J Youth Adolescence DOI 10.1007/s10964-014-0186-8

EMPIRICAL RESEARCH

Distinguishing Between Risk Factors for Bulimia Nervosa, Binge Eating Disorder, and Purging Disorder Karina L. Allen • Susan M. Byrne Ross D. Crosby

•

Received: 27 June 2014 / Accepted: 8 September 2014 Ó Springer Science+Business Media New York 2014

Abstract Binge eating disorder and purging disorder have gained recognition as distinct eating disorder diagnoses, but risk factors for these conditions have not yet been established. This study aimed to evaluate a prospective, mediational model of risk for the full range of binge eating and purging eating disorders, with attention to possible diagnostic differences. Specific aims were to determine, first, whether eating, weight and shape concerns at age 14 would mediate the relationship between parentperceived childhood overweight at age 10 and a binge eating or purging eating disorder between age 15 and 20, and, second, whether this mediational model would differ across bulimia nervosa, binge eating disorder, and purging disorder. Participants (N = 1,160; 51 % female) were drawn from the Western Australian Pregnancy Cohort (Raine) Study, which has followed children from pre-birth to age 20. Eating disorders were assessed via self-report questionnaires when participants were aged 14, 17 and 20. There were 146 participants (82 % female) with a binge

K. L. Allen (&)  S. M. Byrne School of Psychology, The University of Western Australia, M304, 35 Stirling Hwy, Crawley, WA 6009, Australia e-mail: [email protected] K. L. Allen Telethon Kids Institute, The University of Western Australia, Crawley, Australia R. D. Crosby Department of Clinical Neuroscience, University of North Dakota School of Medicine and Health Sciences, Grand Forks, ND, USA R. D. Crosby Department of Biostatistics, Neuropsychiatric Research Institute, Salt Lake City, UT, USA

eating or purging eating disorder with onset between age 15 and 20 [bulimia nervosa = 81 (86 % female), binge eating disorder = 43 (74 % female), purging disorder = 22 (77 % female)]. Simple mediation analysis with bootstrapping was used to test the hypothesized model of risk, with early adolescent eating, weight and shape concerns positioned as a mediator between parent-perceived childhood overweight and later onset of a binge eating or purging eating disorder. Subsequently, a conditional process model (a moderated mediation model) was specified to determine if model pathways differed significantly by eating disorder diagnosis. In the simple mediation model, there was a significant indirect effect of parent-perceived childhood overweight on risk for a binge eating or purging eating disorder in late adolescence, mediated by eating, weight and shape concerns in early adolescence. In the conditional process model, this significant indirect effect was not moderated by eating disorder group. The results support a prospective model of risk that applies to bulimia nervosa, binge eating disorder and purging disorder. Common prevention approaches may be possible for bulimia nervosa, binge eating disorder and purging disorder. Keywords Adolescence  Bulimia nervosa  Binge eating disorder  Purging disorder  Risk factors  Raine Study

Introduction Eating disorders are serious mental illnesses that affect up to 15 % of adolescent females and 3 % of adolescent males (Allen et al. 2013a). They have one of the highest mortality rates of any psychiatric disorder (Berkman et al. 2007) and are the second leading cause of mental health disability for

123

J Youth Adolescence

adolescent and young adult females (Begg et al. 2007). Traditionally, research on eating disorders has focused on anorexia nervosa (characterized by the inability to maintain a minimally normal body weight, combined with marked fear of weight gain) and bulimia nervosa (involving regular episodes of binge eating and purging, together with the over-evaluation of eating, weight and shape). Over the last decade, research has shifted to recognize a broader range of eating disorders, particularly a broader range of binge eating and purging disorders. These include binge eating disorder (repeated episodes of binge eating, with distress about binge episodes but without regular purging or other compensatory behaviors) and purging disorder (repeated episodes of purging intended to influence weight or shape, without binge eating), as well as bulimia nervosa. This range of binge eating and purging disorders is now recognized in diagnostic nomenclature. In the 2013 release of DSM-5, bulimia nervosa and binge eating disorder were listed as standalone diagnoses, while purging disorder was provided as an example of the ‘‘Other Specified Feeding or Eating Disorder’’ category (American Psychiatric Association 2013). It is important, therefore, that binge eating disorder and purging disorder are also recognized in eating disorder theories and treatments. Bulimia Nervosa, Binge Eating Disorder and Purging Disorder There are multiple studies documenting the clinical severity of binge eating and purging eating disorders. Individuals with bulimia nervosa, binge eating disorder, and purging disorder score significantly higher than noneating disordered control participants on measures of eating disorder psychopathology (Stice et al. 2013), depression (Allen et al. 2013a) and anxiety (Fink et al. 2009), and have a greater incidence of non-eating disorder Axis I (Fink et al. 2009) and Axis II (Keel et al. 2005) psychiatric disorders. Differences have also been observed between diagnostic groups. Most available data suggest that bulimia nervosa is associated with greater symptom severity and psychosocial impairment than binge eating disorder or purging disorder (e.g., Roberto et al. 2010; Stice et al. 2013), and some studies have found binge eating disorder to be associated with greater depressive symptoms (Roberto et al. 2010) and impulse control difficulties (Keel et al. 2011) than purging disorder. In contrast, purging disorder may be associated with greater dietary restraint and body dissatisfaction than binge eating disorder (Keel et al. 2011). Physiological differences have also been identified between the three disorders, which lend further support to distinctions between the groups. For example, fasting leptin appears to be suppressed in bulimia nervosa and purging disorder (Jimerson et al. 2010), relative to

123

controls, but elevated in binge eating disorder (Monteleone et al. 2000). Appetite response and satiety appear normal in purging disorder but blunted in bulimia nervosa (Keel et al. 2007) and binge eating disorder (Sysko et al. 2007). When considering the course of binge eating and purging eating disorders, different results have been obtained in adolescent and adult samples. In adolescents, overall remission rates are comparable across bulimia nervosa, binge eating disorder, and purging disorder (Allen et al. 2013b). Diagnostic cross-over seems more frequent, however, for binge eating disorder and purging disorder than for bulimia nervosa (Stice et al. 2013). Further, adolescents with an initial diagnosis of binge eating disorder or purging disorder show a tendency to progress to bulimia nervosa with time, while the reverse pattern (progression from bulimia nervosa to binge eating or purging only) is less common (Allen et al. 2013a; Stice et al. 2013). In adults, available data suggest that remission rates for bulimia nervosa are lower than those for binge eating disorder (Fairburn et al. 2000) but similar to those for purging disorder (Keel et al. 2005). Cross-over between bulimia nervosa, binge eating disorder and purging disorder also seems rare in adult groups (Fairburn et al. 2000; Keel et al. 2005). The above findings relate to differences in symptom patterns and remission rates across binge eating and purging disorders. Less is known about possible differences in the etiology of these disorders. To our knowledge, only one study has directly compared risk factors for bulimia nervosa and binge eating disorder (Fairburn et al. 1998). Fairburn and colleagues utilized a retrospective case–control design to compare exposure to risk in women with binge eating disorder (n = 52), women with bulimia nervosa (n = 102), healthy control women (n = 104), and psychiatric control women (n = 102) (Fairburn et al. 1998). Childhood risk factors for any psychiatric disorder (e.g., negative self-evaluation, adverse events, parental psychiatric disorder) and risk factors for dieting/obesity (e.g., childhood obesity, comments regarding weight/ shape, parental obesity) were significant predictors of binge eating disorder as well as bulimia nervosa. However, exposures in these areas were more pronounced for women who developed bulimia nervosa than for those who developed binge eating only (Fairburn et al. 1998). The authors concluded that while binge eating disorder and bulimia nervosa may share common risk factors, in the form of childhood exposures associated with psychiatric disorders and with obesity, effects are weaker for binge eating disorder than bulimia nervosa. Several other studies have considered risk factors for binge eating and purging eating disorders when treating disorders as a single outcome group (e.g., Allen et al. 2014; Stice et al. 2011; Striegel-Moore et al. 2007). Two studies

J Youth Adolescence

from Australia suggest that similar risk factors may operate for binge eating and purging disorders that develop in early (Allen et al. 2009) and later (Allen et al. 2014) adolescence. These analyses made use of prospective data from the Western Australian Pregnancy Cohort (Raine) Study, and considered predictors of risk relative to general control and psychiatric control participants. Being perceived as overweight by one’s parent/s in childhood emerged as a significant prospective predictor of binge eating and purging eating disorders with onset by age 14 (early onset cases) (Allen et al. 2009) and those with onset between age 15 and 20 (later onset cases) (Allen et al. 2014). Moreover, parent-perceived childhood overweight was the only variable, besides female sex, to predict early-onset eating disorder cases in multivariate comparisons with general and psychiatric control participants. For later onset cases, high levels of eating, weight and shape concern in early adolescence were also significant in predicting risk, and were a stronger predictor than parent-perceived childhood overweight (Allen et al. 2014). These results converge with those of Fairburn et al. (1998) and suggest that variables associated with dieting and obesity may predict risk for the range of binge eating and purging eating disorders. To date, however, Raine Study analyses have considered bulimia nervosa, binge eating disorder, and purging disorder as a single outcome group. As such, it is not possible to comment on whether parent-perceived childhood overweight and/or early adolescent eating, weight and shape concerns are of similar importance in predicting each disorder. With binge eating disorder and purging disorder receiving attention as distinct eating disorder diagnoses, it is now important to distinguish between risk factors for these conditions and those for bulimia nervosa. If risk factors prove to be similar across bulimia nervosa, binge eating disorder and purging disorder, support would be provided for prevention and early intervention approaches that target all three conditions. In contrast, if risk factors are different for each diagnosis, it would be necessary to tailor prevention and intervention initiatives to the etiology of each disorder. A Mediational Model of Risk Results from the Raine Study allow predictions to be made about how risk factors may work together to increase eating disorder risk over time. Specifically, it seems possible that parent-perceived childhood overweight and adolescent eating, weight and shape concerns work sequentially to increase risk for a binge eating or purging disorder, with the former being a potential predictor of the latter (Allen et al. 2014). Relatively few studies have considered how putative risk factors may work together to increase eating disorder risk, but such work is important if developmental

pathways to eating disorders are to be identified, and any mediating and/or moderating effects are to be understood (Jacobi et al. 2004). There are strong grounds for hypothesizing a link between parent-perceived childhood overweight and adolescent eating, weight and shape concerns. Family and parental factors have been shown to influence the development of adolescent weight concerns (Field et al. 2001), dieting behavior (Neumark-Sztainer et al. 2010), and binge eating and purging (Linville et al. 2011). These effects seem to stem from family modelling of dieting or weight concerns, and/or actual or perceived (by the adolescent) family pressure to lose weight or be thin (NeumarkSztainer et al. 2010). When considering parent-perceived childhood overweight, it seems likely that children who are seen as overweight by their parents may also be vulnerable to experiencing or perceiving pressure to be thin from those parents. This pressure may be explicit, in the form of parental comments about eating or overt efforts to influence child weight, or implicit, via reinforcement of dieting behavior or admiration of thinness in others (Haines et al. 2008). In childhood, the family is a key source of information regarding cultural ideals, which make parental beliefs and behaviors regarding weight particularly powerful (Field et al. 2001). There is also a compelling body of evidence linking eating, weight and shape concerns to later onset of an eating disorder (for a review, see Jacobi et al. 2004). In addition, links between perceived pressure to be thin, weight and shape concerns, and binge eating/purging have been outlined in Stice and colleagues’ dual-pathway model of bulimia nervosa (Stice et al. 1998). This model positions perceived sociocultural pressure to be thin (from family, peers or the media) as a predictor of thin-ideal internalization and weight and shape concern, which, in turn, are thought to predict dietary restraint, negative affect, and binge eating/purging. Support for the dual-pathway model comes from both cross-sectional (e.g., Ricciardelli and McCabe 2001) and prospective (e.g., Stice 2001) studies, largely with female adolescent participants. The dual-pathway model does not make specific predictions about the timing of each model component. For instance, it does not specify whether perceived pressure to be thin occurs immediately prior to the onset of weight and shape concerns, or if there is a lag of months or years between the two. Most prospective studies of the model have spanned relatively short (up to 2 years) periods, precluding consideration of this issue. In contrast, results from the Raine Study allow timing predictions to be made. Most notably, it seems that parent-perceived childhood overweight may have long-lasting effects on eating disorder risk, with up to 10 years between parental weight perceptions and eating disorder onset (Allen et al. 2014). If

123

J Youth Adolescence

mediational model. Path a denotes the X and M (mediating) variables, path between the M and Y variables, and path between the X and Y variables taking into

account the M variable. The indirect effect of X on Y through M (path ab) has traditionally been referred to as the mediating effect and is the main effect of interest for testing the study’s first hypothesis

parent-perceived childhood overweight does predict eating disorder risk via early adolescent eating, weight and shape concerns, there may be two key time periods for seeking to screen for, and target, at-risk youth. As shown in Fig. 1, these periods would include middle childhood (age 10 years) and early adolescence (age 14 years). The mediational model in Fig. 1 highlights possible time periods for identifying young people at risk for a binge eating and purging eating disorder. However, the model has not been tested as a whole and requires empirical evaluation if the proposed mediational sequence is to be supported. In order to progress research on the etiology of binge eating and purging eating disorders, it is also important to test for diagnostic differences in the model pathways. These differences may apply to the level of exposure to a risk factor (e.g., rates of parent-perceived childhood overweight in bulimia nervosa vs. purging disorder) as well as the effect of the exposure on disorder risk (e.g., the degree to which parent-perceived childhood overweight predicts bulimia nervosa vs. purging disorder). No prior studies have addressed these issues in a prospective study of risk factors for bulimia nervosa, binge eating disorder and purging disorder.

childhood overweight and risk for a binge eating or purging eating disorder, operating via early adolescent eating, weight and shape concerns (a mediational effect). Given the limited prior research on risk factors for binge eating disorder and purging disorder, specific predictions were not made regarding any diagnostic differences.

Fig. 1 The hypothesised association between the b denotes the association c0 denotes the association

The Current Study Given the above, the current study had two related aims. First, it aimed to test the mediational model shown in Fig. 1, where eating, weight and shape concerns in early adolescence are seen as a mediator of the association between parent-perceived childhood overweight and a binge eating or purging eating disorder in late adolescence. Second, it aimed to determine if this mediational model was applicable for bulimia nervosa, binge eating disorder and purging disorder (i.e., whether any moderation effects were present). Analyses made use of the Raine Study cohort. It was hypothesised that a significant indirect effect (path ab) would be found between parent-perceived

123

Method Design and Participants The Raine Study is a prospective pregnancy cohort study in Western Australia. Full details regarding the study have been reported previously (Newnham et al. 1993). In brief, 2,900 women were recruited from King Edward Memorial Hospital between May 1989 and November 1991 to take part in a longitudinal study of child health and well-being. Women were enrolled in the study at 16–20 weeks gestation. There were 2,868 live birth babies and these children and their parent/s were assessed at birth and ages 1, 2, 3, 5, 8, 10, 14, 17 and 20 years. Detailed eating disorder assessment occurred at the 14, 17 and 20-year assessments. Of the 1,598 adolescents who provided eating disorder data at age 14, 1,383 (86 %) also provided eating disorder data at age 17 and/or age 20. These 1,383 participants (49 % male) form the sample of interest for this research. They represent 76 % of the Raine Study participants who completed at least one of the adolescent assessments (n = 1,878), 59 % of the participants who were eligible to take part in the adolescent assessments (n = 2,344), and 48 % of the original cohort. Previous analyses have shown that Raine Study participants lost to follow-up come from more socially disadvantaged backgrounds than those retained to adolescence, as reflected by lower family income and lower parental employment rates during the earlier years of the study (Allen et al. 2013a). However, Raine Study adolescents who provided eating disorder data at age 14 and were subsequently lost to follow-up (n = 215/1,598; 14 %) did

J Youth Adolescence

not differ significantly from those who remained in the study across adolescence on 14-year eating disorder variables, childhood or 14-year psychosocial variables, or childhood or 14-year body mass index (BMI). As seen at the earlier time points, adolescents lost to follow-up after age 14 did come from families characterized by greater social disadvantage than adolescents who continued in the study. Detailed data on eating, weight and shape concerns were collected for the first time at the 14-year Raine Study assessment. Thus, this research focused specifically on adolescents with a binge eating or purging eating disorder that developed after age 14. Adolescents with an eating disorder at age 14 (n = 74), and those who developed anorexia nervosa or atypical anorexia nervosa at age 17 or 20 (n = 12), were excluded. This gave a sample size of 1,297 adolescents (49 % male). This included 146 participants with a binge eating or purging eating disorder in late adolescence (18 % male) and 1,151 control participants (55 % male). Within the binge eating and purging disorder group, diagnoses were as follows: n = 81 for bulimia nervosa (14 % male), n = 43 for binge eating disorder (26 % male), and n = 22 for purging disorder (23 % male). The use of bootstrapping techniques (see Statistical Analyses, below) addresses some of the limitations usually associated with small group sizes. Measures Eating Disorder Symptoms and Diagnoses Details of the eating disorder assessment items, and diagnostic algorithms for determining DSM-5 eating disorder diagnoses, are published in full elsewhere (Allen et al. 2013a). In brief, eating disorder symptoms were assessed using 24 self-report items adapted from the Eating Disorder Examination-Questionnaire (EDE-Q) (Fairburn and Beglin 1994). Adaptations from the original EDE-Q were intended to make the measure suitable for independent completion by adolescents of all ages, and the greatest change was a simplified response scale (Allen et al. 2013a). Adolescents rated items on a 4-point scale ranging from 0 (‘‘Not at all’’) to 3 [‘‘Most of the time’’ (every day or nearly every day)], rather than the 7-point scale of the original EDE-Q. Eating disorder diagnoses were based on responses to these selfreport items, combined with measured height and weight. There is good convergence between EDE-Q assessment of symptoms over 1 month and interview assessment over 3–6 months (Berg et al. 2012). Diagnoses of bulimia nervosa and purging disorder were made to DSM-5 criteria, with the exception that behaviors were assessed over 1 month rather than 3 months (Allen et al. 2013a). For binge eating disorder, the EDE-Q does

not collect information on DSM-5 criterion B (i.e., it does not determine whether three symptoms relating to dysregulated eating behavior or distress over eating are present). Over-evaluation of weight or shape was included in lieu of this criterion (Allen et al. 2013a) as others have shown that over-evaluation of weight and shape reliably distinguishes between individuals with binge eating disorder and individuals who binge eat without clinical impairment (Mond et al. 2007). We acknowledge that this definition differs from strictly defined DSM-5 binge eating disorder. Eating disorder data at ages 17 and 20 were used to identify adolescents with a binge eating or purging eating disorder at those times. If eating disorder diagnoses changed between age 17 and age 20 (e.g., from binge eating disorder to bulimia nervosa), the initial diagnosis was used for classification purposes. This occurred for 16 % of the participants with an eating disorder at age 17 (n = 15/92) and in most cases was from binge eating disorder or purging disorder to bulimia nervosa. Eating disorder data at age 14 were used to calculate a continuous index of eating, weight and shape concerns, for use as the mediating variable in the hypothesized model (Fig. 1). Previous Raine Study analyses found that 14-year eating, weight and shape concerns were a stronger predictor of eating disorder risk than 14-year dietary restraint (Allen et al. 2014). Those analyses also revealed high inter-item correlations between eating concern, weight concern and shape concern items, which informed the decision to combine them on a single scale. As over-evaluation of weight and over-evaluation of shape form part of the diagnostic criteria for bulimia nervosa, binge eating disorder and purging disorder, these two items were excluded from the concern index, in order to avoid measurement overlap between the mediating and outcome variables. Thus, the concern index was calculated by taking the mean of 12 EDE-Q items assessing eating concern, weight concern, and shape concern. It was internally consistent (a = .90). Given the 4-point rating scale, scores could range from 0 to 3. Parent-Perceived Childhood Overweight A single questionnaire item asked parents to indicate whether they thought their child was overweight at age 10. Response options were ‘‘no’’, ‘‘yes, somewhat true’’, and ‘‘yes, very true’’. As per previous analyses (Allen et al. 2009, 2014), responses were dichotomized to give a group where the child was not perceived as overweight, and a group where the child was perceived as at least somewhat overweight. While there is overlap between responses to this item and objective child body weight (based on measured height and weight and BMI), up to 50 % of overweight Raine Study children were incorrectly classified as healthy weight by their parents at age 10, and a small

123

J Youth Adolescence

proportion (\10 %) of healthy weight Raine Study children were incorrectly classified as overweight by their parents at this time (Allen et al. 2009). Previous multivariate Raine Study analyses found parent-perceived childhood overweight to be a stronger predictor of eating disorder risk than objective weight status or BMI (Allen et al. 2009, 2014). Parental weight perceptions showed stability over time, with 90 % of parents who perceived their child as overweight at age 8 also describing them as overweight at age 10, and 76 % of parents who perceived their child as overweight at age 10 also describing them as overweight at age 14. Similarly, objective weight and BMI showed substantial stability from childhood to adolescence within the Raine cohort (Huang et al. 2011). Procedure Parents (typically mothers) completed the questionnaire item on parent-perceived childhood overweight as part of a broader package of confidential self-report questionnaires. The time point of interest for this study was when children were aged 10 years. Questionnaires were posted to parents for at-home completion prior to attendance at a face-to-face interview. Adolescents completed eating disorder items as part of a broader package of confidential self-report questionnaires, administered at ages 14, 17 and 20. Again, questionnaires were posted for at-home completion prior to attendance at a face-to-face interview. Height and weight were measured by a trained researcher during each face-to-face assessment and BMI calculated for use with eating disorder diagnoses. Data collection was approved by the ethics committees of King Edward Memorial Hospital, Princess Margaret Hospital for Children, and the University of Western Australia. Informed consent for participation was collected from parents at enrolment and each childhood assessment, and from adolescents from the 17-year assessment onwards. Statistical Analyses Analysis of variance (continuous variables) and Fisher’s exact tests (categorical variables) were used to compare participant groups on rates of parent-perceived childhood overweight at age 10, level of eating, weight and shape concern at age 14, and demographic variables at age 14. To test the hypothesized mediational model, a simple mediation analysis was conducted with bootstrapping. Subsequently, a conditional process model (a moderated mediation model) was conducted to determine if model pathways differed significantly by eating disorder diagnosis. Simple mediation was conducted in accordance with Fig. 1, where parent-perceived childhood overweight at age 10 was treated as the X variable, eating weight and shape concerns at age 14 as the M variable, and a binge

123

eating or purging eating disorder at age 17/20 (present vs. absent) as the Y variable. Analyses were run in SPSS Statistics 22 using the PROCESS macro developed by Hayes (2013). PROCESS allows for the estimation of direct and indirect effects in mediation and moderation models using ordinary least squares (continuous variables) or logistic regression (categorical variables) path analyses. It can accommodate moderated mediation models and dichotomous outcome variables, thereby extending earlier mediation analyses (Hayes 2013). In addition, PROCESS applies bootstrapping methods to the estimation of effect sizes. Bootstrapping involves treating the observed sample as a representation of the population originally sampled, and resampling observations to result in a new larger sample that is used for the estimation of effects. In addition to increasing power, this method overcomes problems associated with the assumption of normality in non-normal data sets. It is recognized as superior to traditional mediation approaches in almost all instances (Zhao et al. 2010). To obtain reliable estimates, it is recommended that bootstrapping be conducted with 5,000–10,000 bootstrap samples (MacKinnon et al. 2004). For this study, 10,000 samples were used. For the conditional process model, eating disorder group (no disorder vs. bulimia nervosa vs. binge eating disorder vs. purging disorder) was specified as a potential moderator of the hypothesized relationships. As multi-category moderator variables are not accommodated in PROCESS, Mplus was used for these analyses. Analyses were run using the procedures outlined by Preacher et al. (2007) using weighted least squares estimation and bootstrapping estimates. Moderation was tested for each of the model pathways (a, b and c0 ). Given the focus on model testing, analyses were not adjusted for potential covariates. In the previous Raine Study papers on risk factors for early (Allen et al. 2009) and later (Allen et al. 2014) onset binge eating and purging eating disorders, a broad range of antenatal, childhood, and adolescent variables were considered as possible predictors of eating disorder risk. Parent-perceived childhood overweight and adolescent eating, weight and shape concerns emerged as the only specific multivariate predictors in final multivariate analyses, and parent-perceived childhood overweight was more important than objective child weight (as a categorical variable or BMI) in predicting eating disorder risk.

Results Sample Characteristics Sample characteristics are shown in Table 1, for control (non-eating disorder) participants and participants with

J Youth Adolescence Table 1 Descriptive statistics for non-eating disorder control participants and participants with a binge eating or purging eating disorder at age 17 and/or 20 years No ED (n = 1,151)

Bulimia nervosa (n = 81)

Binge eating disorder (n = 43)

Purging disorder (n = 22)

12.8 %a (n = 147)

24.7 %b (n = 20)

32.6 %b (n = 14)

40.9 %b (n = 9)

Mediation model variables Parent-perceived child overweight at age 10 (% [n]) Eating, weight and shape concerns at age 14 (M [SD]) 14-year old demographic information

0.39a (0.37)

0.79b.c (0.49)

0.71b (0.43)

0.92c (0.54)

Sex (% male [n])

55.7 %a (n = 641)

13.6 %b (n = 11)

25.6 %b (n = 11)

Age (M [SD])

14.02 (0.20)

14.04 (0.14)

13.98 (0.16)

14.06 (0.14)

Body mass index (M [SD])

20.87a (3.86)

22.53b (3.80)

23.06b (4.85)

23.70b (3.91)

Low family income (% [n])

15.8 %a (n = 182)

25.9 %b (n = 21)

20.9 % (n = 9)

22.7 % (n = 5)

22.7 %b (n = 5)

Columns with different subscripts are significantly different at p \ .05. Range for 14-year eating, weight and shape concern scores (prior to eating disorder onset) = 0.00–2.00 for the no ED group, 0.00–2.21 for bulimia nervosa, 0.00–1.86 for binge eating disorder and 0.00–1.93 for purging disorder

bulimia nervosa, binge eating disorder and purging disorder. Descriptive statistics are provided for the predictor variables of interest (parent-perceived childhood overweight at age 10, and eating, weight and shape concern scores at age 14) as well as basic demographic information at age 14. Age 14 precedes the age of onset for adolescents who developed a binge eating or purging eating disorder in this sample. At the time of eating disorder diagnosis (age 17 or 20 years), participants with bulimia nervosa had a mean binge eating frequency of between once per week and a few times per week, and a mean purging frequency of approximately once per week. Participants with binge eating disorder had a mean binge eating frequency of a few times per week. Participants with purging disorder had a mean purging frequency of approximately once per week. Associations Between Parent-Perceived Childhood Overweight, Early Adolescent Eating, Weight and Shape Concerns, and Eating Disorder Risk Simple mediation analysis with bootstrapping revealed a significant positive relationship between parent-perceived childhood overweight at age 10 and eating, weight and shape concerns at age 14 (a pathway), and between eating, weight and shape concerns at age 14 and risk for a binge eating or purging eating disorder at age 17/20 (b pathway) (see Fig. 2; Table 2). The direct effect of parent-perceived child overweight on binge eating and purging eating disorders (c0 pathway) was not statistically significant when accounting for eating, weight and shape concerns at age 14. A bias-corrected bootstrap estimate of the indirect effect (ab pathway) generated a significant positive coefficient (=0.64), supporting a mediated relationship between

parent-perceived childhood overweight, eating, weight and shape concerns, and a binge eating or purging eating disorder (see Table 2). Effects of Eating Disorder Diagnosis Conditional process analysis (moderated mediation analysis) with bootstrapping was conducted to determine if model pathways differed significantly by eating disorder diagnosis (no eating disorder vs. bulimia nervosa vs. binge eating disorder vs. purging disorder). Eating disorder diagnosis did not significantly moderate the a pathway (parent-perceived childhood overweight to eating, weight and shape concern) (p = .977), the b pathway (eating, weight and shape concern to eating disorder risk) (p = .735), or the ab pathway (indirect effect of parentperceived childhood overweight on eating disorder risk) (p = .735). These results suggest that the pathways estimated in the simple mediation model (Fig. 2) may be viewed as unconditional on eating disorder diagnosis and applicable to bulimia nervosa, binge eating disorder and purging disorder.

Discussion This study aimed to test a mediational model of risk that linked parent-perceived childhood overweight at age 10 and adolescent eating, weight and shape concerns at age 14 to risk for a binge eating or purging eating disorder between ages 15 and 20. The study also sought to determine if risk pathways differed for bulimia nervosa, binge eating disorder and purging disorder. There was no prior research comparing the etiology of bulimia nervosa, binge

123

J Youth Adolescence Fig. 2 The simple mediational model with unstandardized coefficients. *p \ .05

Table 2 Mediation model results for associations between parentperceived childhood overweight at age 10 (X variable), eating weight and shape concerns at age 14 (Mediating variable), and a binge eating or purging eating disorders at age 17/20 (Y variable) Path Antecedent X (parent-perceived child overweight)

Coefficient

95 % CI

SE

0.25–0.39

0.03

Consequent: M a

0.32*

F(1, 1,158) = 97.58**, R2 = 0.08 Antecedent

Consequent: Y

X (parent-perceived child overweight)

c0

0.29

-0.19–0.77

0.25

M (eating, weight and shape concern)

b

1.97*

1.54–2.41

0.22

Model log likelihood = 97.81, Nagelkerke R2 = 0.16 XM (indirect effect of X)

ab

0.64*

0.46–0.84

0.10

* p \ .05; ** p \ .001 Note. Coefficients are unstandardized. The c0 path denotes the direct effect of X on Y. The ab path denotes the indirect effect of X on Y

eating disorder and purging disorder, and relatively few studies on how risk factors for eating disorders may work together to increase risk over time. We hypothesized links between parent-perceived childhood overweight and adolescent eating, weight and shape concerns based on previous research linking family ideals and behaviors regarding eating and weight to weight concern in adolescents (Field et al. 2001). Adolescent eating, weight and shape concerns have, in turn, been linked to eating disorder risk in a number of studies (Jacobi et al. 2004), including a previous study with the Raine cohort (Allen et al. 2014). The novelty of the current research lies in the combination of risk factors to form a mediational model that spans a 10 year time period, and in the consideration of diagnostic differences between the binge eating and purging eating disorders. Consistent with predictions, adolescent eating, weight and shape concerns at age 14 were found to mediate the relationship between parent-perceived childhood overweight at

123

age 10 and later risk for a binge eating or purging eating disorder. Analyses also revealed that the mediational model was not moderated by eating disorder diagnosis. That is, model pathways were not significantly different across bulimia nervosa, binge eating disorder and purging disorder. These results provide initial support for shared risk factors across the range of binge eating and purging disorders. At a minimum, bulimia nervosa, binge eating disorder and purging disorder may be seen as having some overlap in etiology. In the only previous study to directly compare risk factors for bulimia nervosa and binge eating disorder, Fairburn and colleagues found that the two disorders shared common risk factors, but noted stronger risk effects for bulimia nervosa than binge eating disorder (Fairburn et al. 1998). This finding was not replicated in the current sample, with conditional process analyses showing that model coefficients did not differ significantly across bulimia nervosa and binge eating disorder, or purging disorder. Differences were seen, however, in absolute levels of eating, weight and shape concern at age 14. Specifically, mean eating, weight and shape concern scores were significantly lower for participants with binge eating disorder than for participants with purging disorder, with neither group differing significantly from those with bulimia nervosa. This pattern is consistent with previous reports of lower dietary restraint and body dissatisfaction in binge eating disorder versus purging disorder (Keel et al. 2011), and confirms that differences in the level of exposure to a variable may not translate to differences in the effect of that exposure on later eating disorder risk. In this adolescent sample, the effects of eating, weight and shape concern on eating disorder risk appeared to be comparable across bulimia nervosa, binge eating disorder and purging disorder, despite differences in mean scores at age 14. The tested mediational model builds on previous analyses (Allen et al. 2009, 2014) to show how childhood and early adolescent risk factors may work together to increase the likelihood of a later eating disorder. As noted in the Introduction, there is some overlap between the mediational model and Stice and colleagues’ dual-pathway model of bulimia nervosa (Stice et al. 1998). However, the dual-pathway model has typically focused on middle

J Youth Adolescence

adolescent participants. Similarly, most prevention and early intervention efforts for eating disorders have been targeted towards middle adolescence (Stice et al. 2007). Results from this study suggest that it may be beneficial to extend eating disorder prevention work to childhood or very early adolescence. Children who are perceived as overweight by their parents form an at risk group that may be easily identified through primary care or school-based screening. Monitoring eating, weight and shape concerns within this group could help to facilitate early intervention in children who show increasing concern as they approach or enter their adolescent years. A further advantage of extending prevention work to earlier time points is that parents may be more readily involved in prevention efforts. It is recognized that parents and other family members provide a key resource in the treatment of established adolescent eating disorders (Le Grange et al. 2010), and they may also be a valuable aid in the prevention of these conditions. As discussed by others, the family is an important source of information regarding weight ideals and dieting behavior (Field et al. 2008; Neumark-Sztainer et al. 2010). Engaging parents in strategies to reduce eating disorder risk, and educating them regarding the possible adverse effects of eating, weight and shape concern, could complement more traditional, adolescent-focused eating disorder prevention initiatives. With rates of childhood overweight exceeding 20 % in Australia (Olds et al. 2010) and 30 % in America (Ogden et al. 2012), it is important to acknowledge that many parents who perceive their child as overweight may, in fact, be accurate in that perception. Moreover, a challenge to childhood obesity interventions is that many parents of overweight/obese children do not perceive their child as being overweight (Doolen et al. 2009). Clearly, it is important to attend to the overlap between parental weight perceptions and child/adolescent weight concerns. A limitation of the current study is that the mechanisms by which parental weight perceptions influenced adolescent eating, weight and shape concerns could not be easily determined; children did not report on perceived pressure to be thin or their experience of weight-related comments. However, others have shown that parents who perceive their child as overweight are more likely to try to restrict their child’s eating (Birch and Fisher 2000) and to encourage weight loss attempts (Neumark-Sztainer et al. 2008). Both of these outcomes may be expected to increase eating, weight and shape concerns. They are also recognized as being ineffective for weight management and, indeed, as predicting weight gain in adolescence (Neumark-Sztainer et al. 2008). To avoid negative eating and weight-related outcomes, parents of overweight youth need to be supported to focus on healthy behaviors instead of weight per se, ideally at a family level

(Neumark-Sztainer et al. 2008). The results of the current study suggest that this kind of parent-focused intervention should occur very early in adolescence, before eating, weight and shape concerns become entrenched. While the current study provides new insights into the etiology of binge eating and purging eating disorders, several limitations do require consideration. These include the single-item assessment of parent-perceived childhood overweight, use of self-report eating disorder data, and the assessment of eating disorder symptoms over one month rather than three. Others have found good convergence between self-report assessment over 1 month and interview assessment over 3 months, in terms of eating disorder detection and classification (Berg et al. 2012), but replication of our results with an interview-based assessment would be beneficial. Group sizes for the eating disorder diagnostic groups also varied, and the number of purging disorder participants, in particular, was small. The use of bootstrapping procedures helped to provide sufficient statistical power, but larger group sizes would serve to ensure the representativeness of results. Future studies may also wish to consider sex differences in the proposed model pathways, as moderating effects of sex were not examined in this study. Lastly, the loss of disadvantaged families is well-documented in longitudinal cohort studies (Wolke et al. 2009) and has occurred in the Raine Study sample. Previous analyses have shown that Raine Study participants who remained in the study to adolescence are broadly comparable to the Western Australian population on a range of sociodemographic indicators (Li et al. 2008), but, again, replication with other samples would help to strengthen our results. Despite these limitations, there are three key strengths associated with this research. First, the study made use of a large, population-based cohort followed from pre-birth to young adulthood. Very few studies worldwide can examine prospective predictors of eating disorders over such an extended time period, with sufficient participant numbers to attend to diagnostic differences in risk. Second, we made use of conditional process analyses with bootstrapping to maximize power and consider a moderated mediation model. This statistical approach extends earlier mediation methods and also allowed for the consideration of categorical predictor (parent-perceived childhood overweight) and outcome (eating disorder) variables. Third, this is the first study to explicitly distinguish between risk factors for bulimia nervosa, binge eating disorder and purging disorder. The finding that shared risk factors may exist for these conditions is significant both theoretically and practically. Future studies may benefit from seeking to replicate and extend our findings, and working to identify the variables that can distinguish between bulimia nervosa, binge eating

123

J Youth Adolescence

disorder, and purging disorder. Understanding shared and specific risk factors for these conditions will allow for an optimal understanding of the etiology of each disorder, and for best-practice treatment approaches that consider similarities as well as differences in disorder development. Plausible mechanisms for distinguishing between the disorders include variations in the degree of exposure to shared risk factors (e.g., higher rates of eating, weight and shape concern among those who develop purging but not binge eating), genetic and/or physiological differences that influence which combination of symptoms develop (e.g., greater propensity to overeating for binge eating disorder; Sysko et al. 2007), and disorder-specific risk factors that may span a number of biopsychosocial domains. As consideration of sex differences was beyond the scope of this study, future research should also consider whether our results apply equally to male and female participants.

National Health and Medical Research Council (NHMRC) of Australia. Core funding for the Western Australian Pregnancy Cohort (Raine) Study is provided by the Raine Medical Research Foundation; The University of Western Australia (UWA); the Faculty of Medicine, Dentistry and Health Sciences at UWA; the Telethon Kids Institute; the Women’s and Infant’s Research Foundation; Curtin University; and Edith Cowan University. Funding for the 14-year Raine Study follow-up was provided by the Raine Medical Research Foundation and NHMRC project grants. Funding for the 17-year follow-up was provided by NHMRC programme grant 35314. Funding for the 20-year follow-up was provided by the Canadian Institutes of Health Research and NHMRC project grants. Author contributions KA conceived of the study, conducted the statistical analyses and drafted the initial manuscript; RC provided statistical advice, assisted with the interpretation of results, and helped edit the manuscript; SB contributed to the study design and helped edit the manuscript. All authors read and approved the final manuscript.

References Conclusions This study used conditional process analyses to provide support for a mediational model linking parent-perceived childhood overweight at age 10, adolescent eating, weight and shape concerns at age 14, and risk for a binge eating or purging disorder in later adolescence. Significantly, support was also provided for shared risk factors across bulimia nervosa, binge eating disorder, and purging disorder. Common prevention and early intervention efforts may be possible for these disorders, by targeting children who are perceived as overweight by their parents and adolescents who report heightened eating, weight and shape concerns. Given the role of parental weight perceptions, parentfocused programs may be a worthwhile initiative that serve to complement traditional, adolescent-focused approaches to eating disorder prevention (Neumark-Sztainer et al. 2010). The results of this study suggest that parents who perceive their child as overweight may need particular support to provide health-focused (rather than weightfocused) messages in the home environment. The results also suggest that parental support should be provided early in adolescence, before weight and shape concerns may be expected to increase (Field et al. 2001). Overall, parental support during early adolescence may help to reduce the likelihood of bulimia nervosa, binge eating disorder and purging disorder developing in later adolescence. Acknowledgments We are extremely grateful to the Raine Study participants and their families who took part in this study and to the whole Raine Study team for cohort management and data collection. We would also like to acknowledge the National Health and Medical Research Council (NHMRC) of Australia and the Telethon Kids Institute for their long-term support of the Raine Study. The first author is supported by an early career research fellowship from the

123

Allen, K. L., Byrne, S. M., Forbes, D., & Oddy, W. H. (2009). Risk factors for full- and partial-syndrome early adolescent eating disorders: a population-based pregnancy cohort study. Journal of the American Academy of Child and Adolescent Psychiatry, 48, 800–809. Allen, K. L., Byrne, S. M., Oddy, W. H., & Crosby, R. D. (2013a). DSM-IV-TR and DSM-5 eating disorders in adolescents: Prevalence, stability, and psychosocial correlates in a population-based sample of male and female adolescents. Journal of Abnormal Psychology, 122, 720–732. Allen, K. L., Byrne, S. M., Oddy, W. H., & Crosby, R. D. (2013b). Early onset binge eating and purging eating disorders: Course and outcome in a population-based study of adolescents. Journal of Abnormal Child Psychology, 41, 1083–1096. Allen, K. L., Byrne, S. M., Oddy, W. H., Schmidt, U., & Crosby, R. D. (2014). Risk factors for binge eating and purging eating disorders: Differences based on age of onset. International Journal of Eating Disorders. doi:10.1002/eat.22299. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders, 5th edition (DSM-5). Washington, DC. Begg, S., Vos, T., Barker, B., Stevenson, C., Stanley, L., & Lopez, A. D. (2007). The burden of disease and injury in Australia 2003. Canberra: Australian Institute of Health and Welfaire (AIHW). Berg, K. C., Stiles-Shields, E. C., Swanson, S. A., Peterson, C. B., Lebow, J., & Le Grange, D. (2012). Diagnostic concordance of the interview and questionnaire versions of the Eating Disorder Examination. International Journal of Eating Disorders, 45, 850–855. Berkman, N. D., Lohr, K. N., & Bulik, C. M. (2007). Outcomes of eating disorders: A systematic review of the literature. International Journal of Eating Disorders, 40, 293–309. Birch, L. L., & Fisher, J. O. (2000). Mothers’ child-feeding practices influence daughters’ eating and weight. American Journal of Clinical Nutrition, 71, 1054–1061. Doolen, J., Alpert, P. T., & Miller, S. K. (2009). Parental disconnect between perceived and actual weight status of children: A metasynthesis of the current research. Journal of the American Academy of Nurse Practitioners, 21, 160–166. Fairburn, C. G., & Beglin, S. J. (1994). Assessment of eating disorders: Interview or self-report questionnaire? International Journal of Eating Disorders, 16, 363–370.

J Youth Adolescence Fairburn, C. G., Cooper, Z., Doll, H. A., Norman, P., & O’Connor, M. E. (2000). The natural course of bulimia nervosa and binge eating disorder in young women. Archives of General Psychiatry, 57, 659–665. Fairburn, C. G., Doll, H. A., Welch, S. L., Hay, P. J., Davies, B. A., & O’Connor, M. E. (1998). Risk factors for binge eating disorder. Archives of General Psychiatry, 55, 425–432. Field, A. E., Camargo, C. A. J., Taylor, C. B., Berkey, C. S., Roberts, S. B., & Colditz, G. A. (2001). Peer, parent and media influences on the development of weight concerns and frequent dieting among preadolescent and adolescent girls and boys. Pediatrics, 107, 54–60. Field, A. E., Javaras, K. M., Aneja, P., Kitos, N., Camargo, C. A., Taylor, C. B., et al. (2008). Family, peer, and media predictors of becoming eating disordered. Archives of Pediatrics and Adolescent Medicine, 162, 574–579. Fink, E. L., Smith, A. R., Gordon, K. H., Holm-Denoma, J. M., & Joiner, T. E, Jr. (2009). Psychological correlates of purging disorder as compared with other eating disorders: An exploratory investigation. International Journal of Eating Disorders, 42, 31–39. Haines, J., Neumark-Sztainer, D., Hannan, P. J., & RobinsonO’Brien, R. (2008). Child versus parent report of parental influences on children’s weight-related attitudes and behaviors. Journal of Pediatric Psychology, 33, 783–788. Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis. London: Guilford Press. Huang, R. C., De Klerk, N. H., Smith, A., Kendall, G. E., Landau, L. I., Mori, T. A., et al. (2011). Lifecourse childhood adiposity trajectories associated with adolescent insulin resistance. Diabetes Care, 34, 1019–1025. Jacobi, C., Hayward, C., de Zwaan, M., Kraemer, H. C., & Agras, W. S. (2004). Coming to terms with risk factors for eating disorders: Application of risk terminology and suggestions for a general taxonomy. Psychological Bulletin, 130, 19–65. Jimerson, D. C., Wolfe, B. E., Carroll, D. P., & Keel, P. (2010). Psychobiology of purging disorder: Reduction in circulating leptin levels in purging disorder in comparison with controls. International Journal of Eating Disorders, 43, 584–588. Keel, P. K., Brown, T. A., Holm-Denoma, J., & Bodell, L. P. (2011). Comparison of DSM-IV versus proposed DSM-5 diagnostic criteria for eating disorders: Reduction of eating disorder not otherwise specified and validity. International Journal of Eating Disorders, 44, 553–560. Keel, P. K., Haedt, A., & Edler, C. (2005). Purging disorder: An ominous variant of bulimia nervosa? International Journal of Eating Disorders, 38, 191–199. Keel, P. K., Wolfe, B. E., Liddle, R. A., De Young, K. P., & Jimerson, D. C. (2007). Clinical features and physiological response to a test meal in purging disorder and bulimia nervosa. Archives of General Psychiatry, 64, 1058–1066. Le Grange, D., Lock, J., Loeb, K. L., & Nicholls, D. (2010). Academy for eating disorders position paper: The role of the family in eating disorders. International Journal of Eating Disorders, 43, 1–5. Li, J., Kendall, G. E., Henderson, S., Downie, J., Landsborough, L., & Oddy, W. H. (2008). Maternal psychosocial wellbeing in pregnancy and breastfeeding duration. Acta Paediatrica, 97, 221–225. Linville, D., Stice, E., Gau, J., & O’Neil, M. (2011). Predictive effects of mother and peer influences on increases in adolescent eating disorder risk factors and symptoms: A 3-year longitudinal study. International Journal of Eating Disorders, 44, 745–751. MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the

product and resampling methods. Multivariate Behavioral Research, 39, 99–128. Mond, J. M., Hay, P., Rodgers, B., & Owen, C. (2007). Recurrent binge eating with and without the ‘‘undue influence of weight or shape on self-evaluation’’: Implications for the diagnosis of binge eating disorder. Behaviour Research and Therapy, 45, 929–938. Monteleone, P., Di Lieto, A., Tortorella, A., Longobardi, N., & Maj, M. (2000). Circulating leptin in patients with anorexia nervosa, bulimia nervosa or binge-eating disorder: Relationship to body weight, eating patterns, psychopathology and endocrine changes. Psychiatry Research, 94, 121–129. Neumark-Sztainer, D., Bauer, K. W., Friend, S., Hannan, P. J., Story, M., & Berge, J. M. (2010). Family weight talk and dieting: How much do they matter for body dissatisfaction and disordered eating behaviors in adolescent girls? Journal of Adolescent Health, 47, 270–276. Neumark-Sztainer, D., Wall, M., Story, M., & Van den Berg, P. (2008). Accurate parental classification of overweight adolescents’ weight status: Does it matter? Pediatrics, 121, e1495–e1502. Newnham, J. P., Evans, S. F., Michael, C. A., Stanley, F. J., & Landau, L. I. (1993). Effects of frequent ultrasound during pregnancy: A randomised controlled trial. Lancet, 342, 887–891. Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2012). Prevalence of obesity and trends in body mass index among US children and adolescents, 1999–2010. JAMA, 307, 483–490. Olds, T. S., Tomkinson, G. R., Ferrar, K. E., & Maher, C. A. (2010). Trends in the prevalence of childhood overweight and obesity in Australia between 1985 and 2008. International Journal of Obesity, 34, 57–66. Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Addressing moderated mediation hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral Research, 42, 185–227. Ricciardelli, L. A., & McCabe, M. P. (2001). Dietary restraint and negative affect as mediators of body dissatisfaction and bulimic behavior in adolescent girls and boys. Behaviour Research and Therapy, 39, 1317–1328. Roberto, C. A., Grilo, C. M., Masheb, R. M., & White, M. A. (2010). Binge eating, purging, or both: Eating disorder psychopathology findings from an internet community survey. International Journal of Eating Disorders, 43, 724–731. Stice, E. (2001). A prospective test of the dual-pathway model of bulimic pathology: Mediating effects of dieting and negative affect. Journal of Abnormal Psychology, 110, 124–135. Stice, E., Marti, C. N., & Durant, S. (2011). Risk factors for onset of eating disorders: Evidence of multiple risk pathways from an 8-year prospective study. Behaviour Research and Therapy, 49, 622–627. Stice, E., Marti, C., & Rohde, P. (2013). Prevalence, incidence, impairment and course of the proposed DSM-5 eating disorder diagnoses in an 8-year prospective community study of young women. Journal of Abnormal Psychology, 122, 445–457. Stice, E., Shaw, H. E., & Marti, C. (2007). A meta-analytic review of eating disorder prevention programs: Encouraging findings. Annual Review of Clinical Psychology, 3, 207–231. Stice, E., Shaw, H. E., & Nemeroff, G. (1998). Dual pathway model of bulimia nervosa: Longitudinal support for dietary restraint and affect-regulation mechanisms. Journal of Social and Clinical Psychology, 17, 129–149. Striegel-Moore, R. H., Dohm, F. A., Kraemer, H. C., Schreiber, G. B., Taylor, C. B., & Daniels, S. R. (2007). Risk factors for bingeeating disorders: An exploratory study. International Journal of Eating Disorders, 40, 481–487. Sysko, R., Devlin, M., Walsh, B. T., Zimmerli, E., & Kissileff, H. R. (2007). Satiety and test meal intake among women with binge

123

J Youth Adolescence eating disorder. International Journal of Eating Disorders, 40, 554–561. Wolke, D., Waylen, A., Samara, M., Steer, C., Goodman, R., & Ford, T. (2009). Selective drop-out in longitudinal studies and nonbiased prediction of behaviour disorders. British Journal of Psychiatry, 195, 249–256. Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of Consumer Research, 37, 197–206.

Karina L. Allen is an Early Career Research Fellow at the University of Western Australia and Telethon Kids Institute. She received her Master of Clinical Psychology and Doctor of Philosophy (PhD) degrees from the University of Western Australia. Her major research interests include eating and weight disorders.

123

Susan M. Byrne is an Associate Professor at the University of Western Australia. She received her Master of Clinical Psychology and Doctor of Philosophy degrees from the University of Western Australia, and a Doctor of Philosophy (DPhil) in Clinical Medicine from Oxford University. Her major research interests include eating and weight disorders. Ross D. Crosby is Director of Biomedical Statistics and Methodology at the Neuropsychiatric Research Institute and a Clinical Associate Professor at the University of North Dakota, School of Medicine. He received his Doctor of Philosophy (PhD) degree from the University of Nevada. His major research interests include statistical applications in psychiatric research and treatment outcome research in eating disorders.