In this study we demonstrated a new methodology for measuring the eating context of young children and related it to fruit and vegetable consumption in a nationally representative sample of UK children. The results provide new insights into how fruit and/or vegetable consumption varies in different eating locations, with different people present, whether or not the TV is on and if eaten at a table. The results from this sample of children suggested that they were more likely to consume higher amounts of fruit and vegetables in structured and formal care settings, such as school and care outside of home than at home, and with carers and other children and their friends, than compared to the EOs taking place with parents alone. These associations were particularly strong for the younger age groups (1.5-3y and 4-6y). Fewer significant associations were seen for older children.
There are a number of possible explanations for the higher fruit and vegetable consumption seen in structured care and school settings. In 2005, the Department of Education and Skills produced guidelines to implement nutrition standards for lunches in nursery schools, which included a recommendation that an item of fruit and vegetables to be provided to each child at lunchtime each day . The Caroline Walker Trust published nutrient-based standards for children under 5 years in child care, which recommended that carers offer 4 to 5 different types of fruit and vegetables to children each day in day care settings . For school children, the School Meals Review Panel was appointed by the UK government in 2005 to introduce new standards for school food, whereby schools had to provide at least one portion of fruit and one portion of vegetables or salad at lunch per pupil per day . Furthermore, children aged 4-6y in England who attend local education authority (LEA) maintained schools are entitled to a free piece of fruit or vegetables each school day , which may explain the higher odds ratios for fruit (Q3) and vegetables (Q2) in this age group, which were not seen in the 7-10y age group. These changes in recommendations may well explain the increased consumption in school and formal care settings.
Older children (7-10y) had a greater tendency to eat with friends as well as alone compared to younger children, thus illustrating the gain in independence and the increasing importance of peers in eating and food choice as they become older [35, 36]. A number of intervention studies have shown that children have increased consumption of, or have expressed a liking of fruit and vegetables, due to the influence of their peers eating or liking fruit and vegetables [37–39]. Our observational study showed that children were consuming more fruit and vegetables when they were with friends than with parents, suggesting a possible effect of peer modelling on consumption of healthy foods. However, higher consumption with friends could also be confounded by the location of the eating occasions such as school or care settings. This is a speculation since we did not test the combined effects of the eating contexts on consumption. Having the TV on appeared to influence the odds of consuming vegetables, but was not associated with the quantity consumed. No strong associations were seen for fruit. Our findings only partially agree with previous studies, where fruit and vegetable consumption has been found to be inversely correlated with the frequency or number of hours of TV watching [22, 40, 41]. Our results are based on data for each eating occasion and may therefore capture differences in behaviour between fruit consumption and vegetable consumption, which may not be seen using questionnaires as in the other studies of the influence of television.
This study also exhibited new information on eating at the table and fruit and vegetable consumption, since this determinant has not been well researched. Children who ate at the table were much more likely to eat vegetables. This was not surprising, since previous studies have shown that family meal frequency is important in establishing positive eating behaviours such as eating fruit and vegetables, and family meals are more likely to occur around a table [42–45]. However, interesting findings were seen when quantifying the amount consumed when eating at the table. While the first part of the model suggested that children were more likely to consume fruit at the table, the second part of the model indicated large portions of fruit (>100g) were consumed when not at a table. We are unsure of the reason for this observation, but a possible explanation could be that the higher odds of fruit consumption in the first part of the model were driven by higher frequency of occasions consuming small amounts of fruit in composite dishes (such as fruit puddings and desserts) at the table; whilst results in the second part of the model might have been driven by fruit in large discrete portions, such as whole apples and bananas, with typical portion sizes of around 100g, that were likely to be eaten on the go and in any environment without being at the table. Unlike fruit, vegetables were less likely to be consumed in large quantities when not at the table. This is because vegetables are usually served as part of a main meal in substantial portions rather than eaten as snacks, and hence vegetables are more likely to be eaten at a table. These findings emphasise that eating fruit and eating vegetables are different behaviours.
There are a number of strengths of this study. It illustrates an alternative methodology for assessing the eating environment using a simple traditional dietary assessment tool – a food diary. The advantage of this method is that it is easy to use, and collects real time information without using advanced technology, which remains important for national surveys where there are different levels of technology access and expertise and the wide ranges of age and sociodemographic background among participants. The burden of collecting the information is also low, since it can be combined as part of an existing dietary assessment tool (food diary or 24 hour recall) without using a separate questionnaire or other psychometric measure. Previously, data on the food environment, such as television watching and screen time and family meal frequency have been collected via questionnaires, while dietary intake was assessed using separate dietary assessment method. Our method is likely to be more accurate since we have detailed dietary data that match with the eating contexts for each eating occasion. Other studies had investigated specific meals such as school lunches using direct observation methods; the method described here dispenses with interviewer burden as well as the discomfort for participants of being watched and monitored.
This study has its limitations however. Children’s fruit and vegetable consumption and the eating context information were reported by their parents. Previous studies have shown low agreements between children and parental reports of fruit and vegetable intakes [46–48], and also parents had to rely on information given by other carers if the children had been away from home. Hence, there could be errors induced during the recording process. Secondly, there were a number of “Not specified” responses when collecting information on the four eating context elements, particularly in answering whether the TV was on, and eating at the table. This may be because the children were in eating contexts where televisions or tables were not present, for instance, in outdoor areas or when eating on the go. These “Not specified” responses were treated as missing data because it would be bias to make assumptions on these data. The fact that only complete case analyses were performed may have resulted in reduced statistical power to detect differences amongst categories, as the number of eating occasions included in the analyses was lowered. Another limitation was that parents were only asked to record if TV was on at the occasion, it was unknown whether the children were actively watching TV during the meal and how much distraction from the TV might have influenced consumption. Moreover, NDNS did not collect data on other sociodemographic measures such as parental education that might have affected consumption. These factors were not adjusted for in the analysis and hence, there could be residual confounding in the results.
Furthermore, this study could not determine the direction of the effects between the eating context and fruit and vegetable consumption and if the effects were causal, because of the cross sectional design of the NDNS survey. For instance, it is possible that fruit and vegetable consumption is driven by the eating contexts, such as the location of the meal, but there may also be circumstances where eating contexts are driven by food choices or consumption. Further research should therefore be done longitudinally to allow clearer causality conclusions to be drawn. In addition, this new method of eating context assessment needs validation of its accuracy as we have only focused on one food group as dietary outcome, it should also be re-tested on other populations and sub-populations, as well as on other dietary assessment tools such as 24 hour recall. Relationships between the eating context and other dietary outcomes such as energy dense food and drinks should be studied as they may also be related to the eating environment. We intend to apply this assessment method to Year 3 and 4 of the rolling programme for validity as well as testing it on other age groups and food groups. Lastly, the four eating context elements in this study have been analysed separately and this limits the understanding the relative weight of each element and the combined effect of the eating contexts on overall consumption. Hence, for future work, these factors could be modelled in such a way that their association with the outcome as well as with one another can be determined.