“Our Results Are Unexpected, So The Participants Must Have Been in Denial”: Exploring a Worrying Trend in Eating Disorders Research

Something I have come across several times when reading ED research studies is a disclaimer that research has been dutifully carried out, but the findings have to be viewed with some scepticism because the participants (and – more specifically – the participants with AN-R) were in denial when completing the self-report questionnaires.

In this post, I will to look at a couple of recent studies that flag-up this issue, to examine what is behind this disclaimer.

The first paper I will explore was published very recently by Gailledrat et al. and touches on body shape concerns in women with eating disorders. Participants were women with a diagnosis of AN or BN selected from a clinic in France. The researchers inform us that, “patients with an ED are much more concerned with their body image and weight than the rest of the population”. The authors do not provide any support for this statement by referencing published research or population norms data.

With this in mind, they selected level of body shape concern to be the key variable in their research: this would be correlated against several other variables. To determine the level of body shape concern present in their participants, they required them to complete the Body Shape Questionnaire (BSQ.)

However, their results were not what they expected. From the discussion section of the article:

“Firstly, we showed that a significant proportion of patients seeking treatment for ED had no or only mild concerns with body shape. This could be of surprise as body shape concerns are core symptoms of ED. However, this situation is not unusual and several explanations can be considered. Patients with low or normal BMI could have self-reported that they were satisfied with their current body shape whereas the interviewer could have detected that the self-evaluation was unduly influenced by body shape and weight. This raises the question of the truthfulness of patients’ responses as a whole (body shape concerns, laxative use, etc.) when using self-questionnaires.”

They later add:

“The results must be viewed in the context of several limitations. As already partly mentioned, it is very difficult to measure a state of denial, which probably represents the major bias of this type of study and more specifically in patients suffering from AN-R.”

This stance is also taken in the second paper I will examine, also published this year by Pilecki et al.  about eating disorders in Poland. They aimed to explore eating disorders in Poland after 1999, in the context of Westernization (note: Andrea wrote a post about this study here).

Participants in this study were asked to complete a “Questionnaire of Socio-Cultural Context”, which included several questions on weight/shape evaluation. Unlike the French study, it reports mean response scores for separate ED diagnoses. And what they found was that scores for “thin ideal” items were considerably lower for patients with AN-R than for the healthy control group. However, rather than accepting this conclusion, the authors write:

“Would the results suggest that cultural context is relevant for BN but not for ANR? For Hilde Bruch [59], a distorted body image is central to the symptomology of ED, and the more body image is distorted, the more pronounced are the symptoms. What may be significant here are the relationships identified by items 12, 16 and 18, which indicate the irrelevance of thin pressure in the ANR group. […] Symptom denial is an important aspect of ED, and may affect not only the patient’s somatic state, but also other relevant aspects of the patient’s self-image and perception of relation to the world. This brings the credibility of any self-assessment test performed by girls suffering from ANR into question, including self-assessment tests of cultural issues.”

Analysis

So, what responses were participants reporting that the authors coded as denial?

In the Polish study, contrary to expectation, the AN-R girls declined to endorse the following “thin pressure” items on the questionnaire:

12. When I see a thin model, I feel fat and unattractive.
13. To be more attractive, I should be skinnier.
14. I am jealous of a model’s slim appearance

The researchers saw the non-response here as anomalous responses and evidence of denial. And they also add: “It is possible that thin pressure was relevant before the manifestation of ED, but then lost its relevance following significant weight loss”.

I personally find it very troubling that they are so in thrall to this crude stereotype. It has been shown, for example in this research that perceptions of AN differ significantly between people who have the condition and people who do not. People with AN tend to emphasize the emotional factors and precursors to the illness, whereas lay people (including some researchers) are more likely to see it as a “slimming disease” and to overestimate the influence of cultural and social factors. This discrepancy can explain the “unexpected” responses in this study.

As we saw from their discussion, the French researchers decided that the high number of patients expressing low or no body shape concern was evidence of denial. They expand on the theory that this is typical in AN-R patients, adding:  “… as a general rule, patients with a lower body weight exhibit a more acute state of denial and alexithymia and are therefore less in touch with their body shape concerns”.

However, their data actually highlights 34 participants (27.6%) who “expressed body dissatisfaction due to a feeling of being underweight”. This statistic doesn’t feature at all in their discussion and the authors don’t appear to have taken it into account before stating that respondents were denying their concerns about becoming overweight.

Perhaps the biggest issue in the study is that the tool they used (The Body Shape Questionnaire) is not actually able to measure what they appear to mean by the term “body shape concern”. They discuss this concept at some length, for example by noting that: “ED are usually associated with disturbances of self-perception and body size estimation”.

And this may be what they hoped to measure, and expected their ED respondents to endorse. However, the BSQ is merely a list of 34 items about physical things like “rolls of fat”. It asks nothing about disturbances in body-size perception or other issues specific to EDs. It should also be noted that the participant sample (across both AN and BN diagnoses) had an average BMI of 17.6, and many of the questions do not necessarily apply to the large number of them who were clearly very underweight (and especially those who readily acknowledged their underweight status – via the Contour Drawing Rating Scale.) For example:

• Have you felt that it is not fair that other women are thinner than you?
• Has being with thin women made you feel self-conscious about your shape?
• Have you thought that you are in the shape you are because you lack self-control?

Another question is asking two potentially incompatible things at once, i.e.

• Have you avoided running because your flesh might wobble?

It’s certainly true that respondents might worry about wobbling flesh regardless of their actual body composition, but many people with AN and BN have a compulsion to exercise and would definitely NOT avoid running – irrespective of any worries on that front.

In addition, the only two references to ED-specific symptoms in the BSQ are two items about vomiting and laxative abuse – behaviours which (according to the diagnostic definitions the researchers used) do not feature in AN-R: so respondents who do not assent to them are actually answering truthfully.

In conclusion, given the nature of the participant sample (mostly very underweight), it is entirely to be expected that a sizable number of them did not score highly on this questionnaire. It is not reasonable to claim that this signifies denial or response bias.

Discussion

I’d like to address the issue of symptom denial in AN, because  I would argue that it is being used as a catch-all to explain research that doesn’t obtain the expected results. It’s certainly true that AN patients will sometimes deny behaviours in order to continue food restriction or maintain other ED behaviours. But that’s because it can feel existentially compelling to do so: that’s the nature of the condition. And that is simply not the case when reporting on cultural issues or shape concerns.

Here, there is no need to dissemble as the answer has no consequences. It is a huge leap for a researcher to decide that merely because a participant has AN (a condition sometimes associated with symptom denial) they can just discount everything they say – especially whenever responses don’t fit their hypothesis:

Symptom denial is an important aspect of ED, and may affect not only the patient’s somatic state, but also other relevant aspects of the patient’s self-image and perception of relation to the world. This brings the credibility of any self-assessment test performed by girls suffering from ANR into question, including self-assessment tests of cultural issues.

It’s a shame that the French study revolves around levels of body shape concern because the researchers report some interesting findings in other areas. For example,  they found that 87% of ED patients had a family history of addiction: with just over half of this being a family history of EDs. This is considerably more than the number of respondents who reported marked body shape concerns (44% of the original group who completed the questionnaires; or 57% of those whose data was analysed): yet it is not discussed. In accordance with their hypothesis, only body–image interventions feature as follow-up items for further research or action.

Issues and Implications

One main reason for concern about this trend in ED research is the issue of citations. The authors state that they detected participants with AN-R answering untruthfully because they were in denial, and this “finding” could potentially now be used to substantiate and perpetuate this idea in future research – leading to further silencing of this group.

Secondly, in both these studies there is a clear issue with invalidation: some of the participants have indicated that body shape concerns or “thin ideal” pressures are NOT driving their ED – and researchers have refused to believe them. In both studies references are made to authors such as Hilde Bruch or to the DSM definitions to back up claims that the respondents are in denial. They appear to take it for granted that if a participant does not endorse a prevailing social construct theory it is the person living with an ED who is suspect, not the theory. If they are unwilling to accept their participants’ responses you have to question what is the point of even carrying out the research.

A third issue is the use of off-the-peg questionnaires to measure complex concepts. Clearly, they can enable comparison of data between studies, but it is surely more important to collect meaningful data by actually measuring what you intend to measure.

Conclusion

Given that it seems that these researchers did not take into account:

1. Their pre-existing assumptions about EDs may not be completely accurate.
2. They are using crude ready-made questionnaires that cannot actually quantify what they want to measure – and contain questions that can be understood in several different ways and/or analysed in multiple directions (for example, depending on whether the respondent themselves is under or over- weight and is expressing either under- or over-weight concerns).

I contend that there is no actual evidence that participants in these studies were in denial and that studies such as these should not be used to support the idea that research among certain groups is typified by respondent denial.

References

Gailledrat L, Rousselet M, Venisse JL, Lambert S, Rocher B, Remaud M, Guilleux A, Sauvaget A, Eyzop E, & Grall-Bronnec M (2016). Marked Body Shape Concerns in Female Patients Suffering from Eating Disorders: Relevance of a Clinical Sub-Group. PloS one, 11 (10) PMID: 27776159

Pilecki, M.W., Sałapa, K. &  Józefik, B. (2016). Socio-cultural context of eating disorders in Poland. Journal of Eating Disorders, 4(11) DOI: 10.1186/s40337-016-0093-3 

•