Emotional, external and restrained eating behaviour and BMI trajectories in adolescence

Highlights

• We model trajectories of adolescents’ BMI development with growth mixture modelling.

• Development of BMI was best fitted in five distinct trajectories with similar moderate BMI increase.

• Restrained eaters had a higher change of being in the higher BMI trajectories.

• Emotional and external eating were unrelated to BMI trajectory membership.

Abstract

Individual differences in eating behaviours might partly explain the variations in development of weight gain and subsequent overweight and obesity. In the current study, identified trajectories of BMI in adolescence and their associations with restrained, emotional and external eating were tested. For the assessment of BMI trajectories growth mixture modelling was used; a method used to identify clusters of individuals within a population that follow distinct developmental trajectories. In total 328 Dutch adolescents (13–15 years old at baseline) self-reported their height and weight at five annual waves and their eating behaviour at baseline. Development of BMI was best fitted in five distinct trajectories that showed similar moderate increase of BMI over time; parallel but at a different level. High restrained eaters had a higher chance of being in the higher BMI trajectories. Emotional and external eating were unrelated to the BMI trajectories. In conclusion, adolescents in this study followed very parallel patterns of moderate increases in BMI which suggests that factors acting on individual differences in weight status have had their influence mostly at a – perhaps much – younger age. Restraint eating was related to BMI in early adolescence, but not to an increases or decreases in BMI over the course of adolescence.

Introduction

The literature on obesity and eating pathology in adolescents has increasingly focused on both environment and individual differences in eating behaviours and physical exercise. Even in times where the prevalence of childhood obesity has increased profoundly in most Western countries (Wang & Lobstein, 2006), the question remains how we can explain that some youths keep their weight in balance for years, while others become overweight. Individual differences in eating behaviours might partly explain the variations in development of weight gain and subsequent obesity (van Strien, Frijters, Bergers, & Defares, 1986). In the current study, we tested how trajectories of BMI development in adolescence were related to differences in eating behaviour.

According to the theory of restrained eating, dieting can cause overweight through overeating. People who diet suppress their feeling of hunger cognitively and eat less. However, when cognitions of self-control are undermined restrained eaters may abandon their diet altogether and are more likely to overeat than non-dieting individuals (Herman and Polivy, 1980, Herman and Polivy, 2004, Polivy, 1996, Polivy and Herman, 1985, Ruderman, 1986). The psychosomatic theory focuses on “emotional eating” which is overeating in response to emotional arousal or (di)stress. In other words, eating as a coping mechanism in order to reduce negative emotions (Macht & Simons, 2011). This is considered an a-typical reaction since the typical response to distress is considered loss of appetite due to physiologic effects (including inhibition of gastric contractions and elevation of blood sugar) (Greeno and Wing, 1994, Kaplan and Kaplan, 1957, Schachter et al., 1968). Emotional eating may occur in people who developed a conditioned association between negative mood and food reward, for example due to childhood experiences of food being used as comfort (Bruch, 1973) or have not learned to express and regulate their emotions in childhood (Topham et al., 2011). A third theory, focussing on external eating, states that certain people are more sensible to external food cues than others, and eat in response to a variety of external stimuli (Schachter & Rodin, 1974), such as time manipulations, taste, visibility and accessibility of food, regardless of internal states of feelings of hunger or satiety. In an obesogenic environment with many (unhealthy) eating cues, external eating is hypothesised to be related to overeating. The predictive value of these three theories in explaining weight status and weight change has been widely tested in adult samples (e.g. Koenders & van Strien, 2011), but in the past two decades increasing attention has been paid to children and adolescents.

Concerning restrained eating, there is consistent empirical evidence that restraint is higher in overweight adolescents. Cross-sectional studies on small samples of UK, French, and Dutch adolescents (Lluch et al., 2000, Rutters et al., 2011, Wardle et al., 1992) reveal a positive association between restraint and overweight status or BMI. In large samples of 2474 Belgium 13–18 year olds, (Braet et al., 2008) and 10,087 Dutch 12–14 year olds, (Snoek, van Strien, Janssens, & Engels, 2007) higher restrained eating scores were found for overweight adolescents. Recently, positive associations have been found between dietary restraint and several adiposity measures (BMI, waist circumference, and skinfolds) for 17–18 year old adolescent boys and girls (Ledoux, Watson, Baranowski, Tepper, & Baranowski, 2011). In prospective studies, restrained eating also predicted increases in weight in a US sample of 13–17 year olds (Stice, 2001), and similar findings were obtained for US pre-adolescent and adolescents over 3-years-time (Field et al., 2003; see also Neumark-Sztainer et al., 2007).

Fewer studies have been conducted on emotional eating in adolescents. Most of these studies found no associations between emotional eating and weight status in this age-group (Belcher et al., 2011, Goldfield et al., 2010, Lluch et al., 2000, Nguyen-Rodriguez et al., 2008, Snoek, van Strien, et al., 2007, Wardle et al., 1992) whereas Braet et al. (2008) found elevated levels of emotional eating for overweight 13–18 year old girls – but not boys, and Snoek, van Strien, et al. (2007) found a negative association with higher emotional eating scores for adolescent boys with lower BMI-scores. There is a considerable amount of research on the relationship between emotional eating and weight status in younger children. The results of these studies show both positive (Jahnke and Warschburger, 2008, Parkinson et al., 2010, Viana et al., 2008) and negative associations (Braet et al., 2008, Ledoux et al., 2011). So far, there is no prospective research on emotional eating and changes in BMI in the course of childhood and adolescence.

The few studies that did focus on external eating found that high external eating is related to lower BMI (Braet et al., 2008, Goldfield et al., 2010, Lluch et al., 2000, Snoek, van Strien, et al., 2007, Wardle et al., 1992) or for younger children unrelated to weight status (Hill et al., 1994, Jahnke and Warschburger, 2008, Ledoux et al., 2011), which is in sharp contrast with experimental research showing that food cues are related to food intake in children (Anschutz, Engels, & Van Strien, 2009) and to food intake and being overweight in adults (Herman and Polivy, 2008, van Strien, Herman, Anschutz, 2012, Wansink et al., 2007, Wardle et al., 1992). In addition, external eating is related to enhanced selective attention (i.e. attentional bias) to food cues (Brignell, Griffiths, Bradley, & Mogg, 2009) which suggests a higher vulnerability to overeating (although others suggest this relation might be due to impulsivity, see Hou et al., 2011). The relation between external eating and weight status is thus unclear especially as survey studies on the associations between external eating and weight status in youth so far have been cross-sectional and prospective studies are lacking.

To test the relation of the eating behaviours with body weight status, it is essential to have longitudinal data in which people are followed over the course of adolescence. Although there is ample research testing age differences in weight status or body weight (e.g. Cole, Bellizzi, Flegal, & Dietz, 2000), there is a lack of studies that examine individual differences in the development of body weight in adolescence. In the area of developmental psychology, growth mixture modelling is a state-of-the-art technique to test whether clusters of individuals exist within a population who follow unique, qualitatively distinct developmental trajectories (i.e. latent trajectory groups; for a detailed description see, e.g., Jones et al., 2001, Muthén and Muthén, 1998–2012, Nagin, 1999). A few studies applied this technique on weight status of adolescents. For example, Li, Goran, Kaur, Nollen, and Ahluwalia (2007) found three trajectories of overweight trajectories: early onset overweight, late onset overweight and never overweight of which the two overweight trajectories were predicted by among others, socio-demographic and maternal lifestyle variables. Mustillo et al. (2003) found four trajectories of obesity, only chronic obesity was associated with psychiatric disorders. Finally, Ventura, Loken, and Birch (2009) found four trajectories of BMI development for adolescent girls, the highest BMI trajectory was related to a higher maternal BMI and shorter duration of being breast-fed. No study so far related eating behaviours of adolescents to developmental trajectories of BMI. This has an added value to the current literature on emotional, external and restrained eating in adolescence. Firstly because prospective studies relating emotional and external eating to BMI are lacking. Secondly, because with this technique not only a positive association can be shown, but also how eating behaviour relates to different patterns of BMI development.

Using growth mixture modelling, we computed the individual body weight trajectories of 328 Dutch adolescents, using data from five annual waves. The first aim of this study is to explore developmental trajectories of BMI in adolescence. The second aim is to examine how eating behaviours at baseline measurement (when adolescents are approximately 13–15 year old) are related to individual membership to one of the BMI trajectories. We controlled for parental BMI since associations between eating behaviour and BMI could be affected by family similarities in BMI (Davison, Markey, & Birch, 2000) and eating behaviour, especially for emotional (Snoek, Engels, Janssens, & van Strien, 2007) and restrained eating (Snoek, van Strien, Janssens, & Engels, 2009).