The health and eating habits of Finns have improved over the past decades, but socio-economic differences in food consumption remain a serious public health challenge(Reference Kaikkonen, Mäki and Hakulinen-Viitanen1,Reference Mattila2) . Childhood is a particularly important period for preventing health inequalities in later life, since childhood living conditions and home environment influence children’s eating(Reference Zarnowiecki, Dollman and Parletta3), and dietary habits formed in childhood tend to track into adulthood(Reference Mikkilä, Räsänen and Raitakari4,Reference Puolakka, Pahkala and Laitinen5) . Socioeconomically disadvantaged children are at higher risk of having a less healthy diet, and there is a link between childhood socio-economic status (SES) and adolescent and adult health(Reference Cohen, Janicki-Deverts and Chen6). In high-income countries, lower SES is associated with a higher prevalence of childhood overweight and obesity(Reference Bammann, Gwozdz and Lanfer7,Reference Vazquez and Cubbin8) and higher risk of metabolic syndrome, impaired fasting glucose and type 2 diabetes in later life(Reference Puolakka, Pahkala and Laitinen9,Reference Tamayo, Christian and Rathmann10) .
Socio-economic differences in children’s dietary habits exist in the majority of European countries(Reference Fismen, Buoncristiano and Williams11). Earlier studies in Finland(Reference Eloranta, Lindi and Schwab12–Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14) and other parts of Europe(Reference Fernández-Alvira, Mouratidou and Bammann15–Reference Pinket, De Craemer and Huybrechts17) show fairly systematically that higher SES is associated with a healthier diet in children. A mother’s lower education has been associated with lower dietary quality and dietary diversity in children(Reference Pinket, De Craemer and Huybrechts17). Studies have shown that children of a lower SES background tend to consume more sweets and sugar-sweetened beverages(Reference Fismen, Buoncristiano and Williams11,Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14,Reference Bjelland, Brantsæter and Haugen18) and fewer fruits and vegetables (FV)(Reference Fismen, Buoncristiano and Williams11,Reference Koivuniemi, Gustafsson and Mäkelä13,Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14,Reference Bjelland, Brantsæter and Haugen18,Reference Papamichael, Karatzi and Mavrogianni19) than children with a higher SES background. Socio-economic differences have also been seen in children’s nutrient intake(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14,Reference Novaković, Cavelaars and Geelen20) . Most typically, studies have reported a lower intake of vitamins C and D(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14,Reference Novaković, Cavelaars and Geelen20) and fibre(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14) and a higher intake of saturated fat(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14) for the low SES group.
In Finland, the association between SES and dietary habits is well-established in adults(Reference Palosuo, Koskinen and Lahelma21) but less is known about small children. A study of 6- to 8-year-old children (n 424) in Eastern Finland showed that children with higher parental education were more likely to eat fish and fibre-rich bread but less likely to use soft margarine(Reference Eloranta, Lindi and Schwab12). In a recent study on 2- to 6-year-old Finnish children (n 766), a higher parental educational level (PEL) was associated with a better diet quality in the children, assessed by the Children’s Index of Diet Quality(Reference Koivuniemi, Gustafsson and Mäkelä13). This meant that they consumed more vegetables, fruits, berries, vegetable-oil-based spread and skimmed milk. Similar results have been obtained in older studies, where the consumption of vegetables, fat spreads, fish dishes and sugar-sweetened drinks differed between SES groups in favour of high SES(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14). However, studies reporting socio-economic differences in preschool-aged children in Finland, especially for nutrient intake, are scarce. The aim of this study was to examine the associations between PEL and both food consumption and nutrient intake in a sample of 3- to 6-year-old Finnish preschoolers.
Subject and methods
Study design and participants
We collected the data in the context of the DAGIS study (Increased Health and Wellbeing in Preschools) in Finland. The DAGIS study is a research project examining 3- to 6-year-olds’ energy balance-related behaviours and stress in the preschool setting, aiming at producing knowledge on children’s energy balance-related behaviours and improving health behaviours. The other main aim was to examine and diminish possible socio-economic differences in children’s energy balance-related behaviours. The study protocol and the sampling design have been described in detail elsewhere(Reference Lehto, Ray and Vepsäläinen22). The study was approved by the University of Helsinki Ethical Review Board in the Humanities and Social and Behavioural Sciences in February 2015 (Statement 6/2015).
The first phase of the DAGIS study consisted of a cross-sectional survey, conducted in eight municipalities in Southern and Western Finland during 2015–2016. We aimed to have municipalities with socioeconomically diverse populations and selected the municipalities based on following indicators: the Gini coefficient of the municipality and the proportion of single parents and people with a low education level. We contacted 169 preschools, of which sixty-seven did not wish to participate and sixteen were excluded because they did not meet the eligibility criteria (having at least one group consisting of 3- to 6-year-old children, providing early education only during the daytime, being Finnish or Swedish speaking and charging income-dependent fees). All families with children in a group of 3- to 6-year-olds were invited by an invitation letter distributed by preschool personnel. Parents provided written informed consent for a total of 983 children. Due to limited research resources, we then excluded preschools with a total parental consent rate less than 30 % in all groups. The final number of participating preschools was 66 (43 % of invited), and the number of children with parental consent was 892 (25 % of invited). From those participants, 864 children (24 % of invited) have at least some study data. Thus, that is considered to be the final number of participating children in the study.
Socio-economic status
Parents filled in a questionnaire concerning their level of education and other background information. The consenting parents reported the highest educational level for themselves and their partners living in the same household. The highest parental educational level (PEL) in the family was used as an SES indicator. The answer options for the question ‘What is your highest educational achievement?’ were (1) comprehensive school (primary and lower secondary school); (2) vocational school; (3) high school; (4) bachelor’s degree or college; (5) master’s degree and (6) licentiate/doctorate(23). A three-class variable was then formed and used in the analyses: low educational level (high school/vocational school/comprehensive school), middle educational level (bachelor’s degree or equivalent) and high educational level (master’s degree or higher).
The parents also reported the average net income of the household per month and the number of people belonging to the household. The relative income of the household was then calculated, taking into account the number of household members and their ages(24). For the analyses, the subjects were divided into thirds based on the relative income. The income thresholds by group were as follows: the lowest third (179–1894 €), the middle third (1895–2500 €) and the highest third (2501–5556 €).
Food record data
Data on children’s food consumption and nutrient intake were obtained with food records, collected both at home and at preschool. The three-day food records were collected between September 2015 and April 2016. Some of the families (n 292, 34 % of participants) kept an additional two-day food record between June and September 2016 in order to capture seasonal variation in the diet (Table 1).
Table 1. Characteristics of the study sample (n 811) according to parental educational level (PEL) in the DAGIS study
Categories for PEL: low PEL: high school, vocational school or lower education; middle PEL: bachelor’s degree or equivalent; high PEL: master’s degree or higher education.
* Net income, taking into account the number of people in the household(24).
Each participating family was sent a 3-day food record including a validated Children’s Food Picture Book to assist with portion size estimation(Reference Nissinen, Korkalo and Vepsäläinen25,Reference Nissinen, Sillanpää and Korkalo26) . The families were guided to fill the food record in exact given dates (two weekdays and one weekend day). The instruction was to record all foods, beverages that their child consumed during the recording days outside the preschool and describe foods as accurately as possible, either with exact brand and product names or by listing all the ingredients for composite dishes. The portion sizes were instructed to be estimated using the Children’s Food Picture Book, by weighing or by using household measures or package labels. At the same time, the preschool personnel were instructed to fill in a separate pre-coded food record for foods and drinks consumed at preschool. They also received the Children’s Food Picture Book and had the same instructions to estimate portion sizes. The parents also reported their child’s dietary supplement use in a separate questionnaire.
Research assistants checked the completed food records and, if necessary, contacted parents or preschool personnel to complete missing details. In the checking process, special attention was paid to consumption of FV and sugar-contained products. Some individual days had to be excluded due to the incomplete recording (unrealistically long pauses between consecutive meals), but all valid days were still included in the data. Thus, the data consisted of 1–5 food record days for each participant. The food data were entered by trained research assistants using AivoDiet dietary software (version 2.2.0.0, Mashie FoodTech Solutions Finland Oy, Turku, Finland). The software included the Fineli food composition database (release 16, 2013)(27) maintained by the Finnish Institute for Health and Welfare. We updated the database with the up-to-date values of vitamin-D-fortified food items and added new food items and recipes. The recipes received from the preschool’s food services were also added to the database. As the food composition database did not provide values for added sugar, we estimated the intake of added sugar by creating a formula to estimate the added sugar content for each food group containing added sugar. The process is described in more detail in another article(Reference Lehto, Ray and Vepsäläinen28). Finally, we checked the entered food data for outlying values in food consumption and nutrient intake, and any data entry errors observed were corrected.
Data processing and statistical analyses
We extracted the food consumption data from the dietary software in two different ways. The first included single food items (e.g. skimmed milk, margarine and apple) and mixed dishes (e.g. pizza and chicken soup). The second, hereafter called the ‘ingredient level’, included single food items and the ingredients of mixed dishes (e.g. minced meat soup was disaggregated into minced meat, potatoes, carrots, etc.). For reporting food consumption in food groups, all dishes and food items were categorised with other similar foods, forming a total of twelve main food groups and several sub-groups. The presented analyses include eleven main groups as well as selected sub-groups (online Supplementary Table 4). The main group miscellaneous (including food items such as mustard, soya sauce and other spices) was excluded from the analyses due to not having nutritional significance. Similarly, each food item at an ingredient level was categorised into main groups and sub-groups. For reporting nutrient intake and food consumption, the means were calculated based on average intake for each participant over their completed 1–5 food record days. The results include the intake only from foods – not dietary supplements. In this study, we do not report intake from dietary supplements, as the supplement use was collected by a different method and is not directly comparable with the food consumption data. Differences in nutrient intake and food consumption by PEL were assessed using a multilevel hierarchical linear model adjusted for energy intake. Three-level models were used to take into account the clustering of the data due to (1) preschool-based recruitment strategy (preschool-level) and (2) multiple participants from the same household (family-level). The highest PEL group was considered a reference group. Analyses were conducted using R Statistical Software RStudio version 2022·02·3 (R Studio, Inc., 2021) and R package ‘lme4’ for linear model.
Sensitivity analyses
Families with low PEL had kept fewer additional food records (days 4 and 5) (Table 1), which may affect the results, especially the consumption of FV. Thus, we did sensitivity analyses with the core food groups and nutrients, including only 1–3 food record days from each participant. Because the results were similar to the one presented, we decided to keep the additional days included in the analyses.
Results
Participants
The sample consisted of 811 (94 % of the participants) children, from whom we received at least a one-day food record and information about PEL. The descriptive characteristics of the studied sample are provided in Table 1. Of the participants, 52 % were boys and most of the children were 4 (36 %) or 5 (35 %) years old. The majority of the participants (95 %) had at least three days of food record data available.
Food consumption
The associations between PEL and children’s food consumption are presented in Table 2, and the mean consumption of the main food groups and selected sub-groups according to PEL are presented in online Supplementary Table 1. Significant differences between PEL groups were found. Compared with high PEL, low PEL was associated with a child’s lower consumption of fresh vegetables and salads, vegetarian dishes, berries, white bread, blended spread (mixture of vegetable and animal fats), skimmed milk and ice cream but higher consumption of milk (fat 1–1·5 %), dairy-based desserts and sugar-sweetened soft drinks. Children in middle-educated families also had lower consumption of fresh vegetables and vegetable salads, vegetarian dishes and skimmed milk compared with children in higher educated families. No PEL differences were found in the consumption of sweets and chocolate or sugar-sweetened juice.
Table 2. Association between parental educational level (PEL) and children’s daily food consumption in the DAGIS study, hierarchical linear model adjusted for energy intake. Main food groups (bold) and selected sub-groups are presented
Categories for PEL: low PEL: high school, vocational school or lower education; middle PEL: bachelor’s degree or equivalent; high PEL: master’s degree or higher education.
At an ingredient level, compared with high PEL, low PEL was associated with the lower consumption of root vegetables and other vegetables, nuts and seeds, berries and fish and seafood but a higher intake of red meat (Table 3). The lower consumption of root vegetables and other vegetables was also seen in children in middle-educated families. The mean consumption of the selected food groups at an ingredient level is shown in online Supplementary Table 2.
Table 3. Association between parental educational level (PEL) and children’s daily food consumption (ingredient level) in the DAGIS study, hierarchical linear model adjusted for energy intake. Selected food groups are presented
CI: Confidence interval.
Categories for PEL: low PEL: high school, vocational school or lower education; middle PEL: bachelor’s degree or equivalent; high PEL: master’s degree or higher education.
Nutrient intake
Table 4 shows the association between PEL and children’s nutrient intake, whereas the mean intake is presented in online Supplementary Table 3. There were no differences in energy intake between PEL groups. Compared with high PEL, low parental education was associated with a lower intake of protein; fibre; EPA; DHA; vitamins C, D, B6 and B12; folate; riboflavin; potassium; Ca; phosphorous; Mg; Zn and iodine but a higher intake of fat, SAFA, trans fatty acids and MUFA. Children in middle educated families also had a lower intake of protein, fibre, vitamin B6, folate, potassium, Mg and Zn but a higher intake of SAFA compared with children in the highest educated families. There were no differences in added sugar or Na intake between the groups.
Table 4. Association between parental educational level (PEL) and children’s daily nutrient intake from food sources in the DAGIS study, hierarchical linear model adjusted for energy intake
Categories for PEL: low PEL: high school, vocational school or lower education, middle PEL: bachelor’s degree or equivalent, high PEL: master’s degree or higher education.
Discussion
The present study examined socio-economic differences in food consumption and nutrient intake according to PEL among Finnish preschool children. Our study produced detailed information at three levels: food group level, ingredient level and nutrient level.
Based on our results, there were significant differences between PEL groups. In general, compared with high PEL, low PEL was associated with a child’s lower consumption of healthy food groups, such as fresh vegetables, berries and skimmed milk, and a lower intake of many beneficial nutrients. There were a few exceptions, such as higher consumption of white bread and ice cream in the higher educated group. Our main findings are in line with previous studies. The majority of studies have shown that higher PEL is related to a healthier diet in children(Reference Eloranta, Lindi and Schwab12–Reference Fernández-Alvira, Mouratidou and Bammann15,Reference Pinket, De Craemer and Huybrechts17,Reference Manios, Kourlaba and Kondaki29) .
Several studies have observed the positive association between SES and the consumption of fruit, vegetables or berries(Reference Koivuniemi, Gustafsson and Mäkelä13,Reference Fernández-Alvira, Mouratidou and Bammann15,Reference Papamichael, Karatzi and Mavrogianni19,Reference Manios, Kourlaba and Kondaki29–Reference Lehto, Ray and Te Velde31) In our study, the consumption of fresh vegetables and vegetable salads and vegetarian dishes was lower in children with low or middle educated parents compared with high PEL. The low PEL group also had lower consumption of berries and nuts and seeds. The use of legumes and plant-based proteins was low in our data, and there were no differences in the consumption between the groups. Typically, children’s vegetable consumption is lower than their fruit consumption, since children tend to prefer sweet and avoid bitter flavours(Reference Forestell32). Higher fruit consumption was also observed in our data, and no differences between the PEL groups were observed in relation to it. In our study, the average consumption of fruit, vegetables and berries was nearest the recommendation (at least 250 g/d) in the high PEL group (248 g/d) but further away from the recommendation in the middle and low PEL groups (231 and 209 g/d, respectively). Our findings on nutrient intake are in line with the observed patterns of FV consumption. Compared with high PEL, middle PEL group had lower intake of fibre, folate and potassium, and the low PEL group also had lower intake of vitamin C. FV are important sources of these nutrients, and epidemiological evidence suggests that the consumption of FV is associated with reduced risk of CVD, cancer and all-cause mortality(Reference Aune, Giovannucci and Boffetta33). Therefore, our findings raise concern on the possible long-term health effects of less healthy food consumption among Finnish families with lower education levels.
It has been shown in many Western countries that lower SES groups tend to consume red and processed meat more often and in higher quantities(Reference Clonan, Roberts and Holdsworth34). The same was observed in our study, where children with low PEL had higher consumption of red meat than those in the high PEL group. The result is also in line with a study of 3- to 10-year-old French children (n 574) that found that children from lower educated families consumed more meat compared with those with higher education (81·1 v. 69·5 g/d)(Reference Manios, Kourlaba and Kondaki29). People in higher SES groups may consume less meat because of the greater awareness of the health effects associated with overconsumption of meat(Reference Clonan, Roberts and Holdsworth34). Instead of meat, they may prefer other more beneficial food groups, such as fish. This was seen in our study, where fish consumption was higher in children from higher educated families. The same has been observed in an earlier Finnish study examining children aged 3 and 6 years old, which found that children with higher educated fathers consumed more fish dishes(Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14). Another Finnish study reported that recommended fish consumption was more common in children in the highest PEL group(Reference Eloranta, Lindi and Schwab12). We also observed that the children in the low PEL group had a lower intake of vitamin D, EPA, DHA and iodine, the source of which fish is one of the most significant in children’s diet(Reference Korkalo, Nissinen and Skaffari35). The average intake of vitamin D (from food sources) has been below recommendations in Finnish children(Reference Eloranta, Lindi and Schwab12,Reference Kyttälä, Erkkola and Kronberg-Kippilä36) . In our data, the average intake was still below the recommended amount (9·1 µg) but highest among the high PEL group (9·6 µg). This supports the need for vitamin D supplementation in preschool-aged children.
We found differences in types of milk consumed. Parents’ lower education was associated with a child’s lower consumption of skimmed milk but higher consumption of a higher fat milk (fat content 1–1·5 %) compared to high PEL. Children in the middle PEL group also consumed less skimmed milk than the high PEL group. These results are in agreement with previous Finnish studies reporting an association between higher parental education and consumption of skimmed milk in children(Reference Kaikkonen, Mäki and Hakulinen-Viitanen1,Reference Eloranta, Lindi and Schwab12,Reference Kyttälä, Ovaskainen and Kronberg-Kippilä14) . The type of milk may partly explain the higher intake of SAFA that was observed in the low PEL group, since milk products are the main sources of SAFA in preschool children(Reference Korkalo, Nissinen and Skaffari35). These differences should be addressed because studies have shown that the intake of SAFA among Finnish children is higher than recommended(Reference Eloranta, Lindi and Schwab12,Reference Korkalo, Nissinen and Skaffari35,Reference Kyttälä, Erkkola and Kronberg-Kippilä36) , while a diet low in saturated fat is recommended because of the health effects(Reference Te Morenga and Montez37). Replacing SAFA with PUFA in children’s diets may help to reduce blood LDL cholesterol and the risk of cardiovascular disease later in life(Reference Wu, Micha and Mozaffarian38).
In our data, the average consumption of sugar-sweetened beverages was relatively low (12 g/d), but the intake was higher in children with lower educated parents. The same has been observed in previous studies(Reference Fernández-Alvira, Mouratidou and Bammann15,Reference Pinket, De Craemer and Huybrechts17,Reference Papamichael, Karatzi and Mavrogianni19) . A study of 3·5- to 5·5-year-old preschoolers (n 7063) from six European countries found that preschoolers with less educated mothers consumed energy-dense and low-nutritious food, such as sugared beverages, more often(Reference Pinket, De Craemer and Huybrechts17). Another study of 2- to 9-year-old European children (n 14 426) reported that children in the low and medium PEL groups had higher odds of more frequently consuming sugary beverages(Reference Fernández-Alvira, Mouratidou and Bammann15). Another study of European children (n 12 041), including children in Finland, found that 5- to 12-year-old children with a high SES vulnerability score were more likely to consume soft drinks(Reference Papamichael, Karatzi and Mavrogianni19). The consumption of sugar-sweetened soft drinks is associated with unfavourable health outcomes, such as increased risk of obesity(Reference Evans39,Reference Malik, Pan and Willett40) , and thus, the results are worrying. Despite the differences in consumption of sugar-sweetened beverages, we did not find differences in children’s added sugar intake. In our data, the intake of added sugar was in line with recommendations(Reference Clonan, Roberts and Holdsworth34), whereas in some previous studies, the intake was higher(Reference Erkkola, Kronberg-Kippilä and Kyttälä41). This may be partly explained by the fact that the food industry has introduced fewer sugar-containing alternatives to yoghurts, for example, and parents’ awareness of a healthy diet has increased.
Observed differences in children’s diets can be partly explained by the better nutrition knowledge of higher educated parents(Reference Zarnowiecki, Dollman and Parletta3). Education is a commonly used SES indicator, and parental education is reflected in the child’s situation(Reference Galobardes, Shaw and Lawlor42). Education expresses mostly non-material resources such as knowledge, skills, attitudes and values that may support health behaviours. In studies, higher SES is consistently positively associated with nutrition knowledge, parent modelling, home food availability and accessibility(Reference Zarnowiecki, Dollman and Parletta3). They may value healthy eating more and aim to eat according to the recommendations, such as preferring vegetables and low-fat dairy products and reducing the intake of red meat. Higher educated families may also have better material resources to eat healthy, since higher education partly explains higher income(Reference O’Neill, Tabish and Welch43). Our previous study showed that lower family income was associated with a less healthy diet in children(Reference Skaffari, Erkkola and Korkalo44). Typically, energy dense but otherwise lower nutritional density nutrient-depleted foods are low cost, while fish, fruits and vegetables, for example, are among the most expensive food groups(Reference Darmon and Drewnowski45). In this study, higher educated families also had higher incomes (Table 1), which may partly explain the observed differences.
The major strength of this study included the relatively large study sample and comprehensive food record data, as we managed to collect dietary data from over 800 children in eight municipalities in different parts of Finland. In comparison with FFQs used in most studies, food records provide more detailed information on food consumption and nutrient intake(Reference Slimani, Freisling, Illner, Lovegrove, Hodson and Sharma46). To the best of our knowledge, reporting food consumption on ingredient level is rarely used, even though it allows more accurate analysis of, for example, FV consumption, as the ingredient-level data includes FV in dishes (e.g. carrot in minced meat soup) as well as separately eaten (e.g. apple).
Additionally, in our study, food records were filled in both at home and at preschool, providing more accurate information on a child’s diet compared with an assessment made by the parent alone.
Possible limitations of this study were the low participation rate in the DAGIS cross-sectional survey (24 %) and the fact that the education level of the participating parents was higher than in general in Finland, where 43 % of 35- to 39-year-olds have at least a bachelor’s degree (in our data 78 %)(47). Although the sample was socioeconomically biased, differences in the children’s diet were seen. Thus, the actual differences may be even greater than observed in our study. We only reported nutrient intake from food, and thus, the results could have been slightly different if intake from dietary supplements had been taken into account. Few studies reporting children’s supplement use have shown an association between higher PEL and more frequent use of dietary supplements(Reference Sichert-Hellert and Kersting48,Reference Sicińska, Pietruszka and Januszko49) . In Finland, vitamin D supplementation is recommended for all children under 18 years old, but the prevalence of supplementation is not monitored. However, in this sample of 3- to 6-year-old pre-schoolers, 83% had used a supplement containing vitamin D during the previous month, and the use was associated with a higher household income (Master’s thesis,(Reference Kinnunen50)). The use of other dietary supplements was clearly lower. The results of this study are only generalisable to children attending day care, since the diet of the Finnish children attending day care outside the home seems to be more balanced and closer to the national nutrition recommendations than the diet of children being cared for at home(Reference Lehtisalo, Erkkola and Tapanainen51). It is also shown that Finnish preschool meals provide a significant proportion of many important nutrients and recommended food groups(Reference Korkalo, Nissinen and Skaffari35), so it is possible that preschool meals might diminish SES differences in children’s diet. The food record as a method involves certain weaknesses, such as possible under-reporting(Reference Shim, Oh and Kim52). During the recording days, parents may offer healthier food than usual or forget to record some dishes. Generally, foods that are considered unhealthy are more likely to be under-reported, whereas those considered healthy are more likely to be over-reported(Reference Macdiarmid and Blundell53). There is no consistent evidence of the impact of education level on reporting.
In conclusion, our study showed that compared with high PEL, low PEL was associated with a less healthy diet and lower nutrient intake in Finnish pre-schoolers. The observed diet-related disparities highlight the need for policy actions and interventions supporting healthy eating patterns such as high consumption of vegetables, nuts and berries in childhood, paying special attention to those with low PEL.
Acknowledgements
The authors thank the preschools, the preschool personnel and the parents for their participation in the DAGIS study and the research staff for data collection. In addition, the authors thank the collaborating partners of the DAGIS study for providing assistance in designing the DAGIS study.
The DAGIS project and this study were financially supported by Folkhälsan, University of Helsinki, The Ministry of Education and Culture in Finland, The Ministry of Social Affairs and Health, The Academy of Finland (Grants: 285 439, 287 288 and 288 038), the Juho Vainio Foundation, the Signe and Ane Gyllenberg Foundation, the Finnish Cultural Foundation/South Ostrobothnia, the Päivikki and Sakari Sohlberg Foundation, the Medicinska Föreningen Liv on Hälsa and the Finnish Food Research Foundation.
E. S., H. V., E. R. and L. K. formulated the research questions. E. S., H. V., K. N., E. L., R. L., E. R., M. E. and L. K. contributed to the planning and design of the study. E. S., H. V., K. N., EL, RL and LK participated in the data collection and/or were involved in processing data. ES analysed the data and performed the statistical analyses. E. S. wrote the manuscript, and H. V., K. N., E. L,. R. L,. E. R., M. E. and L. K. reviewed the manuscript. E. R. and M. E. were responsible for obtaining funding for the study. All authors reviewed and approved the final version of the manuscript.
L. K. was a board member of the company TwoDads at the time of the study.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114523001460
