Demystifying the Genetics of Eating Disorders – Part I

Today I thought I’d take the time to do an overview of what researchers know about the genetics of eating disorders and try to clear up some common misconceptions. The bulk of the content in this blog post comes from a very nice review paper published in 2011 by Drs. Stephanie Zerwas and Cynthia Bulik on the genetics and epigenetics of eating disorders. In an effort to keep blog posts short, this will be a multi-part mini-series.

When it comes to the genetics of eating disorders, there are two main questions that research ask: What is the relative contribution of genetic factors to eating disorder behaviours? And what are those genetic factors? I’ll talk about research attempting to answer the first question in this post and the second question in my next post.

In order to understand the role that genetics plays in influencing eating disorder behaviours, researchers use family, twin, and to a lesser extent, adoption studies.


In family studies, researchers are typically asking What is the probability that a relative of someone with an eating disorder will also develop an eating disorder compared to the probability of a relative of someone without an eating disorder? 

These studies can tell us if eating disorder aggregate (run) in families, but they can’t tell us if that aggregation is due to genetic factors or due to a shared environment. We simply can’t figure that out using this study design. It could be either one or a mix of both.

What do we know about eating disorders from family studies? 

We know that the likelihood of a first-degree relative of someone with anorexia or bulimia is much more likely to have also experienced an eating disorder during their lifetime than first-degree relatives of healthy controls.

Here’s an important thing to remember though: “much more likely” is a probability statement.For example, first-degree relatives of AN patients are 6-10x more likely to report lifetime prevalence of AN than first-degree relatives of health controls. That means that if the lifetime prevalence of AN in first-degree relatives of healthy controls was 0.6%, we’d expect somewhere between 3.6-6% of first-degree relatives of AN patients to also report suffering from anorexia. That leaves 94-96.4% of relatives unaffected (at least with AN).


Twin studies on the other hand are able to give us an idea about the relative contribution of genetic factors to a particular behaviour.

How? Well, identical twins share roughly 100% of their genes and fraternal twin share roughly 50% of theirs. Say a behaviour was 100% genetic and it was governed by one gene. In that case, we’d expect that if one identical twin showed that behaviour, the other one would, too (and vice-versa), but for fraternal twins, if one twin showed the behaviour, there would only be a 50/50 chance that the other one did, too. Conversely, if a behaviour wasn’t genetic at all, then we wouldn’t expect to see this pattern and there would be no difference between the likelihood of identical and fraternal twins exhibiting the same behaviour, because the genes they’d have wouldn’t matter.

In reality, things are way more complex, but we can take a large sample of both types of twins and compare the rates of concordant identical and fraternal twins for eating disorders and see how those numbers compare with our ideal scenarios (like the ones I described above).

Twin studies enable us to estimate the heritability of a particular trait, in our case, eating disorders.

Heritability estimates describe the proportion of the variation between people that is due to genetic variation for a specific population at a specific point in time. Shared environment estimates describe environmental factors experienced by both twins that lead twins to be more similar. Unique environment estimates describe environmental factors that lead twins to be more dissimilar (eg, events experienced by only one twin, or differential reactions to the same experienced event).

Heritability calculations involve lots of assumptions and so, like most things, they are not perfect (and the values depend on the sample size, too). But they do give us an idea of the extent to which genetic factors contribute to eating disorders.

What do we know about eating disorders from twin studies?

Studies have reported the heritability for anorexia nervosa and bulimia nervosa somewhere between 33-84% and 28-83%, respectively. Interestingly, the rest of the variability that we seen in the behaviour seems to be due to unique–and not shared–environmental factors.

Heritability is a concept that is commonly misunderstood, and it is easy to understand why: it is complicated and confusing. Here are some things you should remember, though:

  • It does not apply to individuals. You can’t say your eating disorder is caused 60% by your genes and 40% by the environment because you read a study that showed a heritability of 60%. Heritability is a population measure. It has nothing to say about individuals. Nada. 
  • It is not static. Heritability for a particular behaviour (like, say, purging) isn’t necessarily static. It can vary with time within a population and between populations, too. This is because it depends on the genetic factors that are occurring/present in the population and on the environment.

Think about it like this: in a stable and homogeneous environment, any variation we see in a trait (like height) will be due to genetic factors, but in a heterogeneous environment where access to food is variable (for example), heritability will decrease because now environmental factors will play a role in explaining why some people are tall and some are short. The really important this to remember is that this is true even if the genetic factors that control height are completely unchanged, even if it is the same number of genes acting in the same way.

  • High or low heritability doesn’t tell us anything about the number of genes or the kind of genes that are involved. More importantly, high heritability doesn’t mean the disorder is incurable or the behaviour is unchangeable because the value doesn’t take into account how extreme some treatment interventions can be.

Here’s a nice example to illustrate how confusing the concept might be (source):

Heritability sounds simple, but it can act in unexpected ways. For example, if everyone in a population has the same allele [think a version of a gene, or just gene] for a trait and shows little variation (differences) on that trait, then the heritability for that trait is zero. It is zero because that trait has no genetic variation. One example is hair color among Eskimos (N. Carlson & W Buskirt, 1997). The whole Eskimo population appears to have the same alleles for hair color, so for Eskimos, the heritability for hair color is 0.00, even though the color is under strong genetic control! It’s 0.00 because there is no genetic variation for hair color.


Adoption studies are good for understanding the role of the environment, particularly shared environment, on behaviour. Researchers compare how similar biological and adopted children are to their parents. If biological parents are more similar than adoptive parents, we can conclude that genetic factors contribute to the similarity. On the other hand, if adoptive parents  are more similar than biological parents, then we know that shared environmental factors likely contribute to that similarity.

To date, one adoption study has been done in the field of eating disorders (Klump et al., 2009). In that study, researchers looked at disordered eating symptoms and not eating disorders because of the low prevalence of eating disorders in the population.

What do we know about eating disorders from adoption studies?

The findings from the adoption study confirm what researchers have found in twin studies. Namely, heritability ranged between 59-82% and shared environmental factors didn’t contribute significantly to the variability observed in disordered eating behaviours. 


In summary, evidence to date suggests that 50-80% of the variation we seen in eating disorder behaviour (at least in the populations we’ve studied, predominantly those of European ancestry) is due to genetic factors and about 20-50% is due to unique environmental factors (and a tiny bit is due to shared environmental influences).

Next up, we’ll look at what specific genes have been implicated in eating disorders. (Warning: we know a lot less about that!)


Zerwas, S., & Bulik, C. (2011). Genetics and Epigenetics of Eating Disorders Psychiatric Annals, 41 (11), 532-538 DOI: 10.3928/00485713-20111017-06


Tetyana is the creator and manager of the blog.


  1. This a great post Tetyana! Thanks so much for citing our article. One of my favorite examples of how heritability estimates interact with the larger societal environment comes from smoking research.

    Kendler, Thornton (my colleague at UNC!) and Pederson did a twin study examining heritability estimates for smoking behavior in Swedish men in women. What they found was really interesting. Smoking heritability estimates for men were ~65% no matter when they were born. But for women there was an interesting cohort effect. The estimates varied depending on what year women were born.

    For the women born before 1925, heritability estimates for smoking were very low: smoking was mostly due to environmental factors. For women born after 1940, heritability of smoking was similar to the estimates that they saw in men (63%). So what happened between 1925 and 1940? There were dramatic societal shifts in how acceptable it was for women to smoke. Before 1925, smoking wasn’t really social acceptable for women and very few smoked (low variability in the behavior). For women born after 1940, smoking was now socially acceptable for women to do. There was greater variability in smoking in women, and the variability in smoking was now better explained by genetic factors. It’s a dramatic example of how social norms interact with genetic factors.

    I don’t think we have enough power to look at cohort effects in twins with eating disorders. But it is an interesting thought experiment. If we lived in a different societal environment, one that praised rounder, softer bodies), would heritability estimates for eating disorders be lower? I don’t think we have the answer to that question.

    However, it doesn’t stop me from advocating for a society that celebrates all body types. Anything has to be better than the current toxic culture that makes us hate and shame our bodies in order to sell us products to “fix” ourselves.

    • That’s a great example of the balance between environmental and genetic factors in contributing to the variation in behaviour that we observe. I think the hardest part is explaining that heritability is a population concept, that it depends on the environment, and that it doesn’t tell us anything about the genetic factors involved. Context matters.

      “However, it doesn’t stop me from advocating for a society that celebrates all body types. Anything has to be better than the current toxic culture that makes us hate and shame our bodies in order to sell us products to “fix” ourselves.”

      I agree completely. To me, the two issues are almost (not entirely, though) separate (re eating disorders and body image). I feel that a lot of awareness groups emphasize body image way too much, conflating the sort of “I feel fat but really I feel anxious I just can’t say that/don’t realize it” with body image issues that lead to yo-yo dieting. They are not wholly separate issues, but I think if we were to draw a Venn Diagram, there wouldn’t be as much overlap as some awareness groups and certainly media outlets lead us to believe.

      I feel that body image is a convenient post-hoc rationalization/explanation for why many individuals with eating disorder engage in eating disordered behaviour. It is an explanation that make sense to the individual and to their relatives/friends. It certainly makes more sense than “restricting makes me feel less anxious” or “purging calms me down.”

      • I agree with you partly. I think the emphasis might be placed too much body image and the toxic media environment. But there are both practical and research reasons for this. 1. Science is hard to communicate 2. Body image and toxic media is an easy hook to get the general public to think about eating disorders but also 3. Thin-ideal internalization has been found to be a powerful predictor of who develops an eating disorder (Stice’s prevention work). As long as we have a culture that promotes the thin-ideal at all costs (and I really mean at all costs) we’ll have people who are triggered by their toxic environment.

        Ultimately, I think there is significant heterogeneity in the pathways to which people develop an eating disorder. One person’s pathway to an eating disorder might look very much like how you describe, ‘restricting makes me less anxious and I couldn’t care less about fashion magazines” But I know others for whom “i want my legs to look exactly like my favorite thin actress’s legs” (idealizing a thin ideal) was the spark that started them on that path.

        I want us to find ways to explain and translate the science of what we know about eating disorders as well. I just worry that we have to be careful and deliberate in how we do it because as we’ve seen in some papers (e.g., Easter), people can hear “eating disorders are genetic” or “eating disorders are brain disorders” and then assume they are they are genetically determined or not treatable. The last thing I would want to happen is for our biological messages to raise stigma instead of reduce it. So figuring out the right public campaign message that resonates with people and clarifies these devastating and complicated disorders is essential. It’s also why your blog and Carrie Arnolds are such a service so thanks for everything that you do.

        • In retrospect, I don’t think I was very clear.

          I think very few people at the early stages of their eating disorder will say they are doing it because it makes them less anxious. I think the vast majority will attribute it to wanting to be thin. My question is, for all of those people, what %age are just unaware of the physiological effects and psychological benefits of restriction and claim it is because they want to be thin but really the situation is much more complex. I’m not saying that everyone who claims their ED was sparked by wanting to be thin are lying to others and themselves and they just do it because it makes them less anxious. Not at all. But what I’m trying to say is that thin culture is a very convenient and seemingly “obvious” culprit, but I think the issue, for many, is more complex.

          I don’t doubt that dieting and thin culture trigger many individuals. If dieting wasn’t prevalent, I would bet less people would realize that restricting is anxiolytic and a convenient way to avoid feeling (for a while) or sooner or later realize that bingeing and purging has similar effects. To me, this is the same as with something like drug addiction. If drugs weren’t available, less people would probably be addicted because less would come across their powerful effects and get addicted.

          It is not something that would be easy to test, because I think the explanations that we make up for our behaviours, to try to make sense of our behaviours, aren’t always the reasons for those behaviours. Thanks prefrontal cortex!

          I think the fascinating question to me is why is it that some people find restricting anxiolytic and others don’t. Why do some people fall into bingeing and purging and others don’t? We are all exposed to thin models, but only a small %age develop anorexia nervosa, even though many people dislike their bodies and wish to be thinner.

          Many people diet and fail because they hate dieting. Sure, they might want to look like Kate Moss or whoever is popular now, but they give up pretty quickly when they realize dieting just sucks. Why is it not the case for AN patients? That’s what’s interesting to me. I think we are getting a clearer picture of that, but, it is not complete, that’s for sure.

          I think there’s a difference between what ultimately makes individual realize that restricting and/or bingeing and purging serves or can serve some functional role in their life and the reasons WHY or perhaps what ENABLES them to actually continue restricting and/or bingeing and purging.

          I hope that makes a little sense?

          • Thanks to both of you for posting your thoughtful and extensive responses–these are questions that I mull over often, at times befuddled by the contradictory (or merely incomplete) data. Your answers remind me of why I loved (and miss) grad school so much: a rich discourse grounded in data and fueled by ample curiosity.

          • But as this shows, this kind of discourse can occur outside of grad school and outside of an academic setting! 🙂

  2. Regarding the part that genetics plays in ED I will sibmit this fact for consideration. My aunt (whom I never met) died at the age of 15 due to what Doctors then termed: “Troublesome eating.” My mother and her mother and all of the sisters were large boned and overweight. The one that passed away was tall, small boned and of course too think due to the disorder. If there was an ED in my family while I was growing up, it would have been overeating. Now I have a daughter age 15 who just this year developed ED. She participated in restricting…and in three months went from a size 8 to a size 3 and has been in a residential program for the past two months. So, when I was being questionned about whether or not my family had any ‘history’ of ED…I recalled this ‘aunt’ who obviously did. My own daughter has a huge fear of being ‘fat’…and I suppose rightfully so when she looks at her ancesotrs. I would have to say genetics DOES play a part in this disorder…both ways. Being overweight and then trying to control it by restricting.

  3. Personally I love statistics, and have taken several classes in research methods, so I’m well aware of the difficulties inherent in research like this.

    So I suspect it’s OK to point out that there are other difficulties with studies of this nature. In this case, when one talks about “eating disorders”, there’s always the question of what factors we are actually measuring when we use that diagnosis as the determinant..

    Are we really measuring a vulnerability to compulsiveness and perfectionism ? A tendency to emotional reactivity or alexithymia ? A susceptibility to anxiety or addictive behaviors in general ?

    And is there actually anything that genetically relates to food, appetite or eating that’s involved ?

    While genetics are clearly involved, eating disorders seem to provide a stage upon which many different possible factors play out, so one does sometimes wonder what studies like these are actually measuring.

    Bob J.

  4. I realise I am commenting on a post written nearly 6 months ago, but I am hoping that some of those who left comments are still open to discussing this topic..

    I read, with interest, Stephanie’s comment above where she stated that people can hear that eating disorders are ‘brain disorders’ and assume that they’re untreatable (or all genetic). I blogged on my thoughts about calling AN a ‘brain disorder’ yesterday ( I don’t find the term at all helpful, because it could imply that environmental factors are insignificant.

    Although I cannot personally identify with the idea that my AN was related to internalisation of a thin ideal, I did experience trauma in the months preceding the onset of my anorexic behaviours as a child. I am 100% confident that this trauma played a role in the development of my AN because my anxious thoughts were trauma related!

    I do not doubt that there are inherent factors which influence temperament and personality development, which in turn render an individual susceptible to developing AN. I was a very anxious and obsessive child before I experienced either trauma or AN. However, it would feel invalidating if my experiences of trauma were brushed aside and had been neglected in treatment because of the apparent genetic influence on the development and progression of AN.

Comments are closed.