Are There Any Meaningful Differences Between Subthreshold and Full Syndrome Anorexia Nervosa?

I see this on a daily basis: patients with subthreshold eating disorders feeling invalidated and “not sick enough.” They are struggling so much, but maybe they still have their periods, or maybe their weight isn’t quite low enough, and so they often (but not always, thankfully) get dismissed by doctors, other healthcare professionals, and insurance companies. Do you think you really need this treatment, maybe you can just focus on eating healthier? You know you are not fat, you are perfectly healthy! Just be happy! Or, Sorry, we can’t cover this psychological treatment because you don’t fit the full diagnostic criteria. 

Why do we draw a line between ‘threshold’ and ‘subthreshold’ at arbitrary numerical criteria?

No doubt numbers are important for medical treatment: someone with a very low BMI might have considerably more physical complications that need to be taken into account during treatment than someone with a not-so-low BMI. But do these arbitrary weight and numerical criteria really say as much as we think they do? Is BMI or menstruation really a valid way of demarcating between full and subthreshold patients? And most importantly, do they say anything at all about psychological well-being?

Luckily, in 2012, a study that asked this very question was published in the European Eating Disorders Review. Daniel Le Grange and colleagues compared eating-related and psychopathology measures between 59 anorexia nervosa and 59 subthreshold anorexia nervosa women. They asked them to fill out questionnaires, do interviews, and even record their moods and symptoms in ‘real-time’ using the ecological momentary assessment (EMA) tool (click the link to find out more about it, it is pretty cool!)

In their study, subthreshold AN (or EDNOS-AN) women had to meet all DSM-IV criteria except (i) have a BMI between 17.6-18.5 OR (ii) not experience amenorrhea (loss of periods) OR (iii) have no body image disturbances and no intense fear of becoming fat. Basically, they had to fit 3 out of 4 diagnostic criteria. 


Sample demographics 

  • No differences in age (mean age ~25, range 17-58)
  • Average BMI for AN: 16.6 and EDNOS-AN: 17.7
  • Participants were mostly Caucasian, single, and with some college education


No differences between the AN and EDNOS-AN groups on: 

  • Levels of depression, anxiety, and perfectionism
  • Personality pathology (self-harm, affective liability, stimulus seeking)
  • Harm avoidance, self-directedness and impulsivity measures
  • No significant differences were found for any measure

Eating-disorder related behaviours

No differences between AN and EDNOS-AN groups on:

  • Any Eating Disorder Examination measures (including: eating concern, weight concern, shape concern, vomiting episodes, laxative use episodes, exercise days, objective binge eating episodes, subjective binge eating episodes, and restraint)
  • Presence of mood, anxiety, or substance use disorders
  • Preoccupations and rituals
  • No significant differences were found for any measure

Ecological Momentary Assessment (EMA)

No differences between AN and EDNOS-AN groups on:

  • Mean mood levels (positive or negative) or mood variability (positive or negative)

Differences between AN and EDNOS-AN groups on:

  • Frequency of binge eating and purging behaviours (significantly higher in the AN group)
  • Frequency of thigh and joint body checking behaviours (significantly higher in the EDNOS-AN group)

Yes, you read all of that correctly: there were no differences between the two groups on any of the measures that were evaluated with the exception of bingeing and purging frequency (higher in the AN group) and body checking behaviours (higher in the EDNOS-AN) group.

With few exceptions, and across multiple methods of assessment, AN and EDNOS-AN were mostly indistinguishable. There were no differences on any demographic variables, self-report measures of mood, anxiety, perfectionism or other personality measures. Nor were there differences on interview measures of eating and general psychopathology. Although the diagnostic groups also did not differ in terms of EMA mood, the only statistical differences to emerge were for EMA eating disordered behaviours. Participants with AN reported higher rates on two diagnostic items, that is, binge eating and purging, compared with those with EDNOS-AN.

As always, it is important to think about the limitations and not extrapolate these results further than the data allows. The participants were overwhelmingly Caucasian (96-97%), all female, and all from the mid-West USA. A lot of things could not be verified, such as the lifetime history of eating disorders (whether those with EDNOS-AN had a history of full AN, for example). The authors also didn’t compare how the ED impaired school/work functioning, which would’ve been interesting, though it is not really a limitation but perhaps a consideration for future studies.


Taken together, our findings overwhelmingly point to the similarities between AN and EDNOS-AN as the latter was defined in this study. The many shared features between AN and EDNOS-AN highlight the clinical significance of EDNOS. […]

There is little evidence that participants with EDNOS-AN were any different from those with AN. Therefore, our results confirm the now accepted notion that menstrual status is probably not a helpful diagnostic marker for AN (Attia, Robero, & Steinglass, 2008) and also challenge the generally accepted cut point of 85% of ideal body weight (or BMI 17.5 ) for a diagnosis of AN.

So whenever you are too enraged to speak because someone is insinuating that EDNOS-AN is not as severe as AN and thus doesn’t warrant treatment, just point them here. If someone says that those with EDNOS-AN aren’t struggling as much as those with AN, or don’t know what AN feels like, or should just go eat a sandwich, just point them here. Or point them to this old post, where I compare mortality rates between BN, AN, and EDNOS patients. 


DSM-IV-Defined+Anorexia+Nervosa+Versus+Subthreshold+Anorexia+Nervosa+%28EDNOS-AN%29&rft.issn=10724133&;bpr3.tags=Medicine%2CPsychiatry%2C+Eating+Disorders%2C+Anorexia+Nervosa%2C+Subthreshold+Anorexia+Nervosa%2C+EDNOS%2C+Psychology%2C+Nosology">Le Grange, D., Crosby, R., Engel, S., Cao, L., Ndungu, A., Crow, S., Peterson, C., Mitchell, J., & Wonderlich, S. (2013). DSM-IV-Defined Anorexia Nervosa Versus Subthreshold Anorexia Nervosa (EDNOS-AN) European Eating Disorders Review, 21 (1), 1-7 DOI: 10.1002/erv.2192


Tetyana is the creator and manager of the blog.


  1. Thank you for the work you do through this website. As a person in recovery who hopes to work in the field, I find it really valuable.

    • You are very welcome! I’m really glad you like it! You can’t see it, but I’m smiling right now. I love doing it! Life is so awesome when you get to do things you like and people appreciate you for it!


  2. My personal feeling is that there are two components to having an eating disorder: the psychological component and the physical component. (I am not enamoured by the notion that weight restoration alone is a cure for anorexia nervosa either).

    Whether or not a person is underweight has implications for treatment (e.g. weight gain!), as you know. The rate of weight loss (if there is indeed weight loss) is also relevant to physical risk.

    All people with severe psychological difficulties ‘deserve’ and warrant treatment, irrespective of their weight.

    But why call something anorexia nervosa when it is not anorexia nervosa? I can understand that people with normal/higher weight eating disorders can feel invalidated and neglected by health professionals, but historically, anorexia nervosa has described a condition that is accompanied by underweight – and often gross underweight. Very low weight is in itself a big risk to physical and psychological health.

    My frustrations surrounding diagnosis come from a different angle. For over 30 years I fulfilled the weight criterion for diagnosis of anorexia nervosa (i.e. I was very underweight), but I didn’t exhibit ‘weight and shape concerns’. Because I didn’t report ‘feeling fat’, fat phobia or weight phobia, I was denied treatment. I have an eating disorder that is driven mainly by OCD and FOOD phobia, and no-one has been able to treat me effectively. I am still underweight, hate being underweight, but struggle to bring myself to eat more because of great anxiety surrounding food and eating.

    If a person is struggling with eating in some form or another (including purging) and has psychological difficulties, they need treatment. Period. But I do feel that anorexia nervosa should be used to describe an eating disorder wherein there in low body weight and that EDNOS should be taken very seriously. Most people with eating disorders have EDNOS anyway.

    • Well, exactly. I think expanding the AN (or BN) diagnoses to encompass a huge chunk of those that fit into EDNOS is not ideal because it will create the exact same problem: an uninformative diagnosis.

      But there are better ways to stratify and group individuals with eating disorders than by arbitrary numerical criteria. What is the basis of 85% of “ideal body weight” (whatever the hell that means, usually taken to be a BMI of 20, it seems, in treatment).

      Weight is an important component from a physiological and psychological perspective, as well. But a BMI <17.5 is not a good metric, at all. I think weight loss rate and absolute weight loss are more telling for both the physiological and psychological perspectives.

      I believe that diagnoses should be informative, and right now, it is a mess. A huge mess. If I say I have EDNOS, it tells the doctor or the nurse absolutely nothing about my symptoms. I don't think we should recreate that by expanding AN too much, I just think we need to think critically about the most informative way to group ED patients so that it facilitates an understanding of best evidence-based treatments, prognosis, medical complications, possible comorbid disorders, and so on.

      • I absolutely agree that the current diagnostic criteria are crap – and that includes the precise BMI value and the 85% of ‘ideal’ weight! My understanding is that DSM V is better, yet still not perfect.

        You’ve probably seen me moan incessantly on Facebook about the diagnostic criteria for AN – mainly in the context of the body image dimension. If I tell someone I have AN, they expect me to think certain things (e.g. that I somehow desire thinness) and to be utterly weight phobic – which are not true. I basically get stuck in patterns of behaviour that modulate my anxiety and depression and lead to inadvertent weight loss!

        • They are marginally better. They are not trying to deal with the main problems, I feel. There’s lots of criticism in the literature pointing out the same things I complain about. There’s still lots to fix, as far as I’m concerned. Why can’t we think outside the box? What about a spectrum or Cartesian plane type of deal, to capture the diversity and really help is figure out what treatments work for what people at what time points in their ED.

  3. Great post, Tetyana! I share your frustration with the 85% EBW guideline — it’s not only arbitrary but inconsistently applied ( Interestingly, the DSM-IV Work Group never meant it to be a “cut-off” (just a guideline), so it’s a good thing it’s being omitted from DSM-5. My work also suggests that EDNOS is typically just as severe as AN and BN (, and clinically I think too many patients find their suffering invalidated when they are diagnosed not with a specific eating disorder, but an acronym.

    I also agree with you that DSM-5 represents a big improvement (especially the inclusion of named subtypes like purging disorder), but I’m worried that being diagnosed with “FEDNEC” (i.e., the new acronym for EDNOS) is not going to feel any less invalidating…

    • Hi Jennifer,

      I have your paper on my computer, and I skimmed it, but to be honest… I went for the shorter paper just because I was in a time crunch and wanted to get a post out sooner. I definitely think people find it invalidating, and conversely, people who do get diagnosed with AN or BN often feel validated. I wouldn’t expect the 85% EBW guideline to be applied consistently (because what is ‘EBW’, really). I think rate of weight loss and absolute weight lost are much more important, both to assess severity of the ED and the possible medical side-effects.

      Honestly, though, the DSM has a long way to go before the ED diagnoses are actually reflective of clinical practice AND have good predictive value in terms of treatment outcomes and prognosis. I’m happy with the progress, but disapppointed (though not surprised) at the pace of it, if you know what I mean.


  4. I agree that changes needed to be made to the DSM-IV. The amenorrhea requirement needed to be removed and it needed rewording so there was no area of confusion when it came to people who did not present in a classical way-non fat phobics etc. I’ve been thinking of this topic and related issues over the past few days.

    I can see that mainly it seems to be an insurance issue why the 17.5 marker had to go. I’m not sure how it would have worked if it was kept in but an additional line like ” or based on the clinical picture” was added.

    With the requirement going up to 18.4? and a “healthy” BMI starting at 18.5 and some dieters aim for the lowest healthy BMI and with natural deviations this will put them into the anorexic criteria. Perhaps these people may be perfectly healthy or suffer from disordered eating, but not AN. How do you think this may work? I can see that there may be problems with it. Some of these people may not maintain their low weight for very long before they rebound may be by binging and this is their normal pattern.

    People getting good intervention is important, but I think that there can there be a problem though of labeling some people as anorexic too early. We know that some people who are diagnosed with EDNOS do go onto remit faster than people with a full syndrome ed or they may go onto develop a full syndrome ed or perhaps they may stay in this space.
    Btw, according to that study people with EDNOS also have higher recovery rates, but perhaps there is some litreature that suggests differently that i have not seen.

    I got treatment for the first time at 15 at a borderline underweight, but i don’t even know what i was diagnosed with at 15, i was just getting psychotherapy from a pyschiatrist which i found helpful, but was not aware of any diagnosis and neither were my parents’. Preventative intervention may be useful for adolsecents.

    Also back to the original topic, i agree that there may be little differences in presentation, but i do think that some of the differences between EDNOS-AN and AN may reveal themselves for some in recovery. We know for example that those with a past history of a clinical case of AN <17.5 even if they cross over into BN tend to react differently than BN's with no history of AN, and it is the previous case of AN that is likely causing this. Their previous low weight altered their behaviour and had an impact on their evolution time and these type of studies have useful insights when it comes to treatment.

    • I came across the below study, it’s on treatment response and the question i had so i’d thought i thought that I would post it

      They compared different AN/EDNOS groups and found that those with substhersold AN had the highest remission rates. I don’t know how significant the differences are.

      In the first study that i linked to in my post above they compared those with Partial AN and full syndrome AN and found that at the four year follow-up-
      The Partial AN requirements were-

      Partial anorexia nervosa
      Meets all SCID criteria for AN for the past 6 months except weight ≤ 90% of ideal body weight together with either a) amenorrhea, or b)a cognitive pattern typical of AN (body image disturbance and intense fear of fat).
      Met full criteria for AN in the past 12 months.

      At the 4-year follow-up 57% of the AN group had remitted” and at the four year follow-up “68% of the PAN group” had remitted. I’m not really sure what to make of those findings.

      Looking at treatment response etc is another way to see how these disorders might compare and there aren’t that many studies to go on.

      With the DSM changes more people will be able to get the treatment that they need and that of course is a good thing. Although, i do still wonder a bit about those points which i made in my other post.

  5. Was just perusing around the archives and came across this. Very interesting, but I’m not surprised. You’ve often highlighted the rewarding aspects of engaging in the behaviors themselves, which does seem to be someone independent from body image itself. If I had to bet, the rewarding/control aspects of engaging in an eating disorder explain more variance in ED behaviors than body image….

    Anyway, it is interesting that BMI is such a huge component of the diagnosis of anorexia. I consider the measure to be rather crude myself (it was first created as just a population comparison measure, not necessarily for health purposes). I think I barely met criterion for AN at my worst, and never met the criterion to be admitted to a hospital in the states, despite still suffering health consequences. I wonder if there is a possibility of developing a better measure of “fighting against a healthy setpoint” than BMI that can be used to more accurately categorize individuals with similar psychological symptoms and negative health consequences? Indeed, I had a friend who was still considered to be slightly “overweight” despite suffering similar symptoms and health problems as myself.

Comments are closed.