A recent BBC documentary: Diabulimia: The Worlds Most Dangerous Eating Disorder discussed the high rates of eating disorders in people with Type 1 diabetes; an autoimmune condition with typical onset in childhood and characterised by the body’s inability to metabolise sugar. The documentary has received extensive media attention and raised awareness on the potentially devastating health implications of the co-occurrence of these two conditions (diabetes and eating disorders).
The use of the term diabulimia (which is not a clinical diagnosis) to describe this comorbidity, however, can give the wrongful impression that this comorbidity is somehow unique to eating disorders. Whilst in some ways it is (for instance because of insulin misuse to control weight as a purging/dieting mechanism in this group), in others, this comorbidity mirrors patterns observed for other psychiatric conditions for which autoimmunity is considered as a risk factor.
The existence of high rates of mental health problems in people with autoimmune and autoinflammatory conditions, has been repeatedly observed in relation to several psychiatric diagnoses, including depression (Benros et al, 2013) and psychosis (Benros et al, 2011). However, with some exceptions (see these articles by Raevuori et al and Mårild et al) the association between autoimmunity and eating disorders has primarily focused on type 1 diabetes.
However, an article recently published in Pediatrics by Zerwas and colleagues, has systematically investigated the association between a wide range of autoimmune and autoinflammatory conditions and eating disorders to better explore the nature of these associations.
The research is a large, nationwide, population-based cohort study of Danish citizens born between 1989 and 2006. In Denmark (as in other Scandinavian countries, such as Sweden and Finland) citizens are assigned a personal identification number that allows linking data on their socio-economic circumstances and health outcomes across a number of registers. This allows researchers to conduct analyses on large samples of individuals, which would not be achieved in other cohort studies.
The researchers identified all children diagnosed with an autoimmune or autoinflammatory condition in clinical registers and followed them up over time (until 2012) to see whether they were more likely than the rest of the population to develop an eating disorder. Because they hypothesised that eating disorders in people with autoimmune or autoinflammatory conditions could be driven by gastrointestinal factors, the latter were considered as a separate category of autoimmune disorders.
Eating disorders were defined using International Classification of Disease codes as:
- Anorexia Nervosa (AN)
- Bulimia Nervosa (BN)
- Eating Disorder Not Otherwise Specified (EDNOS).
The study had two main research questions:
- Aim 1: Are people with autoimmune/autoinflammatory conditions more likely to develop eating disorders? [This question aims to explore whether having a diagnosis of an autoimmune/autoinflammatory condition increases a person’s risk of having an eating disorder.]
- Aim 2: Are people with a first degree relative with an autoimmune or autoinflammatory condition more likely to develop an eating disorder? [This type of analysis tries to answer the question of whether shared genetic or environmental factors could explain any associations between autoimmune/autoinflammatory conditions and eating disorders.]
- Children and adolescents who were diagnosed with an autoimmune/autoinflammatory condition had a greater risk of developing any eating disorders
- Having, specifically, a gastrointestinal or auto-inflammatory condition was associated with a greater risk of developing any eating disorder and EDNOS specifically
- The evidence on whether gastrointestinal or autoinflammatory conditions are risk factors for AN and BN was weaker (i.e., these groups still had higher risk, but there was little evidence that this increased risk represented a real difference compared to people without gastrointestinal or auto-inflammatory disorders). This, however, could have possibly been due to fewer people reporting these diagnoses in this group, which could have made it more difficult to detect a real association, as the effect size was comparable across diagnoses.
- Children with a parent or a full-sibling with an autoimmune/autoinflammatory condition had a greater risk of having an eating disorder (irrespective of the diagnosis)
- The authors also found some evidence that children with a cousin who had an autoimmune/autoinflammatory condition were more likely to be diagnosed with an eating disorder.
People with an autoimmune condition seem to have a higher risk of developing an eating disorder, with (overall) little evidence that this risk is different across diagnoses. This association could be partly explained by shared genetic risk between autoimmune and eating disorders, although pathways through which this shared risk is expressed remain to be investigated.
- This was a nationwide study, which included thousands of individuals with a diagnosis of an autoimmune/autoinflammatory condition
- The authors accounted for a wide range of autoimmune (n=50) and autoinflammatory disorders (n=10) and, among these, considered gastrointestinal conditions separately
- They also investigated increased risk of eating disorders in first-degree relatives in order to explore the potential of shared genetic and environmental mechanisms.
- Despite the study having a nationwide scope, the small sample size of individuals with both autoimmune/autoinflammatory conditions and eating disorders might have reduced statistical power and resulted in type II errors (i.e., when an association exists in reality, but is not observed in the analytical sample). This highlights the need for large epidemiological studies to study these rare conditions that might be obtained, for instance, by pooling together different datasets.
- As the authors mention, one potential limitation of using register data is the complete reliance on diagnosed cases. This is particularly problematic for eating disorders as it is known that they are largely under-diagnosed in the population, particularly BN and binge eating disorder (which in these analyses would be captured by the EDNOS category). People who are in contact with the health care system for a condition (e.g. diabetes, crohn’s disease), might also be more likely to receive other diagnoses which are related to that condition (e.g. eating disorders), simply because they have more frequent contact with the healthcare system. This, in epidemiology, is referred to as Berkson’s bias.
- The authors mention that, as transition across eating disorder diagnoses is common, individuals could appear in different eating disorder categories in their study. This is a difficult point for researchers to tackle, especially when numbers are low. However, one approach for future studies could be to employ hierarchical definitions of eating disorders, whereby those who transition to a BN or EDNOS diagnosis from AN (or vice versa) are considered as having BN, and those with AN who never transition to BN are considered AN-only. This could help in distinguishing restricting only from binge-purge phenotypes.
- Finally, it would have been interesting to know what proportion of patients with autoimmune disorders included in the analyses had type 1 diabetes, to get a sense of how much of the observed association could be driven by this condition.
This is a welcome and important study, which suggests that eating disorders (similarly to other psychiatric disorders) might have an inflammatory origin. This has important implications for research, clinicians, and public health:
- More research needs to be conducted to replicate these findings and to better understand what mechanisms at the population and genetic/molecular levels might drive this association
- These findings suggest that clinicians should pay particular attention to potential mental health problems in people with autoimmune disorders as this has become an association observed increasingly frequently. In particular, studies could test the effectiveness and cost-effectiveness of screening children with type 1 diabetes for eating disorder behaviours, as a preventative intervention
- Finally, studies showing that eating disorders might have similar underlying biological risk factors to other psychiatric disorders will (hopefully, eventually) lower the stigma that exists towards them. A previous review about the stigma of eating disorders suggests that anti-stigma campaigns highlighting biological explanations for eating disorders are more likely to be successful. Hence, investigating new biological mechanisms could bring the double benefit of finding new markers of risk (some of which could be modifiable) and increasing help-seeking behaviours.
Zerwas S, Larsen JT, Petersen L, et al. (2017) Eating Disorders, Autoimmune, and Autoinflammatory Disease. Pediatrics. November 2017:e20162089. doi:10.1542/peds.2016-2089.
Raevuori A, Haukka J, Vaarala O, et al. (2014) The Increased Risk for Autoimmune Diseases in Patients with Eating Disorders. Horwitz MS, ed. PLoS One. 2014;9(8):e104845. doi:10.1371/journal.pone.0104845.
Mårild K, Størdal K, Bulik CM, et al. (2017) Celiac Disease and Anorexia Nervosa: A Nationwide Study. Pediatrics. 2017.
Benros ME, Waltoft BL, Nordentoft M, et al. (2013) Autoimmune Diseases and Severe Infections as Risk Factors for Mood Disorders. JAMA Psychiatry. 2013;70(8):812. doi:10.1001/jamapsychiatry.2013.1111.
Benros ME, Nielsen PR, Nordentoft M, Eaton WW, Dalton SO, Mortensen PB. (2011) Autoimmune Diseases and Severe Infections as Risk Factors for Schizophrenia: A 30-Year Population-Based Register Study. Am J Psychiatry. 2011;168(12):1303-1310. doi:10.1176/appi.ajp.2011.11030516.