Teaching our children when to eat: how parental feeding practices inform the development of emotional eating. A longitudinal experimental design.

2 Background: Emotional eating in children has been related to the consumption of 3 high energy dense foods and obesity, but the development of emotional eating in young 4 children is poorly understood. Objective: The objective was to evaluate whether emotional 5 eating can be induced in 5-7 year old children in the laboratory and to assess whether parental 6 use of overly controlling feeding practices at 3-5 years predicts a greater subsequent tendency 7 for children to eat under conditions of mild stress at ages 5-7. Design: Forty-one parent-child 8 dyads were recruited to participate in this longitudinal study which involved parents and 9 children being observed consuming a standard lunch, completing questionnaire measures of 10 parental feeding practices, participating in a research procedure to induce child emotion (or a 11 control procedure) and observing child consumption of snack foods. Results: Children at 12 ages 5-7 exposed to a mild emotional stressor consumed significantly more calories from 13 snack foods in the absence of hunger compared to children in a control group. Parents who 14 reported using more food as a reward and restriction of food for health reasons with their 15 children at ages 3-5 were more likely to have children who ate more under conditions of 16 negative emotion at ages 5-7. Conclusion : Parents who overly control children’s food intake 17 may unintentionally teach children to rely on palatable foods to cope with negative emotions. 18 Further research is needed to evaluate the implications of these findings for child food intake 19 and weight outside of the laboratory setting. 20 21 22


Introduction
Emotional eating can be defined as "eating in response to a range of negative 27 emotions such as anxiety, depression, anger and loneliness to cope with negative affect" (1). 28 In adults and adolescents emotional eating has been linked to heavier weight, obesity, and 29 greater consumption of high energy-dense sweet and salty foods (2, 3, 4 although see 5 for 30 conflicting results). In younger children, around one quarter of parents of 5 year olds report 31 that their children exhibit emotional disinhibition with food (6) and 63% of children aged 5-32 13 report eating in response to mood (7). Parents of 2-6 year olds tend to report great 33 emotional under-eating rather than over-eating (8).Van Strien and Oosterveld (9) suggest that 34 young children lose their appetite as a natural response to stress associated with a loss of gut 35 activity (10). Emotional overeating may be a learned abnormality, likely to exacerbate ill 36 health. Given that eating behaviours track across life, understanding the development of this 37 behaviour is critical (11,12). 38 Parental feeding practices have been shown to impact upon children's developing 39 eating behaviours (13,14,15). Previous research suggests that parental use of food as an 40 emotional tool may teach children to use food to alleviate or distract from negative emotion 41 (see 16,17). Other feeding practices which overly control children's food intake have also 42 been shown to predict unhealthy eating behaviours as they are believed to undermine 43 children's ability to regulate their hunger and satiety. Parental use of food as a reward or for 44 emotional reasons has been correlated with emotional undereating in 3-6 year old children 45 (13), and pressuring children to eat has been shown to predict greater emotional eating in 5 46 and 7-12 year olds respectively (6,18). However, most of this research is cross-sectional and 47 utilises parent report of feeding practices and eating behaviours (13,15), making it difficult to 48 establish causality. 49 The development of emotional eating Previously Blissett et al. (16) developed a novel procedure for inducing child emotion 50 in a laboratory setting, however their original sample was reduced significantly because many 51 3-5 year olds failed to respond to the emotion manipulation. The first aim of this study is to 52 refine this emotional manipulation to assess its efficacy in a sample of 5-7 year olds, who 53 may be more capable of accurately rating their mood. The second aim is to assess whether 54 children aged 5-7 are more prone to eat in the absence of hunger when under conditions of 55 mild stress in comparison to a control group. The final aim is to ascertain whether parents 56 who report using high levels of controlling feeding practices with their 3-5 year old children 57 have children who are more likely to emotionally overeat when they experience stress in a 58 laboratory setting 2 years later.

61
Participants 62 A small sample of 41 parent-child dyads participated in this longitudinal study which 63 was part of a larger study described elsewhere (16). Participants were recruited to this study 64 via advertisements to parents in the East-Midlands area of the UK. Families were eligible to 65 participate if they had a child aged between 2 and 5 years old with no medical conditions 66 affecting eating or feeding. Families were followed up two years later (time point 2).

67
Following data screening, we removed six families from the original dataset for a variety of   slice of cheese, 4 cheese crackers, 3 pieces of chopped apple, 5 carrot sticks and 2 chocolate 90 chip cookies. Mothers' lunches were the same but slightly larger as they included 2 bread 91 rolls, 2 slices of chicken and 2 slices of cheese. Where mothers indicated that they or their 92 child was vegetarian, chicken was not offered and was replaced with additional cheese.

93
Mothers and children were each provided with a drink of water with their meal. Mothers and 94 children were asked to eat from their own plates until they felt full and could ask for 95 additional food if they wished (no families requested more food). Once they had finished 96 eating, mothers and children each took part in separate procedures within the same room.

97
Mothers were asked to complete a series of questionnaires whilst the child engaged in a task 98 The development of emotional eating with the researcher. Children were randomly allocated to either the experimental mood 99 manipulation group, or to the control group, with children being in the same group that they 100 had been involved in at time point 1. Children could not see their mothers but mothers could 101 watch their child through a screen if they wanted to. After the procedures had finished all 102 mothers and children were weighed and measured in light indoor clothes without shoes.

103
Weight and height scores were converted to BMI scores for mothers and to age and gender-104 adjusted BMI SDS scores for children (20). shown to discriminate mood in children in this age range (16,21). Children were then shown 109 a selection of small toys and invited to choose one which they were told they would receive 110 on completion of a colouring task. The toy was placed in sight, but out of reach. Children 111 then took part in an age appropriate mood induction task where they were asked to colour in a 112 drawing. Sections of the drawing were numbered and a set colour was assigned to each 113 number (e.g., red for all the sections marked with a "1"). The colour key was visible to the 114 children and the researcher went through this with them such that the child knew which 115 colour crayon was to be used in each area prior to beginning colouring. The children 116 coloured in the drawing, but when they got to the final number, number 6, they realised that 117 the required purple crayon was missing. The researcher then told the child that, because they 118 hadn't completed the colouring, they couldn't have their chosen toy. The experimenter 119 presented the child with the smiley-face rating scale and asked them to indicate their mood 120 now that they couldn't have their toy. The researcher then stated that she would look for the 121 missing crayon and placed six bowls of snack foods in front of the child, which had each 122 been pre-weighed and were presented in separate bowls. Manufacturer nutritional labels were 123 The development of emotional eating used to calculate kilocalories for each snack food, these included: 6g salted crisps (approx. 124 32kcal), 2 chocolate chip cookies (approx. 115 kcal), 21 chocolate buttons (approx. 115 kcal), 125 3 small breadsticks (approx. 31 kcal), 2 carrot sticks (approx. 6kcal), and 9 green grapes 126 (approx. 32 kcal). The child was told that they could eat any of the snacks, or play with 127 nearby toys, while the researcher looked around for the crayon. After 4 minutes, the 128 researcher 'found' the missing crayon and handed it to the child. The child then completed 129 the colouring task, received their chosen toy and re-rated their mood using the smiley face 130 scale, as a check that all children's mood had returned to pre-manipulation standard or above 131 (in all cases mood was returned to baseline or happier mood).

132
Control group: The procedure was identical for children in the control group except 133 that there was no missing crayon for these children and they received their chosen toy after 134 completing the colouring task. Children in the control group rated their mood before and after 135 the colouring task and were given the option to help themselves to the same range of snack 136 foods (as above), or to play with the toys in the laboratory, whilst the experimenter "tidied 137 up". As above, the bowls of snack food were removed after 4 minutes and re-weighed.

138
Data analysis 139 Six parent-child dyads were removed from the data set for the following reasons: 140 child showed an incongruous mood shift (e.g., mood did not deteriorate in the experimental 141 condition or mood deteriorated in the control group), child attended with a grandparent, or 142 child was deemed to be an outlier according to their BMI SDS score, leaving a final sample 143 of 35. Independent sample t-tests were used to examine whether there were any differences 144 between the control and experimental groups according to child age, maternal education, 145 child BMI SDS score, maternal BMI or parental reported feeding practices. Pearson's 146 correlations were used to explore whether these demographic variables were related to 147 children's snack food consumption data at Time point 2. Independent sample t-tests were 148 The development of emotional eating used to explore whether there were gender differences for children on parental reports of 149 feeding practices or child consumption data. 150 For the first aim, to explore the efficacy of the emotion manipulation at inducing 151 negative emotion, paired t-tests were used to examine whether experimental group children 152 reported significant differences in their mood before and after the emotion induction

164
Screening for confounding variables 165 Independent sample t-tests indicated that there were no significant differences between 166 children in the control or experimental group according to maternal education, maternal BMI, 167 child age, or parentally reported child feeding practices at time-point 1 (see Table 1). 168 However there were significant differences between the control and experimental group 169 children according to child BMI SDS score (see Table 1). Two-tailed Pearson's correlations 170 were run to ascertain whether maternal education, maternal BMI, child age or child BMI SDS 171 were related to child consumption of different foods. There were no significant relationships 172 with the exception of maternal BMI which was significantly correlated with child 173 The development of emotional eating consumption of calories from cookies at time point 2 (r = 0.46, p < 0.05). Therefore in all 174 subsequent analyses the effects of child BMI SDS and maternal BMI were controlled for. 175 Independent sample t-tests indicated that there were no significant differences between girls 176 and boys according to parent reports of feeding practices or child consumption data, therefore 177 child gender was not controlled for in the analyses.

178
Differences between control and experimental group 179 Paired sample t-tests suggested that the emotion induction procedure had been 180 successful at altering child emotion with significant differences between child mood before 181 and after the emotion induction in the experimental group. In comparison, in the control 182 group children there was no significant difference in their mood ratings before and after 183 completing the task (see Table 2). Moreover, independent sample t-tests indicated that there 184 were significant differences in child reported mood after the mood induction/control task  Table 2). There was a significant difference in terms of kilocalories consumed from the snack 189 foods between children in the experimental group and those in the control group (see Table   190 1).

191
ANCOVAs 192 There were no significant main effects or interaction effects in the ANCOVAs using  Table 1). There was no 203 significant main effect for use of food as a reward but there was a significant interaction 204 between group and use of food as a reward in predicting calorie intake (F(1,29) = 6.01, p < 205 0.05). Children in the control group ate fewer calories when exposed to high use of food as a 206 reward (N = 8) compared to those exposed to low use of food as a reward (N = 10). In 207 contrast, children in the experimental group ate more calories when exposed to high use of 208 food as a reward (N = 5) compared to low use of food as a reward (N = 12) (See Figure 1). 209 Restriction of food for health reasons: In an ANCOVA exploring differences in control group who were exposed to high levels of restriction for health reasons (N = 8) ate 216 fewer calories than those exposed to low restriction (N = 10), whereas children exposed to 217 high levels of restriction for health reasons who were also exposed to stress (N = 8) ate more 218 total calories than those exposed to low restriction (N = 9) (See Figure 2).

221
This study aimed to assess the efficacy of a refined age-appropriate emotion 222 manipulation in order to establish whether emotional eating can be observed in a group of 5-7 223 The development of emotional eating year old children. It further aimed to evaluate whether parents who report using more 224 controlling feeding practices with their 3-5 year old children are subsequently more likely to 225 find that their children are prone to emotionally overeat 2 years later. The findings of this 226 small scale study indicate that 5-7 year old children exposed to an emotion induction 227 procedure consumed significantly more calories compared to children in a control group. periods to freely access food. Previously it has been suggested that emotional overeating is a 243 learned and abnormal response to stress in young children, and stress should naturally inhibit 244 the tendency to eat (10). These findings support this suggestion and it may be that at some 245 point between the ages of 4 and 6 the tendency, and perhaps opportunity, to emotionally 246 overeat increases in many children. Our findings suggest that one factor which may 247 contribute to this tendency is the feeding practices that young children are exposed to.

248
The development of emotional eating The findings indicate that, compared to a control group, under conditions of negative 249 emotion children were significantly more likely to consume food at ages 5-7 if their parents 250 had reported greater use of food as a reward or use of restriction of food for health reasons 2 251 years previously. Overly controlling feeding practices such as these are likely to be associated 252 with presenting and forbidding foods in situations which override children's natural signals of 253 hunger and satiety. It may be that these more controlling feeding practices which restrict and 254 reward children with food are teaching children to see palatable foods as a tool which can be 255 used to alleviate distress (16). By exploring these relationships longitudinally over time, we 256 have attempted to tease apart causality in these relationships, but it is of course possible that 257 parents utilise these more controlling feeding practices with children who are already highly 258 responsive to the rewarding properties of food, or have a tendency to overeat (22,23,24).

259
Children's eating behaviour is highly complex and multifaceted and it is likely that several 260 other factors, not least a parent's own tendency to emotionally eat, also impact upon the 261 child's tendency to eat in the context of negative emotion (25,26).

262
Contrary to our hypotheses, there was no effect over time of parental pressure to eat, 263 use of food for emotion regulation, or restriction of food for weight reasons on child 264 subsequent emotional eating. Previous research has found that pressure to eat is associated 265 with fussier eating and lower child consumption of pressured foods (which are often healthy 266 foods such as fruits and vegetables) (27). Although the use of pressure may not be conducive 267 to encouraging greater subsequent intake of pressured foods, it may be that pressure to eat has 268 no particular detrimental effect on the tendency to overeat in the context of stress, given that 269 the foods chosen in such contexts are not often those associated with pressuring feeding  The development of emotional eating The findings of this research indicate that emotional eating can be observed in children as 299 young as 5-7, and that previous exposure to more controlling parental feeding practices can 300 exacerbate the tendency to emotionally overeat in children of this age. Given that emotional 301 overeating tends to increase as children age (12)   The development of emotional eating