Anorexia Nervosa: Can We Blame The Season Of Birth?

You might have heard that individuals born between the months of June – August (or sometimes March – August) have a higher chance of developing anorexia nervosa. But is it true? A lot of studies have been done to investigate the question of whether a season of birth (or a month) correlates with a higher risk of anorexia or bulimia nervosa. The results are inconsistent, weak, and fraught with methodological problems.

But first, how could seasons (or the average temperature during birth, or conception) have an effect on the etiology of eating disorders? What’s the hypothesis?

There seem to be two main ideas (summarized in Winje et al., 2012):

  1. alterations in neuropsychological function as a result of sunlight exposure during gestation or postpartum, maternal infections during pregnancy, or nutritional changes (seasonal variation in nutrients, vitamins)
  2. alterations in fertility/reproductive patterns of the parents due to cultural influences, disordered eating in the mother, cyclothymia (a mild form of bipolar)

I’m going to summarize some of the studies published on this topic, just to give you an idea of where the field is with regard to this question, and then I’ll briefly touch on some of the methodological problems in many of these studies (and why you shouldn’t take it so seriously.)


1. Waller, Meyer, and van Hanswijck de Jonge (2001):

One of the first studies (but not the first, for first studies see Nielsen, 1992 and Rezaul, Persaud, Takei, & Treasure, 1996) to examine this question was published in 2001 by Glenn Waller and colleagues. They  found a statistically significant difference between 117 young females (aged 16-19) who were born between May – August and the scores on the ‘drive for thinness’ and ‘body satisfaction’ measures compared to females who were born during the other months (for those who are interested: p<0.01 and p<0.03, respectively). No relationship was found for any measures with regard to bulimic attitudes.

2. van Hanswijck de Jonge, Meyer, Smith, and Waller (2001):

In a follow-up study, the same group extended their research to younger adolescents and males. They split them by age: 23 males and 15 males in the 16-19 year-old group (to replicate the above study) and 26 males and 65 females in the 10-14 year-old group. The effects for the older group tended toward significant, but didn’t reach it (but it was significant for males). There were no differences between the season of birth and any eating disorder attitude/behaviour measures for the younger group.

3. Willoughby, Watkins, Beumont, Maguire, Lask, and Waller (2002):

Assuming it is established in the northern hemisphere that people with AN are more likely to be born in spring and early summer (warmer temperature), Willoughby and colleagues wanted to see if the trends were the same in Australia (where seasonal and temperature patterns are reversed). They found no significant patterns in the season of birth in young people with AN.

They did find a “significant” difference between restrictive AN and binge/purge type AN patients (the former were less likely to be conceived in cool weather). I put significant in quotes because the p value was 0.047. That’s basically 0.05, which is a low threshold for statistically significant findings (between two groups, if you are just doing one comparison), and means that the chances that you are thinking there’s a difference when there actually isn’t is 5%.

And anyway, what does this even mean.. the transition rates between subtypes of anorexia nervosa are pretty high (see this post, Eddy et al., 2008). So, what does this mean for the ‘findings’ above?

4. Watkins, Willoughby, Waller, Serpell, and Lask (2002):

These authors sampled 259 adolescents with AN and 149 with “other eating disorders” (227 females, 32 males). They found no seasonality effects for early onset AN (meaning individuals with early-onset AN were born all-year-round, equally). They also didn’t find anything for a group they called “other eating disorders” (ie, not anorexia nervosa).

5. Peñas-Lledó and Waller (2002):

This study found no links between temperature at birth and restrictive attitudes in 262 participants (125 women and 136 men) in Southwest Spain.

6. Munn and Klump (2002):

This study found no significant differences between 155 twins and triplets on measures of eating disorder attitudes and behaviours (using the Minnesota Eating Disorders Inventory and Eating Disorders Examination Questionnaire) . The authors concluded that the SoB bias findings are not generalizable to nonclinical samples.

7. Peñas-Lledó, Rodriguez Santos, Vaz Leal, and Waller (2003):

Out of 105 women, found that anorexia nervosa patients (20, or 43%) were more likely than bulimia nervosa patients (14, or 24%) to have been born between June-August (p=0.03). The fact that more bulimia nervosa patients were born between March-May (19, or 33%) versus anorexia nervosa patients (8, or 17%) was not significant.

8. Javaras, Austin, and Field (2011):

Well, these guys found something that went against the grain: in the US, females born in the fall were more likely to develop eating disorder behaviours than those born during other times of the year (p=0.02).

9. Winje, Torgalsbøen, Brunborg, and Lask (2012):

In an attempt to mitigate the small sample sizes of the past studies, this group analysed ~2,800 female anorexia nervosa patients born between 1975-1999 in Iceland, Norway, Sweden, UK, USA, Australia, Argentina, Brazil, and Singapore. They found no significant differences between season of birth and the development of anorexia nervosa.

10. Brewerton, Dansky, O’Neil, and Kilpatrick (2012): 

With all of this emphasis on bulimia, Brewerton and colleagues wanted to see if there are SoB effects in bulimia nervosa individuals (or women who have ever binged or purged). They conducted telephone surveys of 3,006 adult women. This study, like Jarvas et al (#8 above) found that women born during the fall, not summer or spring, had a higher likelihood of developing bulimia nervosa, or engaging in bingeing or purging (p’s around 0.03).

So, is it fall, or spring, or summer? March, August, or October?

Some studies above found those born in the fall were the least likely to have eating disorders or disordered eating symptoms. What do we make of this?

11. Vellisca, Latorre, Santed, and Realse (2013):

Back in Spain, this group found no evidence of significant variation in month or season of birth in AN patients (210 females) compared to the general population.


Tiring right? I bet you forgot all the various differences individual groups found.

I did.

I just know there were lots of weird barely significant differences that didn’t make much sense to me. A difference between younger and older eating disorder patients? A difference between anorexia nervosa subtypes? Between males and females? Those born in the UK versus Spain? And even when the groups found differences, they often didn’t seem to agree with one another (so it is spring, or summer, or fall? Or maybe winter?)

So – what about the significant findings (including the papers I didn’t mention, like Nielsen et al (1992) and Rezaul et al (1996))? Winje et al. (2012) suggest that these might have been type I errors (ie, false positives).

What were some of the methodological or statistical problems in the other aforementioned studies (Winje et al. 2008)? There are quite a few:

  • definition of seasons (not the same in every study; the months and years differ in lengths
  • accuracy of diagnoses (phone survey versus clinical assessment, different things; DSM-IV versus ICD-10, not consistent)
  • most studies did not control for co-morbid disorders
  • some sample sizes were too small (one just had 17 AN individuals!)
  • sometimes the control group wasn’t clear (there was normalization to general birth trends in the population, but it wasn’t clear enough); only one study used a control group of local, same-sex individuals (if studying females, control group must be females only too, for example)
  • possible duplication of participants (a lot of studies were done by the same group of people), the same people could have been included in several studies

Despite these limitations, Winje et al. (2008) concluded that “the studies consistently show a SoB bias for anorexia nervosa (AN) in the spring months, in both the northern and southern Hemispheres. This is especially strong for early onset and restrictive subtype of AN…”

Now, they’ve changed their mind (in the 2012) paper, upon re-analysing the information. Publishing a paper that says the opposite of what you previous wrote is hard.(Practically, getting it through the peer-review process and convincing the editor(s) to publish negative or non-exciting results is hard.) Which is all the more reason that I appreciate the following:

Unfortunately, we succumbed to the tendency to focus on the significant findings that emerged from multiple testing and specifically the findings in support of our own theoretical framework… Had these issues been understood at the time of the review, the conclusion would have differed and illustrated that the question of a season of birth bias for females with AN was still open.

Winje et al. (2012), conclude:

Demonstrating a season of birth bias would have provided an opportunity to generate hypotheses about annually fluctuating factors that could have increased the vulnerability for AN. However, based on the very small effects observed in this study, there is reason to question whether there really is a season of birth bias for females with AN.

Here’s the bottom line: when you have a lot of studies that compare a lot of different variables (different eating disorders, ages, gender, onset of ED, to different months of birth, assumed conception, temperature during birth, seasons of birth) you will get some false positives.

Some differences that barely reach the arbitrary significance threshold (p=0.05), just by chance. If you look at a study and some comparisons are barely significant, while others are tending toward being deemed “significant”, and the findings don’t agree with previous literature (so it is Spring/Summer or Fall?), or make sense (why would there be a difference between AN-restricting type and AN-binge/purge type, but no other differences, when there is so much crossover between these subtypes?), question the results. 

My impression: it is unlikely there is a seasonal birth bias for anorexia or bulimia nervosa. Is it still possible?  I think so, but we would need a few really good, methodologically consistent and statistically solid studies to  confirm this. Do I think it is worth it pursuing this question further? Personally, no. But then, I’m not the one that’s deciding.

And by the way, my birthday is at the end of November. It was cold when I was born.

PS. I know I said I was going to post about vegetarianism and how it relates to eating disorders, but, in my searches, I stumbled upon this topic and changed my mind. I will explore that topic more in the future, though, I promise.


Waller G, Meyer C, & van Hanswijck de Jonge L (2001). Early environmental influences on restrictive eating pathology among nonclinical females: the role of temperature at birth. International Journal of Eating Disorders, 30 (2), 204-8 PMID: 11449454

van Hanswijck de Jonge, L., Meyer, C., Smith, K., & Waller, G. (2001). Environmental temperature during pregnancy and eating attitudes during teenage years: A replication and extension study. International Journal of Eating Disorders, 30 (4), 413-420 DOI: 10.1002/eat.1102

Watkins, B., Willoughby, K., Waller, G., Serpell, L., & Lask, B. (2002). Pattern of birth in anorexia nervosa I: early-onset cases in the United Kingdom. International Journal of Eating Disorders, 32 (1), 11-17 DOI: 10.1002/eat.10057

Willoughby, K., Watkins, B., Beumont, P., Maguire, S., Lask, B., & Waller, G. (2002). Pattern of birth in anorexia nervosa II: a comparison of early-onset cases in the southern and northern hemispheres. International Journal of Eating Disorders, 32 (1), 18-23 DOI: 10.1002/eat.10058

Peñas-Lledó E, & Waller G (2002). Pattern of birth and eating attitudes in young adults: failure to replicate in a warmer climate. International Journal of Eating Disorders, 32 (3), 367-71 PMID: 12210652

Munn, M., & Klump, K. (2003). Season of birth and disordered eating in female college students International Journal of Eating Disorders, 34 (3), 343-348 DOI: 10.1002/eat.10195

Peñas-Lledó EM, Rodriguez Santos L, Vaz Leal FJ, & Waller G (2003). Pattern of birth in restrictive and bulimic eating disorders. Eating behaviors, 3 (4), 325-8 PMID: 15000993

Winje, E., Willoughby, K., & Lask, B. (2008). Season of birth bias in eating disorders-Fact or fiction? International Journal of Eating Disorders, 41 (6), 479-490 DOI: 10.1002/eat.20540

Javaras, K., Austin, S., & Field, A. (2011). Season of birth and disordered eating in a population-based sample of young U.S. females International Journal of Eating Disorders, 44 (7), 630-638 DOI: 10.1002/eat.20864

Winje E, Torgalsbøen AK, Brunborg C, & Lask B (2012). Season of birth bias and anorexia nervosa: Results from an international collaboration. International Journal of Eating Disorders PMID: 23070973

Brewerton, T., Dansky, B., O’Neil, P., & Kilpatrick, D. (2012). Seasonal patterns of birth for subjects with bulimia nervosa, binge eating, and purging: Results from the National Women’s Study. International Journal of Eating Disorders, 45 (1), 131-134 DOI: 10.1002/eat.20898

Vellisca, M.Y., Latorre, J.I., Santed, M.A., & Realse, J.M. (2013) Absence of Seasonal Patterns of Birth in Patients with Anorexia Nervosa. International Journal of Eating Disorders, 46: 86-88 


Tetyana is the creator and manager of the blog.