waist circumference) and oral health (untreated caries) in a sample of adolescents from Northern Norway. Methods. Data from 464 girls and 494 boys from the population-based Tromsø study Fit Futures, which included first-year students attending upper secondary school in 2010-2011 from two municipalities in Northern Norway (1038 participants in total, 93% participation rate) was analyzed. Multivariable binary logistic regression analyses stratified by sex was used to investigate the association between socioeconomic position indicators and body weight, waist circumference, and untreated dental caries. Comparative models were adjusted for age, birth country, household composition, and health/health behavior variables. Results. Boys enrolled in the general studies and sports programs (versus vocational) had lower odds of being overweight/obese (OR 0.39, 95% CI 0.19-0.79 and OR 0.21, 95% CI 0.07-0.62, respectively), of having high waist circumference (OR 0.47, 95% CI 0.24-0.91 and OR 0.30, 95% CI 0.11-0.77, respectively), and more untreated caries (OR 0.54, 95% CI 0.31-0.96 and OR 0.46, 95% CI 0.22-0.97, respectively). Lower mother’s education was positively associated with overweight/obesity (OR 3.68, 95% CI 1.38-9.79) and lower father’s education was negatively associated with untreated caries (OR 0.42, 95% CI 0.18-0.98) among boys. Girls enrolled in the general studies program (versus vocational) had lower odds of having more untreated caries (OR 0.48, 95% CI 0.28-0.85). Conclusions. social study

but dental caries experience as well as overweight/obesity separately were associated with SEP [13].
Another study among children in South Pacific found ethnicity to be common risk factors for oral health and obesity [14]. Ethnicity is one of a social position proxy, and social position takes a central role in generating health inequalities [4]. Therefore, it would be reasonable to look at common risk factors in the broader social context (based on WHO conceptual model of social determinants of health) when investigating links between non-communicable general and oral health conditions [15,16].
Even though Norway is a so-called welfare state and a high-income country, social inequalities in general and oral health have been shown, with those with lower SEP reporting poorer outcomes [17][18][19]. To our knowledge, there are no studies investigating social inequalities in general and oral health conditions among the same adolescents in Norway.

Study aim, design, and population
The aim of the present study was to investigate how social inequalities (based on adolescent's own study program, parents' education and employment status) manifest in general health (BMI or waist circumference) and oral health (untreated caries) in a sample of adolescents from Northern Norway. Therefore, we investigated the association between SEP indicators and measures of overweight/obesity and abdominal obesity (as general health outcomes), and untreated dental caries (as an oral health outcome) among adolescent girls and boys from Fit Futures 1 (FF1). The main hypothesis was that body weight, waist circumference and untreated dental caries differ by SEP as measured by the adolescent's own study program, parents' level of education and employment status. The second hypothesis was that study program social inequalities in general health are similar to social inequalities in oral health among the same individuals.
This cross-sectional study included data from a population-based cohort study, the Tromsø Study Fit Futures (FF1). FF1 was conducted in the Tromsø school district, Northern Norway (the urban Tromsø municipality and the rural Balsfjord municipality) [20]. All first-year students attending upper secondary school (seven schools in Tromsø and one in Balsfjord) in 2010-2011, mainly aged [15][16][17][18][19] years, were invited to participate in FF1. Of a total of 1117 invited students, 1038 participated (93% participation rate), while 1010 (498 girls and 512 boys) volunteered to participate in the oral health portion (90% participation rate), which consisted of a dental evaluation [21,22]. Data on general health was collected at the Clinical Research Unit, University Hospital of North Norway [21]; data on oral health was collected at the University Dental Clinic [22]. All procedures were done during school hours [22]. After exclusion of participants aged 19 years or older, and those with missing data on height, weight, waist circumference and untreated dental caries, 464 girls and 494 boys (86% participation rate) were eligible for inclusion in the analyses.

Variables and measurements
Socioeconomic position indicators, demographic characteristics, health, and health behavior SEP indicators (adolescent's study program, mother's education, father's education, and parents' employment) were taken from the pretested, electronic self-administered Fit Futures questionnaire, as were demographic characteristics (age, sex, birth country, and household composition) [Furberg 2010, cited in 23] ( Table 1).
Following the WHO conceptual framework for the social determinants of health, as intermediary determinants, twelve behavioral, biological and psychosocial variables were selected to reflect health and health behavior: history of chronic disease, alcohol intake, smoking, snuff use, physical activity, sugar-containing sweets and beverages intake, other dietary factors (including intake of omega 3 fatty acids-rich food/supplements, dairy products, fruits/vegetables, vitamin or mineral supplements), tooth brushing frequency, dental satisfaction and self-esteem, mental health and sleep, which were taken from the Fit Futures questionnaire, and vitamin D status, which was measured as serum vitamin D (25-hydroxyvitamin D) level. In order to depict overall health and health behavior, a "composite health/health behavior" variable was composed of these 12 variables (Table 1). The 12 variables were dichotomized (0 and 1) and the scores were added up, resulting in 13 groups (scores 0-12), which were later categorized into very good health/health behavior (composite health/health behavior score 0-3), good health/health behavior (score 4-6), and less good health/health behavior (score 7-12).

Outcome variables
Body mass index (BMI) and waist circumference were used to assess body weight categories and abdominal obesity, respectively. Height and weight were measured to the nearest 0.1 cm and 0. 1  The oral health outcome untreated dental caries was expressed as a decayed teeth (DT) component of a decayed, missed, filled teeth (DMFT) index, which was recorded by a single dentist [22].

Statistical analysis
Statistical analyses were performed in Statistical Package for the Social Sciences (SPSS, Version 26.0, IBM Corp., Armonk, NY, USA). Chi-square test was used for categorical variables to analyze the differences in SEP indicators, demographic characteristics, health, health behavior, body weight, abdominal obesity and untreated dental caries between the study programs (general studies, sports and physical education, and vocational) stratified by sexes. The binary logistic regression analysis was stratified by sex, as it has been shown that health behavior differs between adolescent girls and boys [32,33]. Univariable binary logistic regression analysis was used to identify the characteristics associated with the outcomes.
Multivariable binary logistic regression Model 1 was constructed separately for girls and boys, and was adjusted for all SEP indicators and demographic characteristics (independent of significance), and for all health and health behavior characteristics that achieved statistical significance p≤0.2 in the univariable binary logistic regression analyses for that sex. Multivariable binary logistic regression Model 2 was comparative; therefore it was the same for girls and boys and was adjusted for all SEP indicators, demographic characteristics, and the composite health/health behavior variable. The hierarchical regression (blockwise) method was used. All SEP indicators were entered in the first block and the covariates in the second block. The assumption of multicollinearity (tolerance, VIF statistics and eigenvalues) was not violated in any of the models [34]. The Hosmer-Lemeshow goodness-of-fit statistic yielded p>0.05 for all models constructed. Nagelkerke R 2 was recorded for the first (SEP) block and the whole model for each model [34]. The level of significance was set at p=0.05 and odds ratios (ORs) are presented with 95% confidence intervals (CIs). The present study was approved by the Regional Committee of Medical and Health Research Ethics (reference number 2018/172/REK nord).

Sample characteristics
The mean age of the 464 included girls was 16.16 (standard deviation (SD) 0.48) years and for the 494 included boys it was 16.11 (SD 0.56) years. A higher proportion of adolescents (both girls and boys) in the vocational program had mothers and fathers with lower education, and reported that at least one parent did not work full time, compared to other study programs ( Table 2). A higher proportion of girls and boys in the vocational program were obese and had more teeth with untreated dental caries compared to adolescents in the general studies program, and in the sports and physical education program ( Table 2). Boys in vocational program had a higher prevalence of high waist circumference compared to other study programs, while there was no difference in prevalence of high waist circumference between study programs among girls ( Table 2).

Associations with socioeconomic position indicators
Girls enrolled in the general studies program versus the vocational program had 52% lower odds of having at least one tooth with untreated caries (OR 0.48, 95% CI 0.28-0.85) ( Table 5, Fig. 1).
Boys who had mother with a lower education level (college less than 4 years vs college 4 years or more) had more than three times higher odds of being overweight/obese (OR 3.68, 95% CI 1.38-9.79) and two and a half time higher odds of having high waist circumference (OR 2.47, 95% CI 1.03-5.95) (Tables 3-4). Boys who had father with a lower education level (college less than 4 years vs college 4 years or more) had lower odds of having at least one tooth with untreated caries (OR 0.42, 95% CI 0.18-0.98) ( Table 5).

Discussion
To our knowledge, this is the first study investigating inequalities in overweight and obesity, and dental caries among the same adolescents using the WHO conceptual model for social determinants of health. The present study demonstrated social inequalities in body weight, waist circumference and untreated dental caries among boys, with poorer outcomes observed among those enrolled in a vocational program compared to those enrolled in a sports and physical education program or a general studies program. In addition, among boys, social inequalities in body weight and waist circumference were based on mother's education and social inequalities in dental caries were based on father's education. Social inequalities in untreated dental caries among girls were based on adolescent's study program. This was a cross-sectional study, whose design in general is prone to confounding and does not allow to establish causality [35]. One way to control for confounders is to adopt a multivariable analysis, as was implemented in the present study [35].
The initial participation rate in FF1 was high, reaching 93%, and participation in the oral health part of FF1 was only slightly lower (90%). It is possible that this decrease in attendance to the dental evaluation that constituted the oral health part of FF1 was associated with low parental education, unemployment, and low income [36]. After exclusions, our final study sample represented 86% of all students invited to FF1, but in multivariable binary logistic regression, the number of participants was reduced to 58% among girls and 61% among boys due to missing data; therefore self-selection bias cannot be ruled out. The sample was collected from both a densely populated urban area (Tromsø, seven schools) and a sparsely populated rural area (Balsfjord, one school) including all upper secondary schools in Tromsø school district, Troms County, Northern Norway. In this county, 29% of the population resides in sparsely populated areas; therefore, the population residing in densely populated areas might be overrepresented in the study sample (7 schools in urban area versus 1 school in rural area). It has been shown that living in densely populated areas is associated with higher physical activity and thus probably better health outcomes among adolescents in Norway [32].
It must be noted that 16-18% of adolescents in Troms County do not live in their parents' household.
Indeed, as Troms County is large, adolescents sometimes have to move from where their parents live to where the school is located -creating the household composition of "living without adults". This living situation occurs due to adolescents' need for education; not necessarily because they have a higher level of maturity and hence, oral health may be joepardized. It has been also shown that having an immigrant background was related to worse general and oral health outcomes among children and adults in Norway [37][38][39]. In Tromsø municipality in 2012, 4.8% of immigrants were aged 16-19 years [40]. In our study sample 6% of girls and 5% of boys reported that they were born outside Norway, indicating that our sample might be representative of the national population with respect to immigrant background. A pretested, electronic, self-administered questionnaire was employed to collect data on SEP indicators and most of the covariates. Structure and content of the questionnaire were to a large degree adapted from the Tromsø Study among adults [41]. In general, questionnaires are prone to bias, especially regarding sensitive data, like alcohol intake and tobacco use. However, selfadministration has been shown to decrease reporting bias [42].
In order to get a global measure of health and health behavior, a composite health/health behavior variable was calculated. It has been shown that composite measures might be the best approach to study socioeconomic differences in health among adolescents [43]. Nonetheless, the validity of our composite health/health behavior variable has not been tested and is therefore debatable. This must be kept in mind when interpreting the data based on Model 2. On the other hand, less good health behavior was observed more often among those enrolled in a vocational program compared to those in a general studies program or in a sports and physical education program. This is in line with other studies [43,44]. In addition, the ORs between Model 1 (adjusted for separate health and health behavior characteristics that achieved statistical significance p≤0.2 in the univariable binary logistic regression analyses for that sex) and Model 2 (adjusted for the composite health/health behavior variable) were similar. Construct validity was to some extent demonstrated in that the composite health/health behavior variable among girls discriminated significantly in the predicted direction between overweight/obese and normal-weight participants, as well as between participants having normal and high waist circumference, and having at least one tooth with untreated caries versus no untreated caries. However, the psychometric properties of the composite variable need to be further tested and evaluated.
Previous Norwegian study investigating social inequalities in health behavior among adolescents suggested that study program in upper secondary school is a potential proxy of an adolescents' socioeconomic status [43]. Therefore, study program was chosen as the SEP indicator in this study. In the Norwegian school system, there is a lawful right, but not an obligation, to complete 1 year of upper secondary school. Students can apply for a general studies program, including a sub-path of a sports, or a vocational study program. The general studies program gives admission to higher education after three years. At the vocational study program, normally after two years of school training, a student goes in apprenticeship for two years. The completion rates (by normative length of study) differ according to study path (75% in general study program, 37% in vocational study program, during 2013-2018, respectively), and varies by sex, geographic area and parent's education. Adolescents' choice of study program has been shown to correlate with their social background [45] and health-related behaviors [43,46]. As in the present study, the same previous Norwegian study also indicated that parents' education and occupation are applicable when investigating social inequalities among adolescents. In addition, in Norway, previous studies have also shown that mother's and father's education associated with child's health behavior and adverse health events [43,47,48]. We had no data on parents' occupation, therefore, parents' employment was used as a substitute variable in this study, as it has been associated with health and health behaviors among adolescents [49].
One of the general health indicator in this study, body weight (expressed by BMI), was measured. BMI is commonly used as an indicator of overweight and obesity. Indeed, BMI is a ratio between weight and height, and it cannot distinguish between body fatness and fat-free mass [50]. On the other hand, it has been shown that BMI-for-age was a good indicator of body fatness, especially among heavier children and adolescents [51]. In addition to BMI, as another general health indicator, we have used waist circumference. Waist circumference is a specific measure to define abdominal fatness [52]. In our study, the two approaches gave quite similar results, but using BMI presented results with a higher Nagelkerke R 2 .
The oral health outcome in this study was untreated caries in dentine (D 3-5 T); it was measured and expressed as DT component of DMFT index. The DT component reflects the treatment need, or in other words, the severity of disease, but does not take into consideration dental caries experience (filled and missing due to caries teeth).
Adolescence has been described as a key period for developing health behaviors, thus determining future health. West discussed that it is also a period during which social equality in health is more predominant than inequality, and he suggested that this might be related to the youth culture, secondary school, and school friends that become more important for health equality/inequality than parents' SEP [53]. Despite the equalizing effect, school creates new inequalities related to study program and/or climate and gender [54]. In our study, the association between study program, one of the SEP indicator, and untreated dental caries proved to be statistically significant among girls.
Among boys, the statistical significant associations were between study program and all three outcomes. It must be noted, that adolescents enrolled in the sports study program are probably healthier because of the fact that they are in this study program. In addition, oral health condition, untreated caries, was also better among those boys enrolled in sports study program compared to adolescents in vocational study program. Previous prospective Norwegian cohort study demonstrated social inequalities in health behaviors only among girls based on admission to given study programs [43]. In addition to the findings of the previous study, our study showed that social inequalities based on study program are also observed among boys. This finding may be explained that, in Norway, the choice of study program has been shown to depend mainly on the occupation of role models, role models for adolescents being mostly their friends and acquaintances, persons from the same social environment [55].
Lower mother's education, another SEP indicator used in this study, was positively associated with higher BMI and high waist circumference only among boys (Tables 3-4). This finding might refer to gender orientation in adolescents' behavior. It might be that boys are less mature and more dependent on their mothers, as mothers have been shown to be "the prime mover in the health and welfare of the child" [56]. Our findings regarding mother's education and boy's health is in contrast to a study from the USA, in which father's health-risk lifestyle, which consisted of diet, physical activity, smoking, alcohol use, and sleep, affected boys' health-risk behavior, while mother's behavior affected girls' behavior; however parents' health-risk lifestyle was not included in this study [57]. It has been shown in Norway that father's occupation predicted changes in health behavior among 13-21-year-old girls [43]. We may speculate that father's occupation is linked to father's education, and in the present study, according to Model 2, father's education was associated with untreated caries among boys, and this finding is in contrast with the previously mentioned study. Even though it has been demonstrated that only few adolescents based their choice of the study program on their parents' opinions, the indirect influence may not be ruled out [55]. Given the differences across genders of parents and children, future studies investigating social inequalities in adolescents should address the issue of gender in the relation between parents and their children. The above mentioned findings confirm our main hypothesis that social inequalities in general health and oral health differ by SEP as measured by the adolescent's own study program and parents' level of education.
A study in Hungary showed that incomplete parental employment (unemployed, retired, housewife) resulted in inconsistent associations; it was positively associated with health conditions, like depressive and psychosomatic symptoms, but negatively associated with behavioral factors, like smoking, drinking, and drug use among adolescents [49]. In the present study parents' employment, another SEP indicator, did not associate with any of the outcomes, however lower father's education was negatively associated with untreated caries. Dental caries is a condition that is highly dependent on behavioral factors. Indeed, in the present study untreated caries was positively associated with less good health/health behavior (versus very good health/health behavior) according to Model 2 and with high sugar-containing sweets and beverages intake frequency according to Model 1. Therefore, it could be argued that the negative effect of parents' employment on behavioral factors in the Hungarian study is in line with our findings.
Among the same boys, social inequalities for having higher BMI and high waist circumference were similar to social inequalities in untreated dental caries (Fig.1). This finding confirms our second hypothesis, that social inequalities in general health are similar to social inequalities in oral health.
One may deduce that overweight and obesity, and dental caries are related through common social determinants (in this case SEP indicator, study program) rather than direct oral-general health links; thus supporting the broader concept of social determinants as common risk factors between oral and general health. It must be noted that it is not the study program itself that is a risk factor, but the social context that leads the adolescent to choose a particular program. Body weight, waist circumference and untreated dental caries did not produce similar social inequalities among girls. This may be explained by the fact that adolescence is a relatively healthy period of life, when harmful health behaviors have not yet had time to manifest. In addition, recorded health behaviors and our chosen general health outcome, body weight (expressed by BMI) and waist circumference are prone to change, especially during adolescence, which raises questions about whether a single time measurement of these outcomes can depict the real situation over time [58]. In addition, Nagelkerke R 2 , showed that the explained variance was not above 20% in any of the multivariable models and was higher for boys than for girls, indicating that there are other variables that may explain the outcomes and change the association estimates.

Conclusions
Social inequalities in general health and oral health were based on different social determinants for girls and boys. Based on study program social inequalities in general health were similar to social inequalities in oral health among boys. This suggests that the relationship between overweight and obesity, and dental caries maybe be explained by common social risk factors. Our findings support the broader concept of social context as a common risk factor for general and oral health.

Consent for publication
Not applicable.

Availability of data and materials
The Fit Futures datasets used and analyzed during the current study were supplied by "Helsefak ISM Tromsøundersøkelsen" under the agreement and so cannot be made freely available. Requests for access to these data should be made to tromsous@uit.no.

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable.

Authors' contributions
LSM drafted the manuscript, TAT substantively revised it. TAT made a substantial contribution to the conception of this work. LSM and ASF contributed to the design of the study. ASF contributed to collection of data. LSM analyzed the data, and together with TAT, interpreted it.
All authors have approved the submitted version and agreed to both be personally accountable for their own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature.    Table 5. Odds ratios (ORs) and 95% confidence intervals (CIs) for the association between untreated dental caries in dentine (D 3-5 T) and socioeconomic position (SEP) indicators according to univariable and multivariable binary logistic regression analyses. The number of participants in each analysis differs due to missing data. The Tromsø Study Fit Futures 1.