Access to evidence-based treatment for individuals with eating disorders is low, with approximately only 23% of people receiving treatment (Layard R. et al, 2012). This is particularly worrying given that mortality rates for those with eating disorders are twice as high than that of the general population (Arcelus J. et al, 2011).
Early intervention for eating disorders has therefore been recognised as essential by the Royal College of Psychiatrists (2019). For this to happen, childhood interventions are necessary, which means understanding the childhood eating behaviours which might be associated with later eating disorders, to be able to target those most at risk. Previous research has suggested that behaviours such as over-eating, under-eating and food fussiness might be precursors to adolescent dieting, which has been suggested as a risk factor for anorexia nervosa (Thornton LM. et al, 2017) along with being directly associated with adult anorexia nervosa (Nicholls DE & Viner RM. 2009) and body mass index (Herle M. et al, under review).
Therefore, the authors of this study wanted to investigate the associations between childhood eating behaviours and eating disorder behaviours and diagnoses in adolescence (Herle et al, 2019). They hypothesised that there would be a continuity between these behaviours in childhood (specifically over, under and fussy eating) with adolescent eating disorders behaviours and diagnoses.
Data from 4,760 participants from the ALSPAC (Avon Longitudinal Study of Parents and Children) was used. ALSPAC is a population-based longitudinal cohort of mothers and their children born in the South West of England between 1991-1992.
Exposures: Trajectories of childhood eating behaviours
Trajectories of childhood eating behaviours were obtained from repeated parent-reported measures at eight time points before the children were 10 years old. At each time point parents reported how worried they were about their child’s eating habits in relation to their over-eating, under-eating and fussy eating.
These reports were used to create childhood eating trajectories for each child detailing their levels of these behaviours over time until the age of 10.
Outcomes: Eating disorder behaviours and diagnoses
Eating disorder behaviours (binge eating, purging, fasting and excessive exercise) in adolescence were later obtained by self-report measures when the children were 16 years old. Eating disorders diagnoses were also determined at age 16 by self and parental-report against the DSM-5 criteria.
The authors included covariates to control for any confounding variables between the exposure and outcomes. These were:
- Maternal age at birth
- Maternal education
- Family socioeconomic status
- Size at birth
- Assigned sex at birth.
The results of this study are presented as risk differences relative to a reference group (those in the sample who exhibited no/low levels of the eating behaviour) for each eating behaviour.
The results for the associations and risk differences between the eating behaviour trajectories and eating disorder behaviours and diagnoses at age 16 (compared to the reference group) are as follows:
- Those with low levels of over-eating until age 5, which then increased (late increasing), had:
- a 6% increased risk of engaging in binge eating (95% CI 2 to 8)
- and a 1% increased risk for binge eating disorder (BED; 95% CI 0 to 3).
- Those who displayed some levels of under-eating at a young age, which decreased to none by age 5 (low transient) and age 9 (slowly decreasing), had:
- a 3% lower risk of fasting (95% CI -5 to 0 and CI -6% to 0 respectively) and purging (both groups 95% CI -5 to -1)
- and a 2% risk reduction for excessive exercise (95% CI -4 to -1)
- Girls with persistently high under-eating throughout their childhood had:
- a 6% increased risk of meeting the diagnostic criteria for anorexia nervosa (95% CI 0 to 12)
- Those with high persistent fussy eating throughout their childhood and those displaying fussy eating at 1 year which gradually decreased both had:
- a 2% risk increase for anorexia nervosa (95% CI 0 to 0.4 and 95% CI 1 to 4 respectively).
Sensitivity analyses were also performed to test for the possibility that anorexia nervosa might be driven by those who were both persistent under and fussy eaters. However, only 4 of the 69 girls meeting the criteria for anorexia nervosa fell into both groups, suggesting both eating behaviours to have a unique association with anorexia nervosa.
The results of this study support the hypothesis of a continuity between childhood eating behaviours and later eating disorder behaviours and disorders. Over-eating in childhood was found to be associated with an increased risk for binge eating at 16 years, with increased overeating in mid-late childhood having a greater probability of binge eating disorder. This suggests that overeating behaviour linked to BED may begin around age 5.
Low levels of under-eating (a healthy appetite) was associated with a lower risk of fasting and excessive exercise at 16 years, suggesting this to be a protective factor. Persistent under-eating in girls was associated with an increased risk of eating disorder behaviours and diagnoses. Finally, early and persistent fussy eating was associated with an increased risk of anorexia nervosa, with the authors suggesting this to either be a contributing factor or early manifestation of adolescent anorexia nervosa.
Strengths and limitations
This study has many strengths:
- This is the largest and most comprehensive study to date looking at the associations between childhood eating behaviours and adolescent eating disorder behaviours and diagnoses. The use of the ALSPAC dataset provides repeated-measures data across childhood and adolescence over 16 years in a large sample of 4,760 participants.
- The authors used latent trajectories as the exposures in this study which allowed for heterogeneity in the childhood behaviours associated with later adolescent eating disorders. This meant that the authors were able to explore the unobserved behaviours for individual participants over time and how these manifested into later eating disorder behaviours and diagnoses.
- Covariates were included to combat the lack of follow-up data traditionally seen over time in longitudinal studies, with lack of follow up in the ALSPAC over the course of the 16 years previously being found to be associated with socio-economic class (Howe CJ. et al, 2016)
However, this study has some limitations:
- The measures are all either self-report or parent-report (including eating disorder diagnosis), which has the potential to be biased. For example, some information may have been held back due to a fear of judgement or social desirability
- Recruitment for the ALSPAC took place in the South West of the UK, which limits this study’s generalisability. Future studies may benefit from wider recruitment
- The methodology of the ALSPAC may now be out of date, for example ARFID (Avoidant/Restrictive Food Intake Disorder) is an eating disorder characterised by fussy eating. ARFID was not included in ALSPAC because at the time it was not yet recognised as an eating disorder
- Although the most comprehensive to date, this study still had low statistical power due to eating disorders being relatively uncommon within the sample, particularly in boys.
Implications for practice
The results of this study highlight some important areas for future practice and research. Although the most comprehensive study to date, further research to consolidate these findings is needed, with more up to date measures. For example, not only do future studies need to assess for newly classified eating disorders such as ARFID, but the authors also reported the measures in this study to be largely female-centric. Traditional eating disorder measures are generally geared more towards females and tend to ignore eating disorders presentations seen more often in males; such as concerns for muscularity and the use of muscle-building products (Calzo JP. et al, 2016). Therefore, this may explain the lack of eating disorders in males in this study, with future research needing to include measures also geared towards eating disorders identification in boys and men to improve the validity of the findings.
The results of this study indicate a potential to predict which children might develop eating disorders, which may allow for preventative strategies to be used. Previous research has demonstrated the ability for childhood eating behaviours to be modified through parental interventions (Magarey A. et al, 2016). However, most of the research is related to obesity (Hayes JF. et al, 2018) which although may be beneficial for binge eating disorder, ignores other eating disorders. Therefore, research to further explore potential preventative interventions and the mechanisms behind the associations observed is essential.
Statement of interests
Herle M, De Stavola B, Hubel C, Abdulkadir M, Santos Ferreira D, Loos RJF, et al. (2019) A longitudinal study of eating behaviours in childhood and later eating disorder behaviours and diagnoses. BJPsych 2019 216(2) 1-7.
Arcelus J, Mitchell AJ, Wales J and Nielsen S (2011) Mortality rates in patients with anorexia nervosa and other eating disorders: A meta-analysis of 36 studies. Arch Gen Psychiatry 68(7) 724–31.
Calzo JP, Horton NJ, Sonneville KR, Swanson SA, Crosby RD, Micali N, et al. (2016) Male eating disorder symptom patterns and health correlates from 13 to 26 years of age. J Am Acad Child Psy 2016 55(8) 693–700.
Hayes JF, Fitzsimmons-Craft EE, Karam AM, Jakubiak J, Brown ML, Wilfley DE. (2018) Disordered eating attitudes and behaviors in youth with overweight and obesity: implications for treatment. Curr Obes Rep 2018 7(3) 235–46.
Herle M, DeStavola B, Hübel C, Santos Ferreira DL, Abdulkadir A, Yilmaz Z, et al. (2020) Eating behavior trajectories in the first ten years of life and their relationship with BMI. Int J Epidemiol under review.
Howe CJ, Cole SR, Lau B, Napravnik S, Eron JJ. (2016) Selection bias due to loss to follow up in cohort studies. Epidemiology 2016 27(1) 91–7.
Layard R, Banerjee S, Bell S, Clark D, Field S, Knapp M, et al. (2012) How Mental Illness Loses out in the NHS. London School of Economics and Political Science. Centre for Economic Policy.
Magarey A, Mauch C, Mallan K, Perry R, Elovaris R, Meedeniya J, et al. (2016) Child dietary and eating behavior outcomes up to 3.5 years after an early feeding intervention: the NOURISH RCT. Obesity 2016 24(7) 1537–45.
Nicholls DE, Viner RM. (2009) Childhood risk factors for lifetime anorexia nervosa by age 30 years in a national birth cohort. J Am Acad Child Adolesc Psychiatry 2009 48(8) 791–9.
Royal College of Psychiatrists (2019). Position statement on early intervention for eating disorders.
Thornton LM, Trace SE, Brownley KA, Algars M, Mazzeo SE, Bergin JE, et al. (2017) A comparison of personality, life events, comorbidity, and health in monozygotic twins discordant for anorexia nervosa. Twin Res Hum Genet 2017 20(4) 310–8.