Characteristics of Women with Midlife-Onset Eating Disorders

Since the late 1990’s, Remuda Ranch Program for Eating Disorders has experienced a 400% increase in patients 40 years of age and older, according to the authors of this paper. However, we don’t really know what the similarities and differences are between women who develop eating disorders in adolescence and those who develop their eating disorders in midlife (40-65 years of age).

It has been theorized that EDs in midlife may be triggered by midlife transitions, such as loss of parents, siblings, or children; divorce; traumatic illness; and empty nest syndrome (Harris & Cumella, 2006; Maine & Kelly, 2005; Shellenbarger, 2004). […] Two quantitative studies found a high correlation between the fear of aging and disordered eating in older populations (Gupta, 1995; Lewis & Cachelin, 2001).

In this paper, Edward Cumella and Zina Kally present a summary of 50 women who first developed eating disorders at the age of 40 or above and were admitted to inpatient treatment at Remuda Ranch. They “assessed patients’ sociodemographics, severity-of-illness, comorbid diagnoses, personality profiles, and short-term treatment outcomes.”


Demographic Characteristics

  • Mean age: 49.5 (range: 41 – 63)
  • Mean ED onset age: 45 (range: 40-62)
  • Mead ED duration: 4 years
  • Previous treatment: 40% had prior inpatient treatment
  • Race: 98% of patients were White
  • Marital Status: 58% were married, 28% divorced/separated, 10% never married and 4% widowed
  • Education: 54% had college or post-graduate degrees, 22% had some college education, 20% had a high school diploma, 4% did not complete high school

Cumella and Kally wanted to find out more specifically how this group of women differed from adolescents with eating disorders in terms of diagnosis characteristics, self-harm, suicide, and sexual abuse histories, as well as psychiatric comorbidities and general ED psychopathology.


Diagnosis Characteristics

  • 38% were diagnosed with anorexia nervosa, restricting type (ANR)
  • 20% with anorexia nervosa, binge-eating/purging type (ANBP)
  • 8% with bulimia nervosa (BN)
  • 34% with EDNOS
    • 76% of EDNOS patients engaged in restricting only, no bingeing/purging

Self-harm, Suicide & Abuse

  • 22% of patients had a history of self-harm
  • 28% had previously attempted suicide
  • 64% had a history of sexual abuse

What about psychiatric comorbidities?

Comorbid Disorders (Axis I)

  • 94% were diagnosed with comorbid Axis I disorders
  • 86% had unipolar depression, with 60% diagnosed with major depressive disorder and 28% with depressive disorder not-otherwise-specified
  • 62% had one or more anxiety disorders
  • 18% had substance abuse/dependence issues
  • 6% had bipolar disorder

Comorbid Disorders (Axis II – Personality Disorders)

  • 24% had Cluster B disorders/traits [Antisocial, Borderline, Histrionic, Narcissistic]
  • 20% had Cluster C disorders/traits [Avoidant, Dependent, Obsessive-Compulsive Personality Disorder]
  • 2% had Cluster A disorders/traits [Paranoid, Schizoid, Schizotypal]

These findings are informative, but they become really interesting once you compare them to what has been published about adolescents with eating disorders.

For example, on the Eating Disorders Inventory Questionnaire (EDI-2), which is commonly used during diagnosis and assessment, this sample of women scored rather moderately on the drive for thinness, bulimia, and body dissatisfaction measures.

This is strikingly different from admission EDI-2 scores of younger inpatient ED populations, where the three primary scales are typically elevated into the ED range (Cumella, Kally, & Wall, 2007; Garner, 2004). Instead, on the EDI-2, patients admitted in the ED range on the ineffectiveness, perfectionism, interpersonal distrust, and asceticism scales, suggesting a possibly greater focus on these issues than the more traditional ED issues of body image and binge/purge behaviors.

Notice that the majority of the patients did NOT exhibit any binge-eating or purging behaviours, something that I don’t think we would have seen if this had been a sample of women who first became sick during adolescence and either never truly recovered or relapsed in middle age. The difference between 8% with bulimia nervosa in this sample to 40% in teenagers and young adults (Blinder, Cumella, & Sanathara, 2006) is quite striking.

The majority of midlife-onset ED inpatients (64%) appeared to have ED diagnoses of ANR and EDNOS with only restricting behaviors. This predominance of pure restriction resembles ED patients in the earlier stages of an ED (Fairburn & Brownell, 2002).

Interestingly, comorbid depression and anxiety was similar in frequency but reduced in severity compared to what has been reported in adolescents with eating disorders. Substance abuse/dependence were also less frequent in this sample than in teenage and young adult ED inpatients, and the drugs of choice tended to be of the sedating (as opposed to stimulating) type.

Cluster B (antisocial, borderline, histrionic, narcissistic  personality disorders were really high in this sample compared to what is seen in adolescents. This is especially interesting because those disorders and traits tend to be primarily seen in patients with bulimic symptoms (which were relatively infrequent in this sample). Conversely, Cluster C personality disorders (avoidant, dependent, obsessive-compulsive (not to be confused with obsessive-compulsive disorder!)) were relatively rare compared to what has been reported in younger ED populations.

The authors suggest these findings warrant additional research. Perhaps pursuing this research will provide us with more information about the differences between adolescent-onset and midlife-onset eating disorders in etiology, phenomenology, and treatment.

Self-harm was less common though sexual abuse history was much more common than in younger ED populations. I would love to find out at what age these women first began self-harming and when they experienced sexual abuse (was it in adolescence or closer to the time of their eating disorder?). Hopefully this, along with the finding on personality disorder comorbidities, is something that will be explored in future studies.

Study Limitations

As always, I like to point out the limitations of a study. Please keep in mind this was a small sample (just 50 women, though of course, that’s to be expected given the study was about midlife-onset of EDs). Secondly, the patients were not interviewed for the study, instead their chart data was reviewed sometimes many years after the patients were in treatment. There is no comparison group to young patients in the same facility, and the sample consists of treatment-seeking patients in a faith-based facility. 

Whether these findings are generalizable to the general population of women with midlife-onset eating disorder remains to be observed. In particular, I wonder about midlife-onset of bulimia, and about the characteristics of non-treatment seeking women (and men, of course). How do they differ from these women and from adolescents with eating disorders (or adolescent-onset eating disorders).


Cumella, E., & Kally, Z. (2008). Profile of 50 Women with Midlife-Onset Eating Disorders Eating Disorders, 16 (3), 193-203 DOI: 10.1080/10640260802016670


Tetyana is the creator and manager of the blog.


  1. Tetanya, thanks for putting the spotlight on this topic. I was thinking about suggesting this topic to you earlier this week actually.

    It seems like these people developed their eating disorder for the first time at 41 and over. That makes me wonder if at least some of them perhaps had been dealing with disordered eating of varying severity at least previously at some points in their lives. This article poses the question much better at what i am trying to get at-
    “Eating Disorders in Older Women Does Late Onset Anorexia Nervosa Exist”

    I do think that eating disorders can and do develop at any age, but still find the above question interesting.

    This is another related article.
    They seemed to suggest that grief counseling might be beneficial for middle aged patients aged over 35 in inpatient. I did wonder what they actually meant by this grief counseling. Is it aimed at dealing with the loss of what the ed has caused, if so i can quite understand that. Many people of this age are dealing with a lot of loss caused by the ed. It can start a young age, but can just get worse as time goes on. Or perhaps they are talking about well the obvious, but i would think that it should at least cover the former. That is interesting though and not something i’ve heard of before at least in relation to ed’s.
    This is another one of Cumella’s articles.

    This started for me in my teens. I’m not totally sure how someone might feel who developed this over 40 for the first time. There is likely though a lot more going on than just fear of aging.

    • Hi missr,

      Thanks as always for your comment. It is nearly impossible to verify that people who claim to have no ED history actually have no ED history. The paper specifically looked at women from 40-65, which is why the mean ED onset was in the 40’s. They specifically excluded women who have had eating disorders as teenagers or young-adults. Of course, like I said, it is hard to verify, but, it is like a lot of things in research–we have to go by what participants tell us. I agree that they are rare (and this sample was collected over a long period of time), but I am sure that adult/midlife-onset EDs exist in a sense of them starting at that age with no history of EDs or disordered eating before.

      The paper you linked to looked at patients who presented with EDs in their midlife and found that many have had long-standing EDs (the sample was also very small, just 32 people), but this paper specifically excluded those people, so the studies are differently designed.

      I checked the study you linked to and this is what it said, it doesn’t such much about what the grief/loss group therapy was:

      “Although middle-aged and younger patients rated the Renfrew groups as equally helpful overall, there were some age-related differences in specific group ratings. Women 35 and older rated the Grief and Loss group as significantly more helpful than younger patients (p < .05), a finding that supports Zerbe’s (2003) thoughts regarding the significance of loss for midlife eating disorders. In addition, middle-aged patients rated family therapy as significantly more helpful than younger patients (p < .05). This finding is hard to interpret since family therapy issues may be quite different in the two groups. Specifically, the younger patients typically have parents involved in family therapy, whereas the midlife patients often have spouses and their own children involved in the family therapy process." Unfortunately, I don't have access to Editrice Kurtis journals. And I don't know anybody who does! (So if anyone reading does, let me know, please). My friends at Harvard, University of Chicago, Northwestern, Durham (UK), Berkeley and a bunch of other places don't have access. It is very annoying. "There is likely though a lot more going on than just fear of aging." I agree, there must be more than just that. Probably similar predisposing factors like anxiety and depression, coupled with some event (like divorce or something?). I'm not sure, just hypothesizing. Cheers, Tetyana

  2. Thanks Tetanya

    The Editrice Kurtis journal is a free to access journal. You can register for free and then download articles for free as a non subscriber.

Comments are closed.