The aim of this study was to assess differences in fruit and vegetable consumption and their potential determinants between two samples of 11-year-old schoolchildren in the Netherlands measured in 2003 and 2009. Overall, the results indicated that the mean fruit intake of the children measured in 2009 was higher than that of the children measured in 2003, independent of their gender, ethnicity and maternal level of education; however, vegetable intake was somewhat lower in 2009 than in 2003. Also more children from the 2009 cohort met the WHO recommendation. However, the observation that more children from the 2003 cohort met the Dutch recommendation for fruit intake may be a result of the different definitions. The Dutch recommendation states a minimum of two portions or pieces of fruit, whereas the WHO recommendation combines fruit and vegetable intake and uses total grams per day. Additional analyses revealed that the 2003 children reported a higher consumption of tangerines than did the 2009 cohort. Whereas one tangerine counted as one piece of fruit, in the calculations for grams per day it was considered to contribute 50 grams and most other popular fruits such as apples, pears and bananas were estimated to weigh 100 grams per piece. That the 2003 cohort ate more tangerines is in line with the season of measurement; in September and October these kinds of fruit are widely available in the Netherlands.
Regarding the presumed mediators, significantly more children in 2009 knew about the recommended fruit intake levels than in 2003, and some positive trends for other potential determinants of fruit and vegetable intakes were also found.
The finding that the schoolchildren in 2009 on average reported eating more fruit than in 2003 is encouraging. From the mediation analyses it appears that knowledge of the recommendation, parental demand and parental facilitation explains most of this difference. This may indicate that school programmes were indeed able to improve the children's knowledge of the recommendations for fruit intake and that this influenced their intake. It further suggests that the school programmes or other media activities were able to reach the parents, who subsequently changed their parenting practices regarding fruit intake. It is, however, surprising that liking or school availability could not explain the differences between the two samples. Most school-based programmes not only addressed the knowledge of the recommendations but also individual level determinants such as liking . Results also show that liking was most strongly associated with fruit intake, but also that the level of liking did not significantly differ between the samples. This may indicate that even if the 2009 sample was more exposed to fruit-promoting school programmes, these programmes were not able to positively influence the liking for fruit. Conversely, the school availability significantly differed between the two samples but was not related to fruit intake. This might be because a very small proportion of children in the 2009 sample reported the positive availability of fruit at school.
Our results regarding fruit intake are somewhat similar to the very few earlier studies that are available for approximately the same period of time. Rasmussen et al. showed that in Denmark fruit intake improved between 2002 and 2006 among 11-, 13- and 15-year-olds; in all age groups among girls the proportion which state to eat fruit at least once per day improved from between 52% and 58% in 2002 to between 62.9 and 69.8% in 2006 . The proportion of boys which state to eat fruit at least once per day improved from between 29.2 and 49.6% in 2002 to between 41.3 and 55.0% in 2006. Johnson and Hackett presented evidence for a positive trend in fruit and vegetable intake in Liverpool between 2000 and 2006 for 9-10-year-olds; the proportion of boys and girls reporting eating fruit on the previous day increased from 71.5% in 2000 to 76.8% in 2006 and from 70.7% in 2000 to 80.8% in 2006 respectively . However, no other studies published on potential underlying factors explaining the trends in fruit intake.
The results further indicate that 11-year-olds in 2009 did not eat more vegetables than 11-years-olds in 2003. On the contrary, the 2009 sample reported a lower vegetable intake than the 2003 sample. This finding is consistent with results from previous research, indicating that increasing intake levels of vegetables of children is more difficult than improving fruit intakes. Children have higher preferences for fruit , and in the Netherlands it might be even more difficult to improve vegetable intake by school-based promotion, as vegetable intake during school hours is uncommon and does not fit the normal eating patterns of most native Dutch families. Vegetables are part of the evening meal, but are rather uncommon at breakfast, lunch or in between meals . The potential mediators included in the current study could not explain the difference in vegetable intake between the two samples, but we found that parental demand and bringing vegetables to school had a suppressive effect on the difference in vegetable intake between the samples. This indicates that even though parental demand and bringing vegetables to school were significantly associated with the outcome variables in the expected direction, the so-called direct effect of the cohort variable on the outcome was stronger when these two variables were taken into account. These results suggest that there must be unmeasured factors that explain why the 2009 sample reported a lower vegetable intake than the 2003 sample. Therefore, future studies should include other potential mediators in order to inform future intervention strategies. One possibly important mediator may be feeding strategies, as a study by Zeinstra et al. showed that a feeding strategy in which children could make choice regarding when and what vegetables to eat, was positively associated with their vegetable intake . In addition, since school availability of vegetables was very low, an alternative strategy may be to improve availability and accessibility of vegetables at schools, for example by providing ready-to-eat vegetables as a snack in the morning breaks.
The fact that many factors that have been found to be significant correlates of fruit and vegetable intakes appeared to be more favourable in 2009 than in 2003 is encouraging. The fact in particular that in 2009 children more often agreed that there are fruits and vegetables available at their school may indicate that schools have changed their policies or have participated in programmes that facilitate availability and accessibility, such as the Schoolgruiten project . Analyses of differences in school policies between 2009 and 2003 will be conducted to gain more insight. Nevertheless, the proportion of children who agreed that fruit and vegetables were available at their school remained low in 2009 also, indicating that there is much room for further improvement.
There was one potential determinant that showed a negative trend: fewer children in 2009 than in 2003 reported that they were allowed to eat as much fruit and vegetables as they wanted. It may be that parents have become stricter in controlling their children's eating behaviours. It could also be a statistical artefact since in 2003 the scores were high and could only go down (regression to the mean), but this is mere speculation and should be further explored in additional research.
A strength of this study is that it used representative samples for the Netherlands, using the same validated methodologies , that were sensitive enough to detect changes in intake as well as in determinants . Another strength of the study is that effect modification by gender, maternal educational level and ethnic background was explored and that potential confounders were controlled in multiple regression analyses. Moreover, the current study did not solely explore differences in intake; it also included potential determinants of fruit and vegetable intake and was thus able to conduct mediation analyses. There are also some limitations to this study, however. First, fruit and vegetable intake might have been influenced by seasonal effects, because the Pro Children data were collected in autumn and the Pro Greens data in spring. Since seasons may influence the availability of fruit and vegetables, it is generally assumed that children eat more fruit and vegetables in summer and autumn compared with winter and spring . We explored this assumption in the available data from the control group of the Dutch Pro Children intervention study (n = 188) and indeed found that the children reported significantly higher intakes of fruit and vegetables in the September measurement compared with the May measurement (data not shown). The observation in the current study that the 2009 sample reported higher fruit intakes in May than the schoolchildren in October 2003 is not consistent with this proposed seasonal effect and may thus represent a real higher intake level in the 2009 sample compared with the 2003 sample. Furthermore, this seasonal influence is unlikely to apply to for most determinants, except for the availability of fruit and vegetables.
Second, the Pro Greens questionnaire slightly differed from the one used in Pro Children in the assessment of fruit intake, i.e. a separate question for berries was added. Although it was just a minor adaptation, a sensitivity analysis was conducted. There were doubts as to whether children might have thought that they had to state pieces of berries eaten instead of portions of berries eaten, because there were high intake numbers on this item. Therefore, under the assumption that the children reported pieces of berries instead of portions, the numbers were recalculated into portions eaten to check if it would make a difference in the number of children who reached the recommended intake. Results from these analyses showed that the reported differences between the two samples were not caused by this extra questionnaire item (data not shown).
A final limitation was the use of self-reported measures. Self-reported intake levels may be biased, but a previous study showed acceptable validity of the questionnaire used in the present study. Furthermore, any bias because of self reports is likely to be the same in the two study samples, and will therefore not be of limited influence on the comparison between the two samples.
Taking the limitations into account, the present study does provide an indication of the current situation compared with a period in which interventions and policies aimed at improving schoolchildren's fruit and vegetable intakes were not omnipresent. Although the effect size may not be clinically relevant for individuals, there is a likely relevance for public health, because many children can be reached by school-based interventions and small changes in a large proportion of the population can have an important impact on health indicators, as recently shown in an epidemiological modelling study . In the Netherlands 608 out of around 7000 primary schools reported participating in Schoolgruiten or similar fruit and vegetable promoting projects. The result of the current study might indicate that the increased attention to fruit and vegetable intakes in Dutch schools may have started to have a somewhat positive effect among 11-year-olds.