Personality Traits after Recovery from Eating Disorders: Do Anorexia and Bulimia Patients Differ?

When we think about eating disorders, we tend to think about eating disorder subtypes: anorexia nervosa (AN), bulimia nervosa (BN), and binge eating disorder. A lot of previous work has shown that individuals with AN and BN tend to be anxious, depressed, perfectionistic, and harm-avoidant. Patients with AN also tend to score low on novelty-seeking, impulsivity, and self-directedness, whereas patients with BN score high on novelty-seeking and impulsivity. More recently, however, some researchers began to wonder if there was another way to categorize patients–not according to symptoms, but according to personality traits?

They identified three clusters of personality subtypes that seemed to “cut across” eating disorder diagnoses, outlined below (taken from a previous post):

Three Personality Subtypes in Eating Disorder Patients:

  1. “dysregulated/undercontrolled pattern: characterized by emotional dysregulation and impulsivity”
  2. “constricted/overcontrolled pattern: characterized by emotional inhibition, cognitively sparse representations of self and others, and interpersonal avoidance”
  3. “high-functioning/perfectionist pattern: characterized by psychological strengths alloyed with perfectionism and negative affect

However, that research was done in ill patients, and so the question remained: Do these personality clusters persist after recovery? This is the question that Angela Wagner and colleagues asked in their study, published in 2006.

Specifically, they asked:

  1. What are the personality traits in individuals who have recovered from an ED?
  2. Are there personality-based clusters in individuals recovered from an ED? And if so, do they resemble those present in ill individuals?

For their study, Wagner et al recruited 47 healthy controls and 60 women recovered from an eating disorder. The ED subtype breakdown was as follows: 21 with restricting type anorexia (RAN), 20 with bingeing/purging type anorexia AND women with both anorexia and bulimia diagnoses (BAN), and 19 women with bulimia nervosa (BN).

The average age of the participants was in the mid-20’s. The average age of ED onset was between 15-17, average length of the ED ~3.5 years, and the average length of recovery was ~3.5-4 years.

Recovery in this study was defined as >1 year of no bingeing, purging, restricting, or excessive exercise; regular menses; >90% of average body weight; and no psychoactive medications (like antidepressants).

To answer the first question (What are the personality traits in individuals who have recovered from an ED?) Wagner and colleagues assessed mood and personality characteristics, as well as the prevalence of axis I and II disorders.

Below is a summary of what they found:


  • Recovered ED subgroups did not differ in depression, anxiety, obsessions and compulsions, harm avoidance or persistence
  • All recovered women in the ED groups scored significantly higher on those scales compared to healthy controls
  • Healthy controls scored higher than all recovered women on the self-directedness measure
  • There were no differences between the RAN, BAN, and BN subgroups on any Axis I (major depressive disorder, obsessive-compulsive disorder, PTSD, etc..) or any Axis II disorders (personality disorders)

The researchers then wanted to see if there were any emerging personality-based clusters. Their analysis revealed two clusters in the recovered individuals (for 55 out of the 60 participants).


Women in Cluster 1 scored higher on:

  • novelty seeking (but same as controls)
  • impulsivity
  • “self-transcendence” (“the extent to which a person identifies the self as an integral part of the universe as a whole”) (than Cluster 2 and control women)

Women in Cluster 2 scored higher on:

  • harm-avoidance (than Cluster 1 and control women)

There were no differences between the clusters for:

  • Axis I and Axis II disorders, or ED subtypes
  • Age, lowest BMI, highest BMI, age of ED onset, duration of illness, length of recovery and age of recovery.

In line with previous research, this study confirmed that many traits persist after recovery from an eating disorder, particularly anxious, depressive, and obsessive symptoms. On a positive note, recovered participants had lower levels of mood disturbances compared to published findings of ill patients.

Particularly interesting was the finding that individuals recovered from EDs were more similar than different, regardless of their ED diagnoses:

This was true for personality and temperament scores, as well as core ED symptoms, mood, and lifetime Axis I and Axis II disorders. […] In contrast, there is evidence for differences in ill ED subgroups in terms of personality variables.

This raises at least two possibilities (which are not mutually exclusive).

One, there might be common factors that are associated with a good outcome. The authors note that these might include high harm avoidance and low rate of Cluster B personality disorders (histrionic, borderline, narcissistic, and antisocial PDs), social phobia, generalized anxiety disorder, and alcohol dependence.

(As an aside: Cluster B personality disorders were present in 5-10% of women in this sample, but range from 20-30% in ill women, according to previous studies. Likewise, no recovered BN and 16% of recovered BAN women reported lifetime alcohol abuse/dependence, whereas alcohol abuse/dependence ranges from 20-40% in ill BN and binge-purge AN patients, according to previous studies.)

Two, it could be that these traits are exaggerated as a result of restricting, bingeing, and/or purging. I know that for me, some of these traits get exaggerated or suppressed depending on whether I’m mostly I’m bingeing and purging or mostly restricting (to an extent that others around me can tell).

In their discussion, the authors questioned whether subtyping EDs based on diagnostic groups was justified, and whether subtyping based on personality-traits made more sense:

Studies reporting a high crossover rate between diagnoses and subtypes also call into question the usefulness of diagnostic subtypes. [Two clusters emerged] in an attempt to identify other behavioral groupings of patients recovered from ED. […]

Women in Cluster 2 seem to be more inhibited, with high scores of harm avoidance and state anxiety. Women in Cluster 1 can be described as disinhibited and are characterized by greater impulsivity, but also had some degree of anxiety and obsessionality.

There are only a few studies in the literature that have described personality- based clusters in participants with EDs. We replicated in the current study two of those previously described clusters, the impulsive and the inhibited, anxious subgroups of ED.

Note that this study identified two previously described clusters but not the third: “high-functioning/perfectionistic.”

The authors also pointed out that individuals in this study recovered, on average, within 6 years of being ill. “These data are similar to Strober et al., who described that “the accelerating force of recovery runs fairly steady through year 6, at which point the likelihood of recovering decelerates and reaches a plateau after 10 years.””


These findings are especially important because they might help guide treatment. It is conceivable that different treatment approaches would work for better individuals who fall into different personality clusters. Right now, treatments are mostly being tailored for specific ED subtypes. Perhaps there is benefit in shifting focus and trying to identify treatment approaches that work for particular personality-clusters instead?

Readers, what are your thoughts?


Wagner, A., Barbarich-Marsteller, N., Frank, G., Bailer, U., Wonderlich, S., Crosby, R., Henry, S., Vogel, V., Plotnicov, K., McConaha, C., & Kaye, W. (2006). Personality traits after recovery from eating disorders: Do subtypes differ? International Journal of Eating Disorders, 39 (4), 276-284 DOI: 10.1002/eat.20251


Tetyana is the creator and manager of the blog.


    • Yeah, I understand your dislike. What they could’ve done (and others do this) is have a mixed group of recovered patients (those on antidepressants/other drugs and those off them), and looked at whether there were any differences between those two groups. Some researchers determine there were no differences (or negligible differences) and just group the two together.

  1. I really wish they’d limited the study population to those w/more time out of treatment. It’s hard to say that those who are recovered have persistent personality traits when it’s only been a few years since treatment for some of the folks included in this sample. Just developmentally, if most of the folks in this sample were in their teens when they were treated and w/in 4 years of that to be in this study, it’s hard to know what might change as they have more recovery time. As someone who went through treatment in the late 80’s/early 90’s, even though I would’ve met criteria for inclusion in this sample in my early 20’s, I wouldn’t say I was “recovered” cognitively or emotionally until my late 20’s or early 30’s. The physical behaviors ceased well before other aspects dropped off. I really, really wish someone would do a study of people with long term survival from eating disorders. There were a ton of us going through treatment programs in the late 80’s and early 90’s – and while the treatment modalities were different then, there’s really nothing out there looking at our lives in our 30’s, 40’s, 50’s. Like what predicted some of us would do ok – regardless of treatment modality – and others are still struggling or died? What personality traits stayed stable once out of our 20’s? What factors existed among those of us who didn’t relapse vs those who are now showing up for treatment again in mid-life? I really wish my generation was included in research about AN survival and “recovery”

    • Yes, you raise very very good points RH. I agree with you. I think one of the main roadblocks is money, time, and follow-up (the longer the follow-up, the more people leave/disappear). The roadblocks are often structural, really. I’ll look for any really long-term studies that looked at personality variables, but as you point out, they might not even exist (I haven’t research the topic too much, personally).

Comments are closed.