Predicting treatment outcome in individuals with an eating disorder

Author: Eva Vall

Vall, Eva, 2016 Predicting treatment outcome in individuals with an eating disorder, Flinders University, School of Psychology

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact with the details.


Eating disorders are serious and prevalent mental disorders that are associated with significant impairments across both physical and psychological domains. Achieving good treatment outcomes is an important goal for this group, but many patients do not improve significantly with treatment. At present, it is not clear which factors can best distinguish those who are likely to respond well versus those who are not. The purpose of this thesis was to improve our understanding of what predicts outcome in individuals receiving treatment for an eating disorder.

The first study was an investigation into the current evidence base, and included a comprehensive review and meta-analysis. The most robust predictors of treatment outcome that emerged were early change in symptoms during treatment and baseline motivation. A number of key gaps and limitations across the literature were identified. These included an absence of testing of theoretically informed models with appropriately sophisticated statistical analyses, few investigations into transdiagnostic predictors, and a wide variability in how outcomes were measured. Three empirical studies were subsequently undertaken to address these limitations, and were additionally guided by prominent theoretical models of eating disorders. These models suggested that perfectionism/cognitive rigidity, mood intolerance and ineffectiveness were potentially important predictor variables.

In order to validate a measure of cognitive rigidity, the second study examined a computerized version of the Trail Making Test (assessing set-shifting) in a transdiagnostic sample of inpatients with an eating disorder.

For the remaining empirical studies (third and fourth studies), two treatment samples were obtained. The first consisted of adolescents with anorexia nervosa (AN) who were inpatients in a specialist paediatric unit. Change in two predictor variables during treatment predicted outcomes: greater weight change during treatment predicted greater increases in body mass index (BMI) centile by follow-up, and increased personal standards perfectionism during treatment predicted a greater likelihood of being readmitted within 3 months of discharge. There were interactions between time and driven exercise (higher levels of baseline exercise resulted in a lower BMI centile at follow-up) and set-shifting (faster set-shifting was associated with poorer quality of life at discharge and follow-up). In terms of simple baseline predictors, higher purging, concern over mistakes perfectionism, emotional regulation difficulties, and faster set shifting were associated with higher levels of eating disorder pathology and poorer quality of life over all points of follow-up.

The second sample was transdiagnostic, and consisted of adults admitted to a short-term stay specialist inpatient eating disorder unit. Diagnosis did not predict any outcome measure, with the expected exception of weight change. Individuals with more severe eating disorder symptoms made greater improvements in disordered eating by follow-up, but remained at a higher overall level of illness severity. Greater improvement in symptoms during treatment was associated with greater improvement at follow-up. Baseline concerns over mistakes moderated the relationship between improvements in quality of life as an inpatient and improvements in quality of life at follow-up. For those with anorexia nervosa, a longer duration of illness predicted less improvement in BMI at follow-up, while improvements in quality of life during treatment predicted greater improvements in BMI by follow-up.

Taken together, these findings suggest that across age-groups, encouraging robust symptom change during treatment is likely to result in improved outcomes. While our results provide preliminary support for the selected theoretical predictors, further research in more controlled settings is needed to better understand their contribution to outcomes. Perfectionism appears to warrant particular investigation as a target for therapeutic intervention that could have clinical benefit.

Keywords: eating disorder, treatment, outcome, response, predictor, anorexia, bulimia, moderator, mediator

Subject: Psychology thesis

Thesis type: Doctor of Philosophy
Completed: 2016
School: School of Psychology
Supervisor: Professor Tracey Wade