Predicting Recovery from an Eating Disorder

Helpful clues may lie in measures of temperament.

Recovery from an eating disorder depends upon so many factors that its course is always difficult to predict. Dr. Cristina Degura-Garcia and her colleagues note that outcome is primarily assessed in physical and behavioral domains, while emotional and cognitive symptoms are often not taken into account, and the impact of Axis II comorbidities may further complicate matters. A patient’s temperament may predict long-term modifications of eating disorders after treatment, according to Dr. Degura-Garcia and her co-workers (BMC Psychiatry 2013; 13:288).

To examine the effects of temperament on outcome, the research team studied women diagnosed in the outpatient clinic with anorexia nervosa (AN) or bulimia nervosa (BN) between January 2000 and December 2012 at the University Policlinic Mater Domini, Catanzaro, Italy. All the study participants underwent12 months of individual therapy (once weekly for 6 months, then every other week). Therapy nutritional rehabilitation, psycho-education, and cognitive restructuring. At the end 12 months, the patients saw a psychiatrist and dietitian twice a month for 6 months, then once a month for 6 months, and then from 2 to 6 times a year for the next 3 years.

Of the 53 participants, 38% were initially diagnosed with AN, restrictive type (ANR), 9% with AN, binge-eating/purging type (ANBP), and 53% with BN. The average age of the women was 23 years, and the average body mass index, or BMI, was 18.8 kg/m² (the average for AN patients was 15.8 kg/m², and for BN patients was 21.7 kg/m²).

Many original diagnoses changed by the end of follow-up

By the end of the study, a high percentage of the patients had migrated from their original diagnoses to a diagnosis of EDNOS. Only 23% of participants no longer matched an eating disorder diagnosis at 5-year follow-up. While temperament and measures of character dimensions failed to predict the outcome in terms of DSM-IV-TR diagnostic criteria, they could be used to successfully predict clinical changes. Specifically, Novelty Seeking, Harm Avoidance and Reward Dependence predicated subsequent change in clinically relevant EDI subscale scores. While the study had some limitations, the authors suggest that low Novelty-Seeking scores on the TCI may be a strong predictor of non-response to treatment.