One of the most prominent environmental risk factors described for numerous diseases is chronic exposure to stressful situations. – “Chronic stress and individual vulnerability,” Schmidt MV, Sterlemann V, Müller MB, Annals of the New York Academy of Sciences, December 2008, 1148, 174-183.

Is it really true that autoimmune diseases become more active in response to stress? There are a handful of related but distinct faulty causal inferences about stress and illness in much of the literature on autoimmune disorders. From work on stressful life events and multiple sclerosis relapse (footnote 1), to reports of stress-related onset and exacerbations of Graves disease (footnote 2), negative weighting of major life events and lupus symptomatology (footnote 3), and psychological factors in sarcoidosis (footnote 4) , medical researchers conflate correlation with causation. The possible logical errors underlying this pervasive tendency are not mutually exclusive, and can be categorized as: (1) reverse causality, (2) factual and thus deterministic accounts of patient history, and (3) endogeneity bias.

First, it is well-established – if it ever needed positivist-style data compilation and analysis to be accepted as fact – that being sick is stressful (footnote 5). Getting a chronic illness diagnosed and treated takes time and money, and is not something most people would choose to have done for fun on the weekends – weekend after weekend, year after year. At the same time, it is also well-established that being stressed out can make you sick in the sense of making you more vulnerable to colds, flu, and running your car into the mailbox (footnote 6). Type A personalities are notoriously more likely to suffer heart attacks and strokes (footnote 7). Clearly the causal arrow can flow both ways when it comes to stress and illness. This means that, while reverse causality may be a significant problem in the literature on stress and autoimmune diseases, the error is actually one of indeterminacy. We just don’t know which way the arrow flows in general for autoimmune disorders, or how much it flows each way, or which comes first – the freaked-out chicken, or its seriously scrambled egg.

An easier error to call out definitively is that of deterministic patient history bias. “Factual framings produce searches for deterministic ‘what made it happen’ accounts of the past, whereas counterfactual framings produce searches for antideterministic accounts that keep pushing back the last possible moment when something else could have happened” (footnote 8). People crave meaning – making meaning is fundamentally what human beings do, be it through art, science, law, religion, or cookies – and we especially crave stories that make everything make sense. As a result of this drive to tell a coherent life story, nobody tells his or her own patient history with counterfactuals. (Okay, nobody who hasn’t got a book deal.) When we are down, perhaps we are more likely to attribute our downness to previous bouts of downness, so that the trend is coherent and logical. Rather than allowing for random error – admitting that we exist by an accident of fate, a lucky roll of the universal dice – perhaps we weave meaning by telling stories in which one life tragedy (major stressors such as crime, divorce, or relocation) causes another (illness). This is the Grand Unifying Theory of Self. It’s simplistic and linear. It’s comforting and comprehensible in a way that random error spelling life or death – to most people – is not. Alternately, maybe doctors themselves have a tendency to project doomed, deterministic histories onto the patients with diagnoses that are particularly difficult to identify and treat – to wit, folks with relatively rare and apparently multi-system, chronic illnesses such as MS, lupus, Graves, sarcoidosis, and the rest of the autoimmune gang. Maybe doctors as well as patients can fall prey to the very human need to make sense of the senseless, to order random error, and ultimately to find someone to blame, just to feel better.

A more obvious error still in the “stress causes autoimmunity” spiel is endogeneity bias (footnote 9). Among people with a family history of autoimmune illnesses, as well as among females, ethnic minorities, and poor people, there is generally more autoimmune illness. There is also more stress in these subgroups, because caring for sick family members is a form of unpaid labor that no society known to man compensates for (footnote 10). It is also notoriously stressful to not have a penis, to not be white, and to not have oodles of cash. Lo and behold, healthy people tend to hit the jackpot and people who hit the jackpot tend to be healthy (and then, if covariance is a valid way of drawing causal inferences, they all go out and buy a penis). I guess the world is a meritocracy after all. Quick, somebody tell all the toddlers in sub-Saharan African dying of diarrheal disease.

In conclusion, it’s certainly true that particular aspects of certain autoimmune diseases are associated with mental health problems like anxiety and depression. Iron-deficiency anemia in lupus, poor sleep quality in Graves disease, and impaired breathing in sarcoidosis are only a few examples of this. But in these instances, poor mental health is a direct effect of poor physical health. Subsequent spiraling via feedback loops tells us nothing about the etiology of autoimmunity as a potentially life-threatening disease process. Rather, it distracts medical care practitioners and social support systems alike when it is misinterpreted to suggest that illness is a choice.

Further research might seek to answer the medical anthropological questions of how this victim-blaming set of logical errors has come to permeate the rheumatology literature. Is it a result of patients with autoimmune diseases narrating histories in which their previous tragedies and/or recent changes in mental state caused or correlated with their illness patterns – and well-meaning, empathetic doctors simply listening well and believing them (footnote 11)? Or are there more nefarious forces at work here – are doctors and medical researchers following the herd mentality, exhibiting societal biases against women (who are far more likely than men to suffer autoimmunity), against minorities (who are more likely than Caucasians to have most autoimmune problems (footnote 12)), and against poor people (who are more likely to suffer from chronic health problems in general (footnote 13))? One thing is for sure: When it comes to the supposed causal relationship between psychological state and autoimmune flares, the only solid proof is all in their heads.


1. “The impact of stressful life events on risk of relapse in women with multiple sclerosis: a prospective study,” Mitsonis CI, Zervas IM, Mitropoulos PA, Dimopoulos NP, Soldatos CR, Potagas CM, and Sfegos CA, European Psychiatry, Oct. 2008, 23 (7) 497-504.

2. “Psychosomatic concept of hyperthyroidism – Graves type – behavioral and biochemical characteristics,” Draganiæ-Gajiæ S, Leciæ-Tosevski D, Svrakiæ D, Paunovic VR, Cvejiæ V, and Cloninger R, Med Pregl., Jul-Aug 2008, 61 (7-8) 383-388; “Age and stress as determinants of the severity of hyperthyroidism caused by Graves’ disease in newly diagnosed patients,” Vos X, Smit N, Endert E, Brosschot J, Tijssen J, Wiersinga W, European Journal of Endocrinology, Oct. 30, 2008 (Epub ahead of print, accessed via PubMed); “A patient with stress-related onset and exacerbations of Graves disease,” Vita R, Lapa D, Vita G, Trimarchi F, Benvenga S, Nat. Clin. Pract. Endocrinol. Metab., Jan. 2009, 5 (1) 55-61.

3. “The role of stress in functional disability among women with systemic lupus erythematosus: a prospective study,” Da Costa D, Dobkin PL, Pinard L, Fortin PR, Danoff DS, Esdaile JM, and Clarke AE, Arthritis Care & Research, June 2001 12 (2) 112-119; “Stress, depression, and anxiety predict average symptom severity and daily symptom fluctuation in systemic lupus erythematosus,” Adams Jr. SG, Dammers PM, Saia TL, Brantley PJ, and Gaydos GR, Journal of Behavioral Medicine, July 2005, 17 (5) 459-477.

4. “Psychological factors in sarcoidosis: the relationship between life stress and pulmonary function,” Klonoff EA, Leinhenz ME, Sarcoidosis, September 1993, 10 (2) 118-124.

5. See, for, example: “Psychological Effects of Chronic Disease,” C Eiser, Journal of Child Psychology and Psychiatry, Dec. 2006, 31 (1) 85-98; and “Toward a general model of health-related quality of life,” Romney DM and Evans DR, Quality of Life Research, December 2004, 5 (2) 235-241. To be fair, the second article “suggests that, although a medical model of HRQOL [health-related quality of life] may be more important when it comes to alleviating illness, a psychosocial model of HRQOL may be more important when it comes to maintaining health and preventing illness.”

6. By which I mean in no way to suggest that there is anything wrong with people who run into the mailbox, honey.

7. As genetics hurtles forward, however, even this au courant theory – commonly accepted as fact – may soon be disproven by the advancement of alternate explanations. For example, the recently discovered MYBPC3 variant is said to cause latent or active heart disease in tens of millions of Indians. If Indians are also disproportionately represented in human capital-intensive fields that require so-called Type A characteristics (like intelligence and organizational skills), then already the Type A story of heart disease has been thrown into question.

8. “Counterfactual Thought Experiments,” Tetlock PE and Parker G, in Unmaking the West: “What-if?” Scenarios that Rewrite World History, Tetlock PE, Lebow RN, and Parker G, Ed., citing Philip E. Tetlock and Richard Ned Lebow, “Poking Counterfactual Holes in Covering Laws: Cognitive Styles and Historical Reasoning,” American Political Science Review 95 (2001): 829-43.

9. Endogeneity bias is a logical flaw that pops up in a lot of social scientific and scientific research when the thing being studied has multiple characteristics of interest. For example, if you wanted to know whether gun ownership increased individual citizens’ chances of being murdered, and you studied victims of domestic violence who bought a gun because their partners had threatened to kill them, your research would suffer from a serious endogeneity bias. Your research subjects would be more likely to be murdered by their partners or former partners who had already threatened to do so, and so you wouldn’t be able to tell how their gun ownership affected their likelihood of getting killed as compared to the general population’s murder risk. In my current context of interest, endogeneity bias is at work when people who are already likelier to be operating under stressful conditions – say, African-American women caring for disabled family members while struggling to gain equal pay – are also found to be more likely to develop lupus than WASPy types whose healthy families have worked for the firm of Fancy, Schmancy & Hung for decades. Stress and lupus correlate in certain subgroups, and they may well covary; but those facts establish no causal relationship between the two variables.

10. And by man, I mean one ignorant American writer. If you are a country, and you will pay me to stay home researching my friends’ and family’s illnesses, call me.

11. I’m playing devil’s advocate here. For a few readers, I played it too well – I am emphatically not saying that sick people tend to blame their life histories for their illnesses. Personally, I think it’s obvious that medical researchers who engage in the blame-the-victim error of suggesting that maladaptive psychology causes autoimmunity are defending themselves from their own subconscious guilt at being healthy when others, by the luck of the draw, are not. Them’s sore winners.

12. But oh, what a tangled web we weave! Non-whites have higher rates of autoimmune diseases like lupus, but being racially discriminated against is in turn associated with having health problems. “It’s enough to make you sick: the impact of racism on the health of Aboriginal Australians,” Larson A, Gillies M, Howard PJ, Coffin J, Australian and New Zealand Journal of Public Health, August 2007, 31(4):322-9.

13. Poverty also correlates with increased exposures to environmental contaminants, decreased access to clean water and to safe and nutritious food, less preventive medical care, and other phenomena that translate into chronic disease. Since women and non-whites are also disproportionately likely to experience poverty, covariance is a problem six ways from Sunday.