Harmless Bubble Clouds

Flaws in Pushing the Idea of Harmless Obesity

So much energy goes into amping up the “crisis” of obesity that an equal and opposite reaction is natural. Surely, can’t a person be fat and fit? Aren’t there certain situation where a bit of extra adiposity can actually help a person’s health? But a series of recent papers challenges two concepts of harmless obesity – the so-called obesity paradox and metabolically healthy obesity.

Obesity Paradox

Writing in Obesity, Katherine Flegal and John Ioannidis say that the “obesity paradox” is misleading terminology that we should stop using. It’s a figure of speech with no precise meaning, other than to stand in the way of learning from counterintuitive research findings.

Even with a very specific definition of an obesity paradox – an observed risk reduction associated with obesity – we have an inherent problem with collider bias. Hailey Banack and Andrew Stokes explained this problem in an IJO editorial. Because of collider bias, nothing short of a prospective, controlled study can establish the effect of obesity on the risk of bad outcomes from diseases associated with obesity.

Building on that argument, Anna Peeters says that scientific journals should no longer accept obesity paradox articles. Serious scientists should instead pursue studies that are adequately designed to address the question.

Metabolically Healthy Obesity

Two recent papers in IJO address the concept of metabolically healthy obesity. Ana Espinosa De Ycaza and colleagues show that people who have obesity but no metabolic complications are likely to develop those complications later. Further weight gain amplifies that risk.

Patrick Bradshaw and colleagues studied 3,969 people with normal weight, excess weight, or obesity who were otherwise healthy. After nine years, the people with overweight and obesity were more likely to develop cardiometabolic risk factors. It was rising blood glucose that developed most quickly.

An Excuse for Neglect

Richard Atkinson and Ian Macdonald pull these observations together beautifully in an IJO editorial. They say that these flawed concepts of harmless obesity tend to support a bias for neglecting treatment. “It is shameful that obesity treatment is either not covered or is very poorly covered by the third party payers of many countries,” they conclude.

Every patient deserves access to real medical care for obesity. No excuse for neglect or denial of care is acceptable.

Harmless Bubble Clouds, photograph © espie (on and off) / flickr

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.


April 6, 2018

9 Responses to “Flaws in Pushing the Idea of Harmless Obesity”

  1. April 06, 2018 at 6:45 am, Angela Meadows said:

    Again, just to be clear, what exactly are the long-term effects on metabolic health of living in an environment where stigma toward your group is endemic in practically every domain of life?

    From my reading of the prejudice literature in general (racism, homophobia, Islamophobia etc), it would seem that this is associated with higher rates of hypertension, heart disease, diabetes, and premature mortality. I don’t suppose any of these obesity studies accounted for that?

    • April 06, 2018 at 7:37 am, Ted said:

      Thanks for this reminder, Angela. No doubt, bias, stigma, and discrimination compound the harm of obesity. In some cases, these social harms are far worse than the physiologic harm of the metabolic disease. In other cases, it just adds insult to injury and interferes with finding good medical care. Regardless, we will never address the harm of obesity without first rejecting bias, stigma, and discrimination.

  2. April 06, 2018 at 8:13 am, Allen Browne said:

    1) children with obesity are not healthy either
    2) bias, stigma, and discrimination is a harm of obesity currently and cause real damage to lives.


  3. April 06, 2018 at 10:24 am, Chester Draws said:

    Obesity should be defined as the point where your weight affects your health, not vice versa. The “paradox” is not solved by refusing to talk about it. That’s just denial.

    Two issues: we don’t measure obesity, we use proxies. And some of those proxies are ludicrous, like MBI.

    Too many studies don’t distinguish sufficiently between levels of obesity. Studies that show increased risk for “obesity” arise if the morbidly obese are at risk and the slightly obese are not. Those studies do not tell us where the danger begins.

    The evidence I have seen is that “obesity” that is actually just those with MBIs just into the, entirely arbitrary, range of “obese” is not hugely harmful.

  4. April 12, 2018 at 9:06 pm, Joanne Ikeda, RD said:

    So many “experts” comment as if we had an effective treatment for obesity. If you look at the scientific literature, you will find that we can help people lose weight – TEMPORARILY! However, over 90% or more regain the weight they have lost within 3 to 5 years! In the infamous “Look Ahead” study which started out working with over 5,000 people at multiple academic intitutions across the country for over 9 years, was a complete failure. There was an 84% drop out rate. After suffering 9 yers of treatment, the average weight loss for women was 5 pounds (from 204 to 199 pounds) and the average weight loss for men was l.3 pounds (240 to 238. Why keep harping on fat people to lose weight when we know it isn’t possible?!
    UC Berkeley Retired Nutrition Faculty Member

    • April 13, 2018 at 3:51 am, Ted said:

      Thanks, Joanne, for offering your point of view.

      I agree with a number of things you’ve suggested here. Respect for people of all sizes and their right to make their own decisions comes first. I’m sick of people offering up ignorant opinions and judgements about the health of other people. I don’t like harping, either.

      I also agree that the options for obesity care are less than ideal. But we do have options that can empower people to improve their health. And people who want those options should have access and be free from harsh judgements and second guessing about the decisions they make for themselves.

      Again, thanks for sharing your view.

  5. April 13, 2018 at 2:19 pm, Katherine Flegal said:

    Our commentary was of necessity brief. But we were trying to point out there are many complexities regarding the issue of whether or not obesity or overweight are associated with better outcomes in specific cases. One example, the original “obesity paradox,” had to do with response to a specific medical treatment. Other examples have to with conditions or diseases that are not known to be associated with obesity, in which case the collider bias explanation may not apply. The collider bias explanation is essentially the same argument as put forth by Stamler in 1991, that lean “hypertensives” have hypertension because of stronger risk factors than hypertension related to obesity and thus lean people with hypertension fare worse. If this is the case in a specific situation, be it a disease or a particular treatment, it behooves researchers to investigate what those risk factors among the normal weight might be. Simply attributing such findings to collider bias does not get us very far.

    • April 13, 2018 at 2:39 pm, Ted said:

      Thanks for adding your wise perspective, Katherine.

  6. April 13, 2018 at 3:27 pm, Paul Ernsberger said:

    The hidden story behind the term “obesity paradox” is that allowed previously unpublished data to be published. Until this term was popularized, it was impossible to publish scientific papers showing a beneficial effect of obesity, no matter how limited or small the effect was. Trivial or dubious harmful associations with obesity have always been welcomed by journals.

    There was a treasure trove of data showing positive implications of obesity languishing in file cabinets of scientists across the globe. Introduction of the concept of “obesity paradox” allowed a flood of studies to see the light of day.
    I predict the flood of studies will slow to a trickle once the backlog of suppressed research data is depleted.