Dental caries is one of the most common preventable chronic diseases, and predisposition to the development of caries continues throughout life(Reference Selwitz, Ismail and Pitts1,Reference Pitts, Zero and Marsh2) . A number of physical, biological, socioeconomic, genetic, dietary, behavioural and psychological factors have been associated with caries expression(Reference Selwitz, Ismail and Pitts1,Reference Pitts, Zero and Marsh2) . Caries denotes dissolution of tooth substance resulting from metabolic events in the biofilm on tooth surfaces(Reference Selwitz, Ismail and Pitts1,Reference Pitts, Zero and Marsh2) . Frequent consumption of easily fermentable carbohydrates is a prerequisite for the development of caries, whereas certain foods and drinks such as cheese and milk have been found to prevent caries(Reference Moynihan and Kelly3–Reference Gondivkar, Gadbail and Gondivkar9).
The effects of single food items and nutrients on health may not be separable in real life because diets consist of combinations of foods with complex nutrient compositions. Therefore, improvements in food consumption or comprehensive dietary patterns may be more easily translated to real-life conditions than improvements in the consumption of single foods and the intake of certain nutrients. Dietary quality indices have been developed to assess adherence to desirable diets comprehensively. For example, the Baltic Sea Diet Score (BSDS) has been developed to illustrate a healthy dietary pattern in Nordic countries(Reference Kanerva, Kaartinen and Schwab10,Reference Kanerva, Kaartinen and Schwab11) . A higher BSDS has been associated with better periodontal health in adults(Reference Jauhiainen, Suominen and Kanerva12,Reference Jauhiainen, Ylöstalo and Knuuttila13) . However, there are no published reports on the association of BSDS with caries in any age group. Some eating behaviour traits, assessed by the Children’s Eating Behaviour Questionnaire (CEBQ), have also been associated with caries in children(Reference Wardle, Guthrie and Sanderson14–Reference Nembhwani and Winnier16). However, to our knowledge, no studies have investigated these associations using CEBQ in populations of European children.
Eating and dietary habits are formed in early childhood(Reference Scaglioni, De Cosmi and Ciappolino17,Reference Birch, Savage and Ventura18) and are crucial for the development of caries. It is therefore of utmost importance to determine which eating and dietary habits are the strongest correlates and predictors for caries in childhood. Our aim was to investigate the cross-sectional and longitudinal associations of BSDS, food consumption, eating frequency and eating behaviour with caries experience in a 2-year follow-up study in a general population of Finnish children.
Methods
Study design and study population
The present analyses are based on the baseline and 2-year follow-up data from the Physical Activity and Nutrition in Children (PANIC) study that is a non-randomised controlled trial on the effects of a combined physical activity and dietary intervention on cardiometabolic risk factors and other health outcomes in a population sample of children from the city of Kuopio, Finland (ClinicalTrials.gov NCT01803776)(Reference Lakka, Lintu and Väistö19). The Research Ethics Committee of the Hospital District of Northern Savo approved the study protocol in 2006 (Statement 69/2006). The parents or caregivers of the children gave their written informed consent, and the children provided their assent to participation. The PANIC study has been carried out in accordance with the principles of the Declaration of Helsinki as revised in 2008.
We invited 736 children aged 6–9 years who started their first grade in primary schools of Kuopio in 2007–2009 to participate in the study (Fig. 1). Altogether, 512 (70 %) children (248 girls, 264 boys) accepted the invitation and participated in the baseline examinations. The participants did not differ in gender, age, height – standard deviation score (SDS) or BMI-SDS from all children who started the first grade in the city of Kuopio in 2007–2009. We excluded six children from the study at baseline either owing to their physical disabilities that could hamper participation in the intervention or withdrawal of the families because they had no time or motivation to attend the study. We also excluded data from two children whose parents or caregivers later withdrew their permission to use these data in the study. The final study sample thus included 504 children at baseline.
Fig. 1. Flow chart of children participating in the baseline and 2-year follow-up examinations and having data on oral health for caries analyses in the Physical Activity and Nutrition in Children (PANIC) study.
These children were allocated to the intervention group (306 children, 60 %) or the control group (198 children, 40 %). The intervention group had six individualised and family-based physical activity and dietary counselling sessions over the first 2 years that occurred 0·5, 1·5, 3, 6, 12 and 18 months after baseline examinations. The control group received general verbal and written advice on health-improving physical activity and diet at baseline but no active intervention.
Of the 504 children who participated in the baseline examinations, 440 (87 %) also attended the 2-year follow-up examinations. Altogether, 487 out of the 504 children attending the baseline examinations had data on oral health and 406 out of the 440 children attending the 2-year follow-up examinations had data on oral health needed for the present analyses.
Assessment of oral health and oral self-care
Experienced dentists of the research group (authors A.V. and T.I. and three other research group dentists) performed the oral health examinations at baseline and at 2-year follow-up according to a strict predefined protocol. The dentists registered caries according to the WHO guidelines with the exception that also initial lesions were registered, although separately from decayed lesions. Caries lesions were examined visual-tactilely from five tooth surfaces, and Fibre-optic transillumination was used to transilluminate the tooth surfaces. Initial lesions were defined as primary lesions that were limited to the enamel, whereas decayed lesions were defined as lesions that extended to the dentin of a tooth. The decayed, missing and filled surfaces (DMFS/dmfs) index combining registrations in both primary (dmfs) and permanent (DMFS) teeth was calculated and used as the main outcome, but also the decayed, missing and filled teeth (DMFT/dmft) index was calculated. Missing teeth due to exfoliation were not separated from the DMFS/dmfs index. The Silness–Löe Plaque index was used to register plaque(Reference Löe20). We assessed oral self-care habits, including the frequency of tooth brushing and the frequency of using xylitol-containing chewing gum, by a questionnaire filled out by the parents or caregivers. Because our participants had mixed dentition, we summed up the DMFS, dmfs and initial caries surfaces in permanent and primary teeth into our main outcome variable, describing the total number of tooth surfaces affected by caries, called in this text as ‘caries experience’.
Assessment of dietary and eating habits
Food consumption, eating frequency and nutrient intake were assessed by food records. The records covered four predefined and consecutive days at baseline, including two weekdays and two weekend days or three weekdays and one weekend day(Reference Eloranta, Lindi and Schwab21). Two food records at 2-year follow-up examinations covered three days and consisted of two weekdays and one weekend day(Reference Lakka, Lintu and Väistö19). Clinical nutritionists examined the filled food records together with the family and added any missing information. Food consumption and nutrient intake were assessed using the Micro Nutrica® dietary analysis software, Version 2.5, based on detailed information about the nutrient content of foods in Finland and other countries(Reference Eloranta, Venäläinen and Soininen22).
Overall diet quality was assessed using BSDS, which is based on foods typically consumed by the general population in the Nordic countries(Reference Kanerva, Kaartinen and Schwab10). BSDS was calculated by summing the scores of six components of food consumption and two components of nutrient intake in quartiles of the present population of children, as described earlier(Reference Eloranta, Schwab and Venäläinen23). The six components of food consumption were fruits and berries (scored 0–3), vegetables excluding potatoes (0–3), high-fibre (≥ 5 %) grain products (0–3), low-fat (< 1 %) milk (0–3), fish (0–3), and red meat and sausages (3–0). The two components of nutrient intake were total fat intake as a percentage of total energy intake (3–0) and the ratio of PUFA to SFA (0–3)(Reference Kanerva, Kaartinen and Schwab11). BSDS ranges between 0 and 24, and a higher score indicates better adherence to the diet.
Eating behaviour was assessed using CEBQ(Reference Wardle, Guthrie and Sanderson14). The questionnaire was translated into Finnish, and parents were given instructions on how to fill out the questionnaire on behalf of their children(Reference Jalkanen, Lindi and Schwab24). CEBQ contains thirty-five questions and divides eating behaviour into eight subscales: enjoyment of food, food responsiveness, satiety responsiveness, slowness in eating, emotional undereating, food fussiness, desire to drink and emotional overeating. Each subscale consists of 3–6 statements, each of which was rated on as a five-point Likert scale (never = 1, rarely = 2, sometimes = 3, often = 4, and always = 5). The scores of statements that belonged to the same subscale were summed, and the means and standard deviations were calculated. We used the mean of each eating behaviour subscale in the analyses. A higher mean value in the subscale indicates a greater likelihood for the given eating behaviour trait in a child’s eating behaviour. Cronbach’s α coefficients as measures of internal consistency were calculated for each eating behaviour subscale and all statements of the subscales(Reference Jalkanen, Lindi and Schwab24). We used the modified mean of emotional undereating because one of the statements (‘My child eats more when s/he is happy’) showed a low total correlation(Reference Jalkanen, Lindi and Schwab24). After excluding this statement, the subscale showed good internal consistency(Reference Jalkanen, Lindi and Schwab24).
Assessment of body size, pubertal status and household income
Body height and weight were assessed by trained research staff in the morning. The children fasted for 12 h(Reference Eloranta, Schwab and Venäläinen23,Reference Väistö, Haapala and Viitasalo25) . Body height was measured three times using a calibrated wall-mounted stadiometer to an accuracy of 0·1 cm for the children standing in the Frankfurt plane without shoes. The mean of the nearest two values was used in the analyses. Body weight was measured twice using a weight scale integrated into the Inbody 720® bioelectrical impedance device (Biospace) to an accuracy of 0·1 kg; the children had emptied the bladder and wore light underwear. The mean of the two values was used in the analyses. BMI was calculated by dividing body weight (kg) with body height (m) squared, and BMI-SDS was computed using Finnish reference data(Reference Saari, Sankilampi and Hannila26). Overweight and obesity were defined using the International Obesity Task Force criteria corresponding to an adult BMI cut-point at 25 for overweight and at 30 for obesity(Reference Dietz and Bellizzi27,Reference Cole, Bellizzi and Flegal28) . A research physician assessed the pubertal status for girls according to breast development (B 1–5) and for boys according to testicular volume (G 1–5) using the Tanner staging method(Reference Marshall and Tanner29,Reference Marshall and Tanner30) . Girls were defined having entered clinical puberty if their breast development had started (Stage ≥ 2). Boys were defined having entered clinical puberty if their testicular volume measured by an orchidometer was ≥ 4 ml (Stage ≥ 2). We used household income reported by the parent who had higher income without taken taxes, and it was categorised as ≤ 30 000€, 30 001–60 000€ and ≥ 60 001€.
Statistical methods
The sample size calculations for the PANIC Study have been described earlier(Reference Lakka, Lintu and Väistö19). The statistical analyses were performed using the IBM SPSS Statistics software, Version 27.0 (IBM Corp.). The normality of distributions of the variables was verified visually from histograms and by the Kolmogorov–Smirnov test. Differences in basic characteristics between genders were tested using the independent samples t test for continuous variables with normal distributions, the Mann–Whitney´s U-test for continuous variables with skewed distributions and the Pearson’s χ2 test for categorical variables. Linear regression analyses with various adjustments were performed to study the association of BSDS, food consumption, eating frequency and eating behaviour with caries experience at baseline. First, the dietary variables were entered one by one into a model adjusted for age and gender (Model 1). Second, the dietary variables were entered one by one into a model adjusted for age, gender, household income, BMI-SDS, the frequency of tooth brushing and the frequency of using xylitol-containing chewing gum (Model 2). These potential confounding factors were chosen based on earlier evidence on their associations with caries(Reference André Kramer, Petzold and Hakeberg31–Reference Tseveenjav, Suominen and Hausen33). Third, all these dietary variables were entered simultaneously into a stepwise model (Model 3) to study what would be the best combination of variables to be independently associated with caries experience according to model selection criteria (F-test). Differences and associations with a P-value < 0·05 were considered statistically significant.
We examined the associations of changes in BSDS, food consumption, eating frequency and eating behaviour with caries a change in caries experience over 2 years using generalised linear mixed-effects models. Separate generalised linear mixed-effects models adjusted for the fixed effects of gender, study group and household income at both time points and the repeated effect of time were performed to test the associations of each variable with caries experience. The random subject-specific intercept was used in the models. A Bayesian information criterion was used to measure model adequacy and find the best predicting parameters for the change of caries experience, a lower value indicating a better model. We prioritised choosing a model with an optimal balance between good fit and complexity. We also tested models allowing dependent variable clustering on a subject within the school level. However, because the three-level structure resulted in unnecessary complexity and did not improve model fit, clustering within schools was not taken into account in the final models. We also tested a generalised linear mixed-effects model only with the study group and time to determine whether dietary and physical activity intervention had an effect on change in caries experience and found no such effect. However, to minimise residual confounding due to the effect of intervention on dietary factors that could affect caries risk, we adjusted the data for the study group in the generalised linear mixed-effects models.
Results
Characteristics of children at baseline
More than half (54 %) of the participants had DMFS/dmfs = 0, and more than a third (37 %) of the participants also did not have initial caries lesions. Most participants (62 %) brushed their teeth at least twice a day, and 43 % used xylitol-containing chewing gum every day. There were no statistically significant differences in DMFS/dmfs, the frequency of tooth brushing or the frequency of using xylitol-containing chewing gum between genders (Table 1). However, girls had less primary teeth, more permanent teeth and more initial caries on their permanent teeth than boys. Girls were also lighter and shorter than boys (Table 1). Furthermore, girls had less main meals per day, consumed less milk, red meat and sausages, and had higher satiety responsiveness and slowness in eating than boys (Table 2).
Table 1. Basic characteristics of Finnish children at baseline of the Physical Activity and Nutrition in Children study, n 487, in years 2007–2009
The values are means (standard deviations) from independent samples t tests for normally distributed variables and from Mann–Whitney’s U-tests for variables with skewed distributions as well as numbers (percentages) from Pearson’s χ2 test for categorical variables. P-values < 0·05 are bolded.
There were 487 children (234 girls, 253 boys) for all other variables than pubertal status (485 children, 233 girls, 252 boys), household income (479 children, 232 girls, 247 boys), tooth brushing (464 children, 223 girls, 241 boys) and xylitol chewing gum (462 children, 222 girls, 240 boys).
BMI-SDS, BMI-standard deviation score calculated using Finnish reference values(Reference Saari, Sankilampi and Hannila26); DMFT, Decayed, Missing and Filled Teeth; dmft, decayed, missing, and filled teeth; DMFS, Decayed, Missing and Filled Surface; dmfs, decayed missing and filled surfaces. Written in lowercase letters dmft/dmfs refers to primary dentition.
* Body weight status was defined using the International Obesity Task Force criteria corresponding to an adult BMI cut-point at 25 for overweight and 30 for obesity(Reference Dietz and Bellizzi27,Reference Cole, Bellizzi and Flegal28) .
† Pubertal status was defined using Tanner 5-stage criteria(Reference Marshall and Tanner29,Reference Marshall and Tanner30) .
‡ Includes DMFS, dmfs and initial caries surfaces in primary and permanent teeth.
Table 2. Eating frequency, diet quality, food consumption and eating behaviour at baseline
The values are means (standard deviations) from independent samples t tests for normally distributed variables and Mann–Whitney’s U-tests for variables with skewed distributions. P-values < 0·05 are bolded.
The numbers of children vary from 400 to 485 in different variables: 485 children (234 girls and 251 boys) for desire to drink; 484 children (234 girls and 250 boys) for slowness in eating; 483 children (233 girls and 250 boys) for enjoyment of food and food responsiveness; 480 children (233 girls and 247 boys) for emotional overeating; 479 children (232 girls and 247 boys) for satiety responsiveness; 473 (226 girls and 247 boys) for emotional undereating; 413 children (199 girls and 214 boys) for eating frequency and food consumption variables; 400 children (200 girls and 200 boys) for Baltic Sea Diet Score.
* Includes bread, cereal, porridge, flour, pasta and rice.
† Includes vegetables, nuts, roots, beans and mushrooms, excluding potato.
‡ Includes yoghurt, curdled milk, sour milk and quark.
§ Includes fish, shellfish and fish products.
|| Includes pork, beef, lamb, reindeer, game meat and sausages.
¶ Includes carbonated and non-carbonated beverages.
** Includes popcorn, crisps and salted nuts.
Associations of dietary factors with caries experience at baseline
BSDS and the consumption of high-fibre grain products and milk were inversely associated with caries experience adjusted for age and gender (Table 3, Model 1). The inverse association between BSDS and caries experience weakened and was no longer statistically significant and the inverse associations of high-fibre grain products and milk with caries experience remained similar after further adjustment for household income, BMI-SDS, the frequency of tooth brushing and the frequency of using xylitol-containing chewing gum (Table 3, Model 2).
Table 3. Associations of dietary factors with caries experience in children at baseline
Caries experience includes DMFS, dmfs and initial caries surfaces in primary and permanent teeth.
The values are standardised regression coefficients (β), their 95 % CI, and P-values from linear regression models by entering each variable first one by one into a model with age and gender (Model 1), second entering each variable one by one into a model with age, gender, household income, the frequency of tooth brushing and the frequency of using xylitol-containing chewing gum (Model 2), and third entering all these variables simultaneously into a stepwise model (Model 3) to study what would be the best combination of variables to be associated with caries experience according to model selection criteria (F-test). Standardised βs refer to how many standard deviations a dependent variable will change per standard deviation increase in the predictor variable. P-values < 0·05 are bolded.
* Data on variables were available for the following numbers of children in Model 1 and Model 2, respectively: eating frequency and food consumption variables 392 and 385; Baltic Sea Diet Score 382 and 377; desire to drink 461 and 454; slowness in eating 460 and 453; enjoyment of food and food responsiveness 459 and 452; food fussiness and emotional undereating 458 and 451; emotional overeating 456 and 449; satiety responsiveness 455 and 448. For Model 3, the data were available for 327 children.
† Includes bread, cereal, porridge, flour, pasta and rice.
‡ Includes vegetables, nuts, roots, beans and mushrooms, excluding potato.
§ Includes yoghurt, curdled milk, sour milk and quark.
|| Includes fish, shellfish and fish products.
¶ Includes pork, beef, lamb, reindeer, game meat and sausages.
** Includes carbonated and non-carbonated beverages.
†† Includes popcorn, crisps and salted nuts.
Desire to drink and emotional overeating were positively associated with caries experience adjusted for age and gender (Table 3, Model 1). The positive association between emotional overeating and caries experience weakened but remained statistically significant, and desire to drink was no longer associated with caries experience after further adjustment for household income, BMI-SDS, the frequency of tooth brushing and the frequency of using xylitol-containing chewing gum (Table 3, Model 2).
BSDS and the consumption of high-fibre grain products and milk were inversely associated, and the consumption of potatoes and emotional overeating were positively associated with caries experience when age, gender and all dietary factors were entered simultaneously into a stepwise model (Table 3, Model 3).
Associations of changes in dietary factors with a change in caries experience over 2 years
The mean (sd) of caries experience was 3·89 (5·15) for all 487 children at baseline and 3·82 (4·70) for all 406 children after 2 years. Increases in the number of snacks, slowness in eating, food fussiness and desire to drink were associated with an increase in caries experience, and increases in BSDS and enjoyment of food were associated with a decrease in caries experience over 2 years (Table 4).
Table 4. Associations of changes in dietary factors with a change in caries experience over 2 years
Caries experience includes DMFS, dmfs and initial caries surfaces in primary and permanent teeth. The values are fixed coefficients, their 95 % CI and P-values from generalised linear mixed-effects models, adjusted for the fixed effects of gender, study group and household income at both time points and the repeated effect of time. The random subject-specific intercept was used in the models. P-values < 0·05 are bolded.
Data on variables were available for the following numbers of children at baseline and 2-year follow-up, respectively: caries experience 487 and 406; main meals, snacks, BSDS and food consumption variables 423 and 389. Eating behaviour variables vary from 500 to 495 at baseline and 431 to 427 at 2-year follow-up.
* Includes bread, cereal, porridge, flour, pasta and rice.
† Includes vegetables, nuts, roots, beans and mushrooms, excluding potato.
‡ Includes yoghurt, curdled milk, sour milk and quark.
§ Includes fish, shellfish and fish products.
|| Includes pork, beef, lamb, reindeer, game meat and sausages
¶ Includes carbonated and non-carbonated beverages.
** Includes popcorn, crisps and salted nuts.
Discussion
This study showed that higher adherence to the Baltic Sea diet and a higher consumption of high-fibre grain products and milk were independently associated with lower caries experience in a general population of school-aged children. In contrast, a higher consumption of potatoes, higher desire to drink and higher emotional overeating were associated with higher caries experience. Consistent with these cross-sectional results, an increase in adherence to the Baltic Sea diet was associated with a decrease in caries experience and an increase in desire to drink was associated with an increase in caries experience over 2 years in a general population of school-aged children. Moreover, an increase in the enjoyment of food was associated with a decrease in caries experience and increases in snacking frequency, slowness in eating and food fussiness were associated with an increase in caries experience over 2 years.
To our knowledge, there are no previous published reports on the associations of overall diet quality assessed by BSDS, and several single dietary factors with caries experience in any age group. One explanation for our finding on the inverse association between BSDS and caries experience is that BSDS consists of several foods, such as fruits, berries, vegetables, high-fibre grain products, milk and fish, which are considered protective for dental health(Reference Moynihan7,Reference Gondivkar, Gadbail and Gondivkar9,Reference Moynihan and Petersen34,Reference Anderson, Rothballer and Altenburger35) , and that diet including these foods tends to restrict the consumption of high-sugar products, such as sweets and ice cream(Reference Kanerva, Kaartinen and Schwab10). Although BSDS does not include high-sugar products, a higher BSDS was associated with lower caries experience at baseline and with a decrease in caries experience over 2 years. A lower BSDS has been associated with higher abdominal adiposity in Finnish adults(Reference Kanerva, Kaartinen and Schwab11), and adiposity has been associated with caries in children and adolescents(Reference Modéer, Blomberg and Wondimu32,Reference Willerhausen, Blettner and Kasaj36) . Moreover, we found that the association between BSDS and caries experience was attenuated after controlling for adiposity. Our results are consistent with the observations of previous studies concerning the inverse association between overall diet quality and caries in children(Reference Kennedy, Ohls and Carlson37,Reference Nunn, Braunstein and Kaye38) . For example, children with a higher Healthy Eating Index, indicating a better overall diet quality based on healthy eating guidelines for Americans that recommend a diet rich in fruits, whole grains, milk and low on saturated fats and added sugars(Reference Kennedy, Ohls and Carlson37), were less likely to exhibit severe early childhood caries compared with children with a lower Healthy Eating Index(Reference Nunn, Braunstein and Kaye38). One’s diet consists of a wide variety of foods and drinks, and there are complex interactions and cumulative effects of multiple foods, drinks and nutrients within the diet. Therefore, the indices of overall diet quality, such as BSDS, could be useful tools when studying the association between diet and caries.
Our finding that the consumption of high-fibre grain products and milk were inversely associated with caries experience independent of various other dietary factors is in line with the results of previous studies(Reference Giacaman4,Reference Moynihan7,Reference Gondivkar, Gadbail and Gondivkar9,Reference Moynihan and Petersen34) . High-fibre foods stimulate saliva secretion, cleanse teeth, modify the composition of oral biofilm microbiota and encase cariogenic carbohydrates in a non-fermentable covering(Reference Anderson, Rothballer and Altenburger35,Reference Coogan, MacKeown and Galpin39,Reference Laine, Tolvanen and Pienihäkkinen40) . Milk is an excellent dietary source of calcium, proteins and phosphate, which all have been considered to have features that inhibit caries(Reference Moynihan7,Reference Lin, Lin and Hu8) . Milk proteins, such as casein, form stable calcium phosphate complexes on the tooth surfaces that promote remineralisation and prevent demineralisation in enamel(Reference Moynihan7,Reference Moynihan and Petersen34) . Moreover, milk, excluding organic milk, is in Finland fortified with vitamin D(Reference Piirainen, Laitinen and Isolauri41,Reference Soininen, Eloranta and Lindi42) , which has been suggested to reduce the risk of caries(Reference Hujoel43). However, the results of some studies suggest that a disaccharide lactose from milk could increase the risk of caries by acid production due to oral bacteria metabolism(Reference Signori, Hartwig and Silva-Júnior44). Studies suggest that more prolonged and frequent exposure to lactose from milk is warranted to increase caries risk(Reference Ricomini Filho, de Assis and Costa Oliveira45). The protective components present in milk, such as minerals and proteins, might overcome the adverse effects of lactose(Reference Ricomini Filho, de Assis and Costa Oliveira45,Reference Peres, Coppi and Volpato46) .
The direct association between potato consumption and caries experience in children was a surprising finding. Potato is a recommended dietary source of carbohydrate when served cooked or baked without added sources of fat, and starch is the predominant carbohydrate in potatoes(Reference Camire, Kubow and Donnelly47). Previous studies have found that highly processed starch products, such as potato chips, are associated with caries, but unprocessed starch seems to have little or no association with caries(Reference Giacaman4,Reference Hancock, Zinn and Schofield5,Reference Moynihan and Petersen34) . We observed no association between salty snacks, including potato chips, with caries experience in children. Interestingly, one study suggested that the cariogenic potential of sucrose increases when it is combined with starch(Reference Ribeiro, Tabchoury and del Bel Cury48). One possible explanation for our result may be that children who preferred potatoes consumed less high-fibre grain products, such as rice and pasta, that were inversely associated with caries experience. Further studies are warranted to confirm our surprising finding on the direct association between potato consumption and higher caries experience in children.
Our observations on the direct cross-sectional association between emotional overeating and caries experience and the direct longitudinal association between snacking and increased caries experience in children are potentially important. Our results are in accordance with the findings that emotional overeating was associated with caries in children aged 3–6 years(Reference Nembhwani and Winnier16) and that constant snacking predisposed to obesity and caries in children and adolescents(Reference Johansson, Holgerson and Kressin49,Reference Alm, Fåhraeus and Wendt50) . These observations suggest that children who have higher eating frequency and food consumption are at increased risk of not only overweight but also caries. Emotional stress had been found to stimulate appetite in some children but has been argued to inhibit appetite in non-obese children(Reference Braet and van Strien51). Therefore, more research is warranted to clarify the association between emotional overeating and caries in children.
We found that desire to drink was cross-sectionally associated with higher caries experience and longitudinally associated with an increase in caries experience over 2 years. This observation is in line with the results of few previous studies in children(Reference Shqair, dos Santos Motta and da Silva15,Reference Nembhwani and Winnier16) . Desire to drink does not sort out the type of drink but indicates food approach, and the association with caries may reflect the drink preferences of children. Other dietary factors partly explained the association between desire to drink and higher caries experience in our study. Earlier studies have shown that children prefer foods and drinks that are not consistent with a healthy diet and contain lots of sugar and fat(Reference Cooke and Wardle52,Reference Chung, Backholer and Zorbas53) . Children with higher desire to drink also prefer sugar-sweetened soft drinks and consume carbonated soft drinks frequently(Reference Sweetman, Wardle and Cooke54). However, it is interesting that we did not observe an association between the consumption of sugar-sweetened beverages and caries experience in children. It may be that the drinks have been consumed as part of a meal and therefore are not as cariogenic than those drunk between meals(Reference Gustafsson, Quensel and Lanke55).
In our study, food fussiness and slowness in eating had direct longitudinal associations with increased caries experience over 2 years, whereas enjoyment of food was associated with a decrease in caries experience. Food fussiness is common in childhood and has been defined as avoiding familiar and unfamiliar foods, leading to consuming an inadequate variety or quantity of foods(Reference Kutbi56). Slowness in eating represents the lack of enjoyment and interest in food, whereas enjoyment of food represents higher interest and responsiveness for food(Reference Wardle, Guthrie and Sanderson14). Similarly to the desire to drink, food fussiness may describe the food preferences of children towards cariogenic foods(Reference Cooke and Wardle52). Indeed, it has been shown that food fussiness is associated with a lower consumption of vegetables, cheese and meat in children, whereas enjoyment of food has been associated with a higher consumption of these foods(Reference Jalkanen, Lindi and Schwab24). It may be that children who enjoy food more and have less food fussiness may, in general, consume healthier foods and therefore have more favourable eating behaviour traits for oral health and preventing caries. One explanation for the association between slowness in eating and increased caries experience could be a more prolonged exposure time for demineralisation when eating slowly. These observations of our study are in line with those of previous studies(Reference Nembhwani and Winnier16,Reference Anandakrishna, Bhargav and Hegde57) . It is interesting that the intervention was not associated with a change in caries experience over 2 years, although we have previously shown that the intervention improved diet quality in children(Reference Lakka, Lintu and Väistö19,Reference Eloranta, Schwab and Venäläinen23,Reference Viitasalo, Eloranta and Lintu58) . However, 2 years is a relatively short time given the slow progression of caries(Reference Meyer-Lueckel and Paris59) and the complex interrelationships between the pathogenic and protective factors for caries progression(Reference Meyer-Lueckel and Paris59). The oral self-care habits of the children in our study were relatively good, and their caries prevalence was low. Protective factors such as frequent tooth brushing may partly explain the low caries experience in this study population. It may be that a more extended time period would be needed to examine whether the intervention is beneficial in preventing caries.
The strengths of our study include the relatively large representative population sample of children and the longitudinal study design. Oral health examinations were also performed by two experienced dentists. Moreover, we evaluated caries experience by calculating initial caries surfaces and the DMFS/dmfs index, one of the most common caries indices in epidemiological studies(Reference Pitts, Zero and Marsh2,Reference Moradi, Mohamadi Bolbanabad and Moinafshar60) . By including initial caries lesions in the analyses, we were able to assess caries status more precisely than by using the DMFS/dmfs index alone. In addition, initial caries lesions represent a considerable portion of children’s caries lesions in the Nordic countries(Reference André Kramer, Petzold and Hakeberg31,Reference Alm, Wendt and Koch61) . Another strength of our study is that we assessed eating behaviour using CEBQ, which is an internationally validated questionnaire(Reference Viana, Sinde and Saxton62). We also used food records to assess diet quality, food consumption and eating frequency. Our food records covered 4 days which offers higher reporting accuracy and provides more representative information on the diet than retrospective methods(Reference Crawford, Obarzanek and Morrison63) that are subject to recall bias. Food records have been stated as an accurate method for estimating food consumption in children(Reference Crawford, Obarzanek and Morrison63). Furthermore, parents were carefully instructed to fill out the food records that were examined and completed by a clinical nutritionist, if needed. However, the food records are also subject to inaccuracy in reporting food consumption and eating frequency and tendency not to follow a regular diet during reporting. For example, parents may have omitted the consumption of foods that are generally recognised as non-healthy. Our result may not be directly generalised to children other than those living in the Nordic countries because BSDS consists of foods typically grown and consumed in the Nordic countries, such as wholegrain rye, oat and barley. An essential point to consider is that children participated in a physical activity and dietary intervention study. Thus, the families involved in the study may have been more health conscious than the general population. Moreover, participation in the study may have affected our results on caries because the children and their caregivers in the intervention group were given individualised dietary counselling and the control group also received general verbal and written guidance on a healthy lifestyle. The PANIC study was not designed to study the effect of the lifestyle intervention on caries, and no difference was found in caries experience between the groups. However, we adjusted the data for the study group to control for possible confounding due to the intervention.
In conclusion, we demonstrated here that enjoyment of food, better overall diet quality and a higher consumption of high-fibre grain products and milk were associated with lower caries experience in a general population of school-aged children. Furthermore, higher snacking frequency, slowness in eating, food fussiness and desire to drink were associated with increased caries experience over 2 years in these children. Further research is needed to confirm these observations in other pediatric populations. If confirmed, these results could be utilised in the prevention of caries among children.
Acknowledgements
We are grateful to all children and their parents participating in the PANIC Study. We are also indebted to the research group and personnel of the PANIC study for their invaluable contribution throughout the study.
The PANIC Study has financially been supported by Ministry of Education and Culture of Finland, Ministry of Social Affairs and Health of Finland, Research Committee of the Kuopio University Hospital Catchment Area (State Research Funding), Finnish Innovation Fund Sitra, Social Insurance Institution of Finland, Finnish Cultural Foundation, Foundation for Paediatric Research, Diabetes Research Foundation in Finland, Finnish Foundation for Cardiovascular Research, Juho Vainio Foundation, Paavo Nurmi Foundation, Yrjö Jahnsson Foundation and the city of Kuopio. Moreover, the PhD students and postdoctoral researchers of The PANIC Study have been supported by Program for Clinical Research and Program for Health Sciences of Doctoral School of University of Eastern Finland, Finnish Doctoral Programs in Public Health, Päivikki and Sakari Sohlberg Foundation, Paulo Foundation, Jalmari and Rauha Ahokas Foundation, Aarne and Aili Turunen Foundation, Finnish Medical Foundation, Finnish Medical Society Duodecim, Jenny and Antti Wihuri Foundation, Kuopio Naturalists’ Society and Olvi Foundation. This work was supported by a personal grant from the Finnish Cultural Foundation, North Savo Regional Fund (V. F. V., 65211987). The funding sources had no role in the design, analysis or writing of this article.
V. F. V., A.-M. E., A. L. S., J. V., U. S. and T. A. L, contributed to the study conception and design. A. V. and T. I. conducted clinal oral health examinations. A.-M. E., A. V., T. I., J. V., M. N., U. S. and T. A. L. participated in the data collection. V. F. V., A.-M. E., A. L. S., J. V., S. M. and T. A. L. planned the statistical analyses. A.-M. E., A. L. S., J. V., M. M., H. T. V., J. L., M. N. and T. A. L. provided scientific advice for the conduction of the study. S. M. provided scientific advice for the conduction of the statistical analyses. V. F. V. conducted the statistical analyses, interpreted the results and drafted the manuscript. All authors critically revised the manuscript and approved the final version of the manuscript.
There are no conflicts of interest.
