An Exploration of Weight Bias in the Primary Source Literature

by Health At Every Size® Blog

by Annie Goldsmith, RD, LDN

Recently, a research article with an interesting title landed in my inbox: “Metabolically Normal Obese People are Protected from Adverse Effects Following Weight Gain¹”. As a Dietitian who works from a Health At Every Size® perspective, it certainly caught my eye! At first scan, the article appeared to support what I already believe to be true: weight as an isolated variable is not a predictor of heath risk or outcomes. However, as I read closer, my heart began to sink. I found in this article what I so often find in the world around me: a blatant, unfounded weight bias.

The Study: The Bare Facts

Before I launch into commentary, I’ll present here the objective facts of the study design and the authors’ conclusions. The sample consisted of 20 participants identified to be obese and divided into two groups: Metabolically Normal Obese (MNO; n=12) and Metabolically Abnormal Obese (MAO; n=8). The marker used to classify the participants as MAO was increased intrahepatic triglyceride (IHTG), which is associated with adverse metabolic effects such as insulin resistance and cardiovascular disease risk factors. The question the authors investigated was whether the MNO group was somehow protected from adverse metabolic outcomes as compared with the MAO group when both groups gained a similar amount of weight. They also were interested in the possible mechanism driving any differences between the groups.

In order to test their hypothesis, the authors instructed each group to consume 1000 extra kcals per day from one of 5 restaurants: McDonald’s, Burger King, KFC, Pizza Hut and Taco Bell. The subjects continued on with this diet until they had gained ~5-7% of their original body weight and maintained it for 2 weeks. The authors then re-tested IHTG and associated metabolic risk factors. Results showed that the MNO participants displayed no adverse metabolic effects from the short-term calorie surplus and weight gain, while the MAO did suffer increased risk factors. The authors concluded, “MNO people with normal IHTG content represent a distinct phenotype and are not simply in a transition phase toward MAO¹” In other words, the results suggest that obesity alone is not an indicator that can be used to identify heath risk!

The Study: The Underlying Message

At first pass, this study might seem encouraging – one more result to add to the mounting body of work that demonstrates the obesity researchers need to ditch this weight bias thing once and for all. However, what the bare facts don’t reveal, the specific language the authors use surely does. They write in the introduction that a primary purpose of the article is,

“To evaluate some of the putative molecular mechanisms in adipose tissue responsible for the adverse metabolic effects of weight gain.¹” (italics mine).

This statement is made before collecting one ounce of data! The authors simply want to see if the results will help them explain the weight-related mechanism for poor health – whether or not it exists at all is never questioned.
The assumption that weight is the causative factor for poor health drives this article forward, and it is particularly evident in the discussion (where the authors have freedom to speculate!). They spend an entire paragraph hypothesizing about,

“…the cellular mechanisms responsible for the differences in metabolic function between MNO and MAO people.¹”

(spoiler alert: adipose tissue is implicated). However, they ultimately concede that the mechanisms are “not clear”. Then there’s the kicker: after being forced to conclude that MNO people are in fact phenotypically different from MAO people (in other words: healthy), the authors throw in for good measure,

“Nonetheless, we cannot exclude the possibility that our MNO subjects would develop metabolic abnormalities with greater weight gain.¹”

Wait – what?? The data reveal a perfectly clear outcome, and the authors undermine the entire result with one throw-away statement. Which, by the way, is quite useless and absurd. What do they think would happen to any of us at ANY BMI if we overate fast food into the unforeseeable future?  (Morgan Spurlock gave us a pretty good idea in his movie “Super Size Me” – and he started out in perfect health!)

A Missed Opportunity

Sadly, a study that started out with potential ended as a missed opportunity. What might the authors have found if they had controlled for the myriad of confounding variables left glaringly unaddressed in the research design? Not one mention is made of how lifestyle factors such as eating habits, physical activity, sleep hygiene, and stress management might have contributed to the differences between groups. Or how about genetics? Family history is not mentioned either. I will go out on a limb and make some speculations of my own: it’s very likely that had the authors been able to see past their blind certainty that weight alone is the mediator of disease risk, they might actually have tapped into some of the possible underlying factors differentiating the MNO and MAO groups. Additionally, conspicuously absent are control groups of metabolically normal and metabolically abnormal normal-weight participants. Thin people who are sick exist in this world too! What impact would a steady diet of Taco Bell have had on them? We can only speculate.

In their wonderful article “Weight Science: Evaluating the Evidence for a Paradigm Shift²” Linda Bacon and Lucy Aphramor state,

“Researchers have demonstrated ways in which bias and convention interfere with robust scientific reasoning such that obesity research seems to ‘enjoy special immunity from accepted standards in clinical practice and publishing ethics.’ “

Here we see an example of such research. Until there is a shift in how society as a whole – or at least the scientific community – views the relationship between weight and health, NIH funding will continue to be wasted on superfluous studies that tell us what we in the HAES® community already know: weight is not a proxy for health.

References

1. Fabbrini, E et. al. Metabolically Normal Obese People are Protected from Adverse
Effects Following Weight Gain. J Clin Invest. doi:10.1172/JCI78425: 2015.

2. Bacon, Linda and Aphramor, Lucy. Weight Science: Evaluating the Evidence for a
Paradigm Shift. Nutrition Journal 2011, 10:9

Annie Goldsmith

Annie Goldsmith is a Registered Dietitian practicing one-on-one nutrition counseling in Charlotte, NC. She has an undergraduate degree in Brain and Cognitive Sciences, a field she was drawn to because she possesses an infinite curiosity about that place where biology and psychology come together to guide behavior. Her transition to a career in nutrition was a natural one, as she believes that how we choose to feed ourselves is deeply rooted in human physiology and psychology. She sees health as multidimensional, and has embraced the HAES® philosophy in her own practice. When not working, Annie spends time hanging out with her wonderful husband Brian and their intrepid cat, Chicken.

9 Comments to “An Exploration of Weight Bias in the Primary Source Literature”

  1. The bias is almost certainly not that of the authors. The “gotcha” statement is typically required by the anonymous reviewers or by the editor. In science, as in journalism, one is supposed to acknowledge the possibility that the other point of view is correct.
    Another kind of bias is that authors working in the old paradigm never have to acknowledge that HAES may be correct.

    • Point well taken that offering up an alternative explanation for the outcome is a common practice in primary source literature. I will contend, however, that the bias does not lie solely in that one “gotcha” statement. Rather, it is imbedded into the fundamental study design as well as language scattered throughout the entire article. So assuming the authors are the ones who designed the experiment and submitted the grant, I would still hold them accountable.
      And yes, excellent point about the double standard that seems to exist!

  2. Thank you, Annie, for a well-written post. You are spot on about this:

    “…conspicuously absent are control groups of metabolically normal and metabolically abnormal normal-weight participants. Thin people who are sick exist in this world too!”

    As one individual, “rg,” observed in last week’s comments:

    “Remember, if you are fat, you can never die of natural causes. You will only be able to die of conditions ’caused or exacerbated by obesity’ — even if you live to 110.”

    It is wearying to be enlightened by the HAES® paradigm while so much of the world still believes as axiomatic that fat = bad. I thank each of you who fights this bias every day and remind you that it is not in vain: little by little, the truth must surface.

    • Hi Sandra,

      Thanks for your feedback! I know what you mean by it feeling wearying walking around in a world where weight bias is pretty much ubiquitous. It’s also incredibly frustrating! I guess despite that we keep fighting it because…what is the alternative?

  3. Thanks for this Annie, really enlightening.

    Amanda
    Amanda Hallson (MSc., RD)
    Team Lead Mansionhouse Unit
    Greater Glasgow & Clyde Weight Management Service
    100 Mansionhouse Rd.,
    Langside, Glasgow
    G41 3DX
    Tel: (0141) 201 6115
    Email: amanda.hallson@ggc.scot.nhs.uk
    [cid:image001.jpg@01D036F0.9103CA20]

  4. So much bias and another bias and massive assumption is that fat people just continue to get fatter. Huge assumption there folks. What we in HAES aim for is stable weight which is more likely especially after exiting the weight cycling treadmill and embracing a health focused paradigm, and that is not even considered. Grrr

  5. While I agree with the entire underlying premise of this piece, I have to also agree with Paul that most (he just mentioned one, but I’ll go with most) of your objections to this particular paper are related to practical and pragmatic issues rather than bias.

    I don’t doubt for a second that anti-fat bias exists in academia, and may even be held by these researchers, but there are limitations to any given research study – in terms of numbers recruited and scope. The fact that they were limited to 20 people tells me more about their funding than it does about their bias, and I think they went with the best choice to answer that particular research question given these constraints. And I say this as someone who frequently rails at the conspicuously absent ‘unhealthy normal weight’ group in most epidemiological studies.

    The spin you put into your intro will determine where (and if) the piece can be published. The gotcha clause in the discussion is no more than good practice and would also have been required by reviewers. The absence of testing for confounds is more of an issue, because as you say, it could theoretically help explain the differences observed. However, in a sample size of 20, this is extremely unlikely – it wouldn’t be powered to pick up those kinds of differences.

    In the end, I am thoroughly delighted that they managed to get this small study into the Journal of Clinical Investigation (impact factor 13.8, one of highest ranked in Medicine, Research & Experimental), and showed that MHO who gained weight were still MHO afterwards. I am more amazed that they got permission from their ethical review board to conduct this study in the first place!!

  6. While I agree with the entire underlying premise of this piece, I have to also agree with Paul that most (he just mentioned one, but I’ll go with most) of your objections to this particular paper are related to practical and pragmatic issues rather than bias.
    I don’t doubt for a second that anti-fat bias exists in academia, and may even be held by these researchers, but there are limitations to any given research study – in terms of numbers recruited and scope. The fact that they were limited to 20 people tells me more about their funding than it does about their bias, and I think they went with the best choice to answer that particular research question given these constraints. And I say this as someone who frequently rails at the conspicuously absent ‘unhealthy normal weight’ group in most epidemiological studies.
    The spin you put into your intro will determine where (and if) the piece can be published. The gotcha clause in the discussion is no more than good practice and would also have been required by reviewers. The absence of testing for confounds is more of an issue, because as you say, it could theoretically help explain the differences observed. However, in a sample size of 20, this is extremely unlikely – it wouldn’t be powered to pick up those kinds of differences.
    In the end, I am thoroughly delighted that they managed to get this small study into the Journal of Clinical Investigation (impact factor 13.8, one of highest ranked in Medicine, Research & Experimental), and showed that MHO who gained weight were still MHO afterwards. I am more amazed that they got permission from their ethical review board to conduct this study in the first place!!

    • Really interesting feedback. If I am interpreting your comment correctly, you are making the point that the issues that I raised could also be explained by politics in academia (the spin that helps you get published) and logistics (numbers recruited and scope), not necessarily simply the bias of the authors. Important points, and thanks for bringing them up!

      My reaction to that is, it highlights the issues of weight bias in scientific research got an even greater extent. Not so much the logistics part (and clearly this was a small, limited study), but more the idea that you need to “spin” an article in a certain way to be more competitive for publishing. If presenting this study differently, from a more neutral perspective, lessened the chances of it getting published (or lessened the quality of the journal it was accepted into) – then that seems to be a problem. Why should the “pragmatic” approach be to lead with the assumption that fat = unhealthy? Even if testing for the confounding variables was not feasible, surely a mention of them as another possible alternate explanation for differences between groups would have presented a more balanced picture.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: