The Role of Reward and Punishment Sensitivity in the Development of Eating Disorders. A Prospective Cohort Study

Eating disorders most commonly affect people for the first time in adolescence and can involve behaviours like under/over eating, with ensuing low or significantly high weight having a potential longer-term impact on individuals' health and wellbeing. Known risk factors like childhood anxiety and dieting behaviours contribute to the development of these serious and debilitating psychiatric disorders. Another risk factor, which is a potentially modifiable target within prevention programmes, may be reinforcement sensitivity, an individual's response to signals of reward and punishment. For example, a cautious person who is risk averse and fears failure may make different decisions under conditions of risk, compared to a more impulsive, reward driven person. The latter individual may be motivated by the seeking out of rewards (high reward sensitivity), whereas the behaviour of the former may be motivated by wanting to avoid punishment (high punishment sensitivity). These differences may be important for both people acutely unwell with eating disorders and those who have recovered. For example, those who struggle with under-eating and resulting low weight tend to have high levels of punishment sensitivity and low levels of reward sensitivity and are driven by the desire to avoid failure (e.g., weight gain). On the other hand, those who are more reward sensitive and impulsive may struggle with bingeing and purging and lose control over their eating. For some, this can result in overweight and obesity. What is not known is whether these differences in reward and punishment sensitivity are present before the illness, occur as a result of the impact of starvation on the brain, and/or act as a scar of the illness in spite of symptom recovery.

Thus, the overarching aim of this longitudinal, prospective, cohort study is to explore whether reward and punishment sensitivity in childhood and early adolescence play a role in the development of eating disorders in later adolescence. This will be addressed by accessing already available quantitative data from the Millennium Cohort Study, a large UK general population study of about 19,000 people born in 2000-2001 followed up to the present day. At ages 11 and 14, Millennium Cohort Study participants completed the Cambridge Gambling Task, a computerised experimental task requiring participants to make decisions about where a token is hidden and how many points they wish to gamble on correctly identifying the token's location. Responses across the task allow for variables to be computed which measure risk-taking and deliberation time (reward sensitivity) and risk adjustment and delay aversion (punishment sensitivity). Examining how participants performed on these variables will reveal how reward and punishment sensitivity impact the later development of eating disorder symptoms (measured in the Millennium Cohort Study at ages 14 and 17). We will also explore in detail the rich raw Cambridge Gambling Task data for each participant to explore how Millennium Cohort Study participants' with eating disorders change their responses over the course of the task and how specific patterns of change can predict later symptoms. We hypothesise that responses on the Cambridge Gambling Task at ages 11 and 14 will predict the probability of developing eating disorders at 14 and 17, controlling for comorbidity (e.g. mood) and factors like puberty onset. Statistical analyses (logistic regression, latent class and latent transition analysis) will be used to examine these predictions. This work will enhance understanding of the extent to which a psychological factor currently identified in the acute and remitted phases of illness may contribute to illness develop. This new knowledge will be shared through an international academic presentation, three journal articles and events with key patient, carer, teacher, practitioner, GP and public stakeholders around the UK to improve illness understanding and detection

1) Design
This prospective, observational, population-based study will adopt a longitudinal design and utilise data from the Millennium Cohort Study, a UK-based, national cohort study of about 19,000 children born in 2000-2001 and followed to the present day. This will permit an understanding of the putative risk conveyed by differences in reward and punishment sensitivity measured using the Cambridge Gambling Task, administered in the Millennium Cohort Study at ages 11 and 14, on the development of symptoms of eating disorders, measured in the Millennium Cohort Study at ages 14 and 17. The Cambridge Gambling Task is an experimental measure of decision-making under risky conditions and a subtest of the well-validated Cambridge Neuropsychological Test Automated Battery.

2) Hypotheses and Data Analyses
Hypothesis 1: Responses on the Cambridge Gambling Task at 11 and 14 will affect the odds of developing eating disorder psychopathology at 14 and 17.
Hypothesis 2: Responses on this task will also be associated with the development of specific clusters of eating disorder symptoms.
Hypothesis 1 will be tested using logistic regression. Hypothesis 2 will be tested using latent class analysis and latent transition analysis. In collaboration with our Academic Advisory Board, raw Cambridge Gambling Task data will be further exploited to further investigate underlying computational mechanisms in eating disorders.

3) Application and Exploitation of the Research
The project involves the exploitation of an already generally available data resource to address currently unanswered questions in eating disorders. To ensure broad impact beyond academic communities and to address misinformation about the causes of eating disorders, 6 research surgeries will be held in London, Nottingham, Manchester and Edinburgh to engage with patient, carer, teacher and practitioner stakeholders. A conference for the general public will also disseminate new knowledge to a wider audience.

The proposed impact activities aim to increase knowledge of how eating disorders develop through engaging with stakeholders like patients, carers, practitioners, academics, teachers, GPs and the lay public to disseminate key project findings. It is hoped that this will a) contribute to reduced stigma which can aid help-seeking; b) improve symptom knowledge in those in regular contact with young people, thus supporting early detection and preventing onset. As Millennium Cohort Study data are already freely available for use, all support research costs are being requested to focus purely on delivering impact. Through engagement with the PI's professional network and our knowledge of the field, we have identified the audiences beyond academia that the project looks to engage with or impact upon. These include:
1) Clinicians working with people with eating disorders who will be able to use the new knowledge generated to inform the individual formulations they develop with patients to understand the illness and develop treatment plans;
2) General Practitioners and teachers who play a key role in detecting eating disorders and will benefit from greater knowledge about the development of eating disorders;
3) Patients with eating disorders who will be able to better understand factors that may have contributed to the development of their illness;
4) Caregivers of people with an eating disorder who frequently request more information and knowledge about eating disorders, and;
5) The lay public who often have poor knowledge of eating disorders and their development.
Patient, carer and teacher stakeholders were all consulted during the development of the proposal for this project.

To deliver impact to these audiences, the following impact activities are planned:-
a) A free, one day conference on 'The Causes of Eating Disorders' will be held in central London due to its strong transport links. Its focus will be to engage the wider lay public as well as the other stakeholder groups described above to share knowledge and information about eating disorders and how they develop.
b) Three research surgeries will be held in central London. Representatives from the stakeholder groups outlined above will be invited to discuss the research findings and their implications for their own practice and to work with the PI to generate new ideas for future research.
c) Three research surgeries will also be held in locations outside of London to ensure broader impact beyond London and the South East, utilising the PI's professional networks. These surgeries will involve a carers' group in Manchester, primary and secondary schools in Nottingham and a service for people with eating disorders in Edinburgh.
The effectiveness of these impact activities will be measured through attendance figures, social media presence and the building of new and enhanced networks. This will allow the team to self-assess what impact approaches are working best in terms of outreach and to focus more on them to ensure the project delivers.

Knowing more about the involvement of individual differences in reward and punishment sensitivity in the development of eating disorders will also impact eating disorder prevention programmes. Impact here will be achieved through utilising the PI's professional network to engage with leading developers of ED prevention programmes (e.g., Dr Eric Stice in the USA, Prof Ulrike Schmidt in the UK and Dr Peter Musiat who develops online mental health prevention programmes at Flinder's University, Adelaide, Australia) to build this new knowledge into existing interventions.
Disseminating the findings through an oral presentation submitted to the world's largest annual international conference on eating disorders, the Academy for Eating Disorders' International Conference on Eating Disorders, to be held in June 2020 in Sydney, will offer opportunities to further impact the agenda around eating disorder prevention.