The Obesity Paradox Revisited

As Peter and I discuss frequently here at Obesity Panacea, the relationship between body weight and health is not always as neat and tidy as you might expect (For all the details, check out Peter’s 5-part series on metabolically healthy obesity).  A recent paper published in the International Journal of Obesity by Drs DK Childers and David Allison examines a number of these issues, and suggests ways that they may be at least partially resolved.

In the intro to this new paper, the authors point out 3 confusing issues surrounding the relationship between body weight and health:

  1. The relationship between body mass and mortality in epidemiological studies is “U-shaped” – high health risk for individuals with a body mass that is very high or very low, and low risk for individuals with intermediate body mass (e.g. 18-28 kg/m2 or so, depending on sex and ethnicity).  However, the “ideal BMI” is far from clear (some studies have found that having a BMI well into the “overweight” range may be associated with minimal health risk), as is the causality of the relationship between low BMI and health risk (e.g. does being underweight cause disease, or does having a disease cause you to lose weight?).  Given the wealth of research on obesity, it’s surprising that we really have only a very general idea of the relationship between body weight and health!
  2. Even more confusing, whatever the “ideal” BMI is, it appears to increase with age.  The nadir of the U-shaped curve shifts slightly to the right in older populations.  Why?  No one really knows.
  3. Finally, there is the so-called “obesity paradox” – the idea that although obesity is typically associated with increased risk of death and disease in the general population, among individuals who have certain illnesses or injuries, obesity is actually associated with increased survival time. In other words, obesity is generally bad for your health, but in certain situations it may actually offer a protective advantage.

All of the above issues are quite controversial and not terribly well understood.  However, in their new paper, Drs Childers and Allison suggest that the obesity paradox may actually help to explain the U-shaped nature of the relationship between body weight and disease, as well as the reason why the ideal body weight increases with age.  From the paper:

Specifically, we suggest that the obesity paradox in which obesity hypothetically (a) causally monotonically increases MR among persons in the absence of major injuries or other diseases; yet (b) causally monotonically decreases MR among persons in the presence of certain major injuries or other diseases; can explain both the U-shaped curve and the increasing nadir with age. The existence of major injuries and diseases that satisfy (b) at the association level (not necessarily causal) is well founded. Let the collection of all such injuries or diseases be called afflictions of type A. We take no position here regarding the veracity of the causal aspects of point (b) above, but merely evaluate the consequences and potential explanatory power of it being true.

Now I should point out that this paper is entirely stats based, and what the authors did was simply calculate what the relationship between body weight and health would look like if the obesity paradox is true.  Here’s basically what they assume:

In the absence of disease or injury, BMI has an exponential relationship with health risk – the higher the BMI the higher your estimated health risk.  However, when you have those diseases for which body weight is protective, there is a general decrease in risk as your BMI goes up.  So what happens when you make those assumptions?  Well, you wind up with a relationship between BMI and health risk that very similar to the one we see in the general population – a U-shaped relationship between BMI and health risk at any given age, but with the “ideal” BMI increasing slightly as age goes up.

Now this paper is incredibly speculative, the models exclude a number of potentially important factors (e.g. body fat distribution, physical activity levels, diet, etc) and doesn’t necessarily mean that the obesity paradox is the driving force behind the observed relationship between BMI and health.  But it does seem to suggest that the obesity paradox could be one explanation for this relationship, and one worthy of further study.  Now that’s not terribly straight-forward in humans (for obvious reasons we can’t randomize people to different body weights and then subject them to injury or illness to see what happens!), but the authors suggest that animal models could be used to see whether the findings in the paper hold up under closer scrutiny.  The authors also point out that for some individuals at high risk for specific illnesses, weight gain may actually be protective, although I think they realize that it’s not really a viable option, especially for the time being (talk about personalized medicine!).  I should point out that for the record, there is still very little evidence that you should gain weight to live longer (and even then only in incredibly specific circumstances – e.g. right before you are going to suffer a debilitating injury), despite what the mainstream media may claim from time to time.  As usual, lots of questions, not a whole lot of answers, but it at least sheds some light on the reasons behind the complicated relationship between body weight and health risk.

Thanks to Amby Burfoot for suggesting we look into this interesting topic!


ResearchBlogging.orgChilders, D., & Allison, D. (2010). The ‘obesity paradox’: a parsimonious explanation for relations among obesity, mortality rate and aging? International Journal of Obesity, 34 (8), 1231-1238 DOI: 10.1038/ijo.2010.71

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Obesity Research and tagged , , . Bookmark the permalink.

17 Responses to The Obesity Paradox Revisited

  1. Ed says:

    People like to say that the ideal BMI is far from clear. And yes, some (usually pretty small) prospective studies have found that classically overweight people have the lowest all-cause mortality.

    But there *is* a clear overall picture. This ( is a collaborative analysis of data from 57 separate prospective studies, including a sample of 900,000, and lead-authored by the one of the chaps who invented meta-analysis in the first place. It says that the optimal BMI is 22.5, pretty much smack-dab in the middle of the traditionally healthy range.

    Nice post, Travis.

  2. Obesity is a risk factor for coronary heart disease and heart failure. Yet obese and overweight patients with coronary heart disease and heart failure seem to have better clinical outcomes than others so afflicted except at “healthy” BMIs of 18.5-25.

    So should we tell our obese and overweight heart patients NOT to try losing the excess weight? I was hoping for a clear answer when I read the article at hand, but was disappointed to find no answer. I don’t think anyone really knows.

    • DebraSY says:

      Dr. Parker: I know you’re asking Travis and Peter, but since it is an open blog, I’ll risk jumping in, just as a lowly patient. I would suggest, since an enormous percentage of us will not be able to maintain losses, even if we attain them to begin with, you’re kindest care would be to set other benchmarks than weight, and treat weight loss as a lovely bonus when it happens as a result of other measures we take to improve our health. In this way you establish an atmosphere of support not weight judgement. That way, if we regain the weight, we don’t feel the additional burden of having disappointed you, and we won’t go AWOL and quietly find another doctor. Weight regain happens for a variety of reasons, not because we suddenly go crazy for fast food. Often, people just cannot maintain an energy equation that is radically lower than what their body establishes at its highest weight (and there are all kinds of theories about why this happens). However, regardless of their weight control savvy, they can eat vegetables and lean proteins, and get in a work out most days, and they’ll be particularly motivated to do so if they feel their doctor still cares about their health and honors their work toward better health, regardless of what happens with their weight.

    • Travis says:

      This is admittedly not my area of expertise as I’m not a clinician, but I think that Debra has the right idea. I think it’s important to focus on the healthy lifestyle changes first and foremost. Dr Arya Sharma ( has argued that weight maintenance (e.g. prevention of further weight gain) is itself a worthwhile goal. But I think this is one of those issues where the specific course of action and goals really depend on the individual.

      BTW – Dr Sharma and Yoni Freedhoff have recently published an obesity management book titled Best Weight, which you may find interesting.

  3. Jennifer says:

    I think that Jeffrey Friedman would probably argue that you wouldn’t be able to gain a lot a weight and keep it on, even if you wanted to. Eventually your metabolism would adjust to reduce your weight closer to what it was before.

    • Travis says:

      I think this depends on the individual. Some individuals are resistant to weight gain when consciously overfed, but most are not. For example, Jim Levine overfed participants by 1000 calories/day for 8 weeks ( The average weight gain was >10lbs, although there was a huge variation, and some gained hardly any weight at all. I don’t know of any longer studies off the top of my head, but given that the majority of the population is now overweight or obese, I would wager that most people are susceptible to at least some weight gain if they changed their lifestyle sufficiently.

  4. Travis says:

    Thanks for the comment, and for pointing out the Lancet study, Ed. I absolutely agree that the the general picture clearly shows that a high BMI is bad for your health at the population level. But there are still a few important limitations to BMI and general, and the “ideal” BMI in particular. For example, 92% of the participants in the Lancet review are from Europe, Israel, Australia, and the USA – e.g. mainly white. And we know that the ideal BMI is almost certainly a few units higher in white populations than in Asian populations. Then there are important age-related differences, like those discussed in this post, as well as others related to age-related changes in body fat distribution. So while there may be an “ideal” BMI at the population level, it’s really difficult to apply that clinically, and tricky to even make truly useful BMI guidelines in multi-ethnic societies like Canada and the USA (E.g. Should we use the current guidelines based on Caucasians, or should we make separate guidelines for each ethnic group? And what about individuals of mixed ethnic background?).

    The other issue is that the Lancet study, like several others, doesn’t suggest that 22.5 itself is the ideal – it suggests that the range from 22.5-25 is ideal (I think I’m reading this correctly, let me know if I’m not). And the additional risk of having a BMI up to 27.5 still seems pretty slim. This is similar to other studies, that haven’t necessarily suggested that the overweight category is the “ideal” one, but that it might not be tremendously different from the typically prescribed range of 18.5-25.0. All of this obviously doesn’t mean that it’s not a useful measure, just that it has some limitations.


  5. Travis, thanks again for yet another thoughtful post. As a molecular biologist turned science writer who has been following these phenomena for several years, I’ve come to believe, based on the work of people like Paul Ernsberger, Jeff Friedman (thanks Jennifer), and Rachel Wildman, that the old setpoint theory is more true than ever. There’s no reason to suggest to any patient that they either gain or lose weight – the important thing is to encourage healthy eating behaviors and physical activity that the individual finds enjoyable. Their body will take care of the rest, and their weight will stabilize wherever it stabilizes. One thing that becomes apparent from your post is that way too much emphasis is put on the number on the scale or tape measure. Our culture has encouraged, in many ways, a loss of connectivity with our own bodies. Most people in developed countries have learned from the media to hate their bodies. A century of this attitude has yielded nothing but problems.

  6. I bet diet does a long way towards explaining the obesity paradox. I’ve been making this comment a lot lately. Fat isn’t inert, so being overweight may not be a benign condition (and may be very involved in making it hard to maintain weight loss), but I think that the villain re disease is far more likely diet-induced inflammation etc which just so happens leads to fat gain as well.

    Certainly would explain the “even a 5-10% weight loss results in improved CVD markers” phenomenon that mainstream medicine likes to push as justification for weight loss. But it’s probably not the weight loss per se, it’s the result of eating nutrient-rich food rather than the normal quantity of industrial food in our diets.

    Or as Jonathan Christie puts it at the tail end of his very long (and interesting) essay re health and diabetes:

    Apparently, Dr Dean Ornish, Dr Robert Atkins, Nathan Pritikin and so many others among us have been as the blind men who described the elephant: each solved a part of the puzzle and thought they had deciphered the mystery. It does not much matter what is in the diet as long as the food is unrefined.

  7. Quinlan says:

    Do these obesity paradox studies adjust for age? Maybe the obese encounter these injuries/illnesses at a younger age and therefore they are less life threatening.

    Also I have always wondered about BMI as a measure of obesity, BMI >30 is very specific. My BMI hovers around 30 so when my BMI drops to 29.9 statistically my obesity is cured, I can eat a little less and exercise a little more and statistically I’m no longer obese. Compare that to my wife who has a BMI in the 40s I think, she could eat a lot less and exercise a lot more and still be obese.

    Now because bmi is normally distributed there are far more “obese” people my size than obese people her size and because people my size are much closer to the obesity threshold we have much greater sway when it comes to statistical findings. Essentially someone my size has far more control (and lifestyle factor influence) over their obesity status than someone like her.

    So I suppose what I’m getting at is are we stigmatizing and breeding stereotypes about very fat people by focusing too much on obesity “rates” (which are easily swayed by people like me)? Basically because my obesity status and those like me that have gained a little bit and crossed the arbitrary threshold (causing the obesity epidemic) is mostly environmental/willpower does this mean that we end up underestimating the power of genetics for people with weights much higher than that?

    • Travis says:

      Many of the studies that have led to the idea of the “obesity paradox” are actually those studying elderly individuals – the ideal BMI seems to creep up in these individuals, and it’s thought that this is because this age group is more likely to suffer from injuries like hip fractures or certain types of cancer and heart disease than their younger counterparts.

      What is interesting is that the number of people that are slightly obese (e.g. BMI of 30-35) hasn’t really increased much in the past few decades. What has increased the most has been the number of individuals who have very high BMIs (for example:

      • Quinlan says:

        Even though the bigger categories are growing faster the moderate obese category is still numerically much bigger isn’t it? So if all of the people with BMIs just above 30 went for twenty minute walks and lost a few pounds the overall obesity rate would plummet.

        If instead all of the people with BMIs around 40 went for 20 minute walks and lost a few pounds this would go unnoticed and the overall obesity rate would stay the same.

        • Travis says:

          I think you’re under-estimating the difficulty of losing even a few pounds. But even if that were the case, I’m not sure that I understand the importance of it. Regardless of what specific BMI we choose to classify an individual as obese, the BMI distribution of the population has shifted to the right in recent decades (more people are heavier, less are lighter), which has important health consequences. Looking at temporal trends in the prevalence of obesity is a simple way to illustrate that, but it’s just one way to do so.

          • Quinlan says:

            Thanks for the replies Travis! I don’t actually know if I have a point. 😀

            I think basically what I’m getting at is those fixed thresholds are good for telling us what might cause inter-group differences (USA today compared to 30 years ago) in body weight but not inter-individual differences in body weight within those groups. We get so caught up in the fact that the distribution is shifting that I think we sometimes forget or downplay the shape of that distribution. At the individual level people are fatter now, but they still roughly resmble their parents and grandparents.

  8. Pingback: Quick Links | A Blog Around The Clock

  9. Jennifer says:

    @ Travis: when I said that Friedman would not agree that people could gain weight and keep it on, any more than they could lose weight and keep it off, I was referring to a LOT of weight – certainly more than ten pounds (more like 50 pounds) and keep it on (over the long-term; much longer than 8 weeks).

  10. Eric Melse says:

    Check this research that shows that the “obesity paradox” is actually a statistical bias.