We studied the changes and tracking in intakes of fruit, vegetables and SSB in a group of Norwegian children at three time points (18 months, 36 months and 7 years), and examined the association between maternal education and dietary tracking in the same group. The largest changes were seen for the intake of fruit and SSB from age 18 months to 7 years of age, while fair to moderate tracking coefficients for the intake of fruit, vegetables and SSB were found. Close to 50% of the children assigned to low, medium and high frequency of consumption at 18 months remained in the same group at age 36 months and 7 years. Children of mothers with low education were more likely to have a less frequent consumption of fruit and vegetables and a more frequent consumption of SSB, compared to children of mothers with high education at 18 months of age. Children of mothers with a high education level had lower odds for increasing fruit intake or decreasing SSB intake compared to children with a stable intake having mothers with a lower education level.
Comparison of children in the present study sample with all children in the 18 months sample showed significant differences for the dietary behaviours and maternal education. The level of maternal education in Norway has increased during the recruitment period (1999–2008)
. Moreover, healthier eating habits among preschool children have been associated with higher maternal education
[18–24]. Taken together, this may explain the differences in the characteristics of the dietary behaviours and the maternal education.
The median intake frequencies of fruit and vegetables for both genders were below the national recommendations for all time points, while the consumption frequency of SSB was relatively low. Relative high proportions of stability were seen for the intakes of SSB in both genders, as found among Norwegian 11–13 year olds as well
. From 18 months to 36 months, 50% of the boys and girls increased their intake of fruit and 40% decreased their intake of vegetables. These findings may be affected by the number of questions and different frequencies used in the questionnaires
. In the present study there was one question for fruit at both 18 months of age and 36 months of age, but the frequency was higher at 36 months. This may have resulted in an overestimation of intake at 36 months of age (compared to 18 months of age), meaning that the proportion of children that actually increased their intake of fruit between the two time points is a bit lower than 50%. The same could be the case for the intake of vegetables, but in the opposite direction. The number of questions about vegetables was reduced from three to two between 18 months of age and 36 months of age, and the frequency was lower at 36 months of age. This may have resulted in an underestimation of intake at 36 months of age (compared to 18 months of age), meaning that the proportion of children that actually decreased their intake of vegetables between the two time points is a bit lower that 40%. However, our findings for the changes between 18 months and 36 months/7 years of age are in line with other studies that report percentages around 50% of children who remained in the same percentile/group during a certain period
The number of studies investigating tracking in the childhood period is limited. Our results from Norway are in line with the findings in previous studies, even if they are not directly comparable; fair to moderate tracking was found both from 18 months to 36 months and from 18 months to 7 years. Stein et al.
 reported fair to moderate tracking, as estimated by agreement of classification within quintiles of intake, of energy, fat, cholesterol, protein, carbohydrates, sodium, potassium and calcium in 3–5 year old children over a 19 month period of follow-up. Boulton et al.
 found that tracking of energy and fat intake became more stable from 2 years of age in 106 children recruited at birth and followed prospectively until 15 years of age. Children with large energy intakes remained big eaters while children with low food intake became evenly spread across the distribution curve over time. Singer et al.
 followed 95 children for 6 years, covering three age periods (3–4 years, 5–6 years and 7–8 years), and concluded that tracking of nutrient intake begins as young as 3–4 years of age. Moreover, extreme intakes tended to sustain over time. Robinson et al. reported tracking of dietary patterns characterised as “infant guidelines” (high consumption of fruit, vegetables and home prepared foods) and “adult foods” (high consumption of bread, snacks, biscuits and chips) between children at 6 and 12 months of age, suggesting stability in eating habits that persist beyond infancy. When assessing the stability of dietary patterns like “processed”, “traditional” and “health conscious” in children at 3, 4, 7 and 9 years of age, Northstone and Emmet
 found weighted ĸ for quintiles of dietary pattern scores in the range of 0.31 and 0.38 between 3 and 7 years of age. Moreover, the finding that children assigned to low, medium and high frequency of consumption stayed in the same group over time has also been found among Norwegian 11–13 year olds
 and 14–21 year olds
. Finally, Boulton et al.
 found a similar pattern and level of tracking among males and females in intake of energy, fat and calcium intake in 2–15 year olds.
Gender differences in frequency among the 7 year olds were observed in our sample for fruit, vegetables and SSB. No significant differences were found between boys and girls in intake of fruit, vegetables and SSB among the 4 year olds in the Norwegian Ungkost study from 2000
. Gender differences in consumption of fruit, vegetables and SSB were larger in samples of Norwegian 10–13 years olds
[14, 15] indicating that girls start to eat healthier than boys already during early childhood. Additionally, previous research has found that girls compared to boys have a greater liking for and consumption of fruits and vegetables, while boys give higher ratings to fatty and sugary foods
An important factor related to children’s dietary habits is maternal education. Results from previous studies are in concordance with our findings, suggesting that children of mothers with low education consumed fruit and vegetables less often and SSB more often compared to children of mothers with high education at 18 months of age. Several European studies of young children (from 6 months of age) have reported that lower maternal education is associated with a less healthy diet in the children compared to children who have mothers with higher education
[18–24, 39]. Finally, Vereecken et al.
 found differences by educational level in children's and mothers' consumption frequencies of fruit, vegetables and soft drinks, and in the use of restrictions, verbal praise, negotiation, discouragement of sweets and restraining from negative modelling behaviour. Differences in children's food consumption by mothers' educational level were completely explained by mother's consumption and other food parenting practices for fruit and vegetables but not for soft drinks.
Finding that children of mothers with a high education level had lower odds for increasing fruit intake or decreasing SSB intake over time may be a consequence of limited possibilities for change, due to an already high consumption of fruit and a less frequent intake of SSB compared to those with a stable intake and having mothers of low education. Bere et al.
 have reported that Norwegian adolescents of parents with higher education had a higher intake of fruit and vegetables, greater access to and preference for fruit and vegetables, greater knowledge of national recommendations, stronger intentions to eat 5-a-day and stronger role models. Their results support our findings. According to our study, the potential for improved intake of vegetables seems to be the same regardless of maternal education level.
In a public health perspective, our results indicate two main challenges; how to improve dietary habits among children of mothers with low education and how to maintain healthy dietary behaviours in children of mothers with high education during early childhood. Targeting mothers, and first-time mothers in particular, in nutrition education interventions has the potential to impact the dietary behaviours of young children indirectly and the mothers’ diet directly
[25, 26]. The interventions should target nutrition knowledge, attitudes, strategies and increase the awareness related to role modelling, regulation of unhealthy dietary habits and encouragement of healthy dietary behaviours
[25, 26, 41].
Strengths of this study were the longitudinal study design based on a large national representative sample of children, and the use of multiple methods to describe tracking patterns over time. Both healthy and unhealthy dietary behaviours were studied, giving the opportunity to look at different dietary behaviours of young children. The main limitation of the study is the different variables and frequencies used at each time point to collect information about intake. The variances in intakes are small due to the large sample and the categories used. Analyses revealed only minor differences in the dietary behaviours at 18 months of age and maternal educational levels between the included children and the total sample at 18 months of age. Additionally, those with the most unhealthy dietary behaviours and lowest education were well represented in the sample included in the present study.
Finally, regression towards the mean as observed in the analyses presented in Figure
1, showed a decrease in frequencies among high consumers. This is the phenomena whereby the same variable is measured on two or more occasions, cases that are extreme on the first occasion will be somewhat less extreme on the second and third occasion