Is it really healthier to be a few pounds overweight? That’s not what the study says.

IMG_1330A new study published in the Journal of the American Medical Association (JAMA) indicates that a body mass index or BMI of 25-29.9 (overweight) is associated with the lowest risk of death and that class 1 obesity (BMI 30-34.9) is not associated with an increased risk of mortality. As this study hit the presses January 2nd (and I’m sure no editorial thought was given by JAMA to such a study coming out at the first of the year) when many people are thinking about weight loss resolutions, it was covered widely in the press and I read several op-eds claiming vindication for obesity. One op-ed on a major news site was indignant that CT scanners couldn’t accommodate a friend (some CT scanners have difficulty accommodating patients over 300 lbs). The author’s solution? Build bigger CT scanners because obesity isn’t bad at all. This new study proves it.

My answer? Back the truck up.

First of all the study doesn’t say that being overweight is good for you and that being an ideal weight is bad. What the study does tell us is that people who have a BMI of 35 or greater are more likely to die. This is not new information. A BMI of 35 is a lot of extra weight, depending on your height it could easily mean 70 extra pounds or more. 15% of Americans have a BMI of 35 or greater. Only people with a BMI over 35, way over 35, need bigger CT scanners. I’m not saying that severely obese people shouldn’t have access to imaging studies, but the answer to the epidemic of severe obesity is not to claim vindication based on the inaccurate interpretation of one study and simply build bigger equipment.

What about the lower risk of death in the overweight and class 1 obesity groups compared with the normal BMI group? Well, this can be explained by a variety of factors:

  • The wrong control group. Many researchers question whether the control group should really be a BMI of 22-24.9, not the wider range of 18.5-24.9 used in this study. The reason, many people at the thinner end of the scale are thin because of illness and this obviously skews mortality statistics.
  • BMI is an imperfect tool with which to predict mortality when the result isn’t one extreme (< 18.5) or the other (>34.9). This is not a new finding. BMI just looks at weight, not the proportion of weight that is muscle mass vs. fatty tissue. Many people with a normal BMI have very little muscle mass and thus are carrying around excess fat and are less healthy than their BMI suggests. There are better metrics to look at mortality risk for people who have a BMI in the 18.5-34.9 range, such as waist circumference, resting heart rate, fasting glucose, leptin levels, and even DXA scans (just to name a few). The problem is that not all these measurement tools are practical on a large-scale.
  • A small amount of fat may provide an extra energy reserve for someone who becomes chronically ill, thus skewing the survival stats. For example, consider the dramatic weight loss associated with chemo…if you can’t eat due to extreme nausea and you have a little extra fat then you burn fat, but if you have no fat and can’t eat then you start breaking down muscle. This is a phenomenon has popped up in a few studies and definitely requires more research, because obesity is definitely associated with worse outcomes in many cancers.
  • Not all fat is created equal. Belly fat, the metabolically active muffin top, is what contributes to diabetes and other inflammatory conditions. Having a few extra pounds around the middle is far worse than having a few extra pounds on the hips. Again, not new information. BMI doesn’t distinguish between belly fat and thigh fat.

What is very important is that we don’t take erroneous messages from this study (hello, health reporters for major news outlets looking for attention-grabbing headlines). This study says nothing more than we need better tools than BMI to assess mortality risk for people who have a body mass index between 18.5 and 34.9 and that BMI doesn’t predict “ideal weight,” it only tells us that extremes are bad. This study also confirms that the 15% of Americans with a BMI of 35 are at increased risk of dying prematurely, a point sadly missed by many.

Body mass index simply doesn’t convey enough information to assess mortality risk for 85% of the population, but that fact (which isn’t new) shouldn’t stop each and every one of us from striving everyday to be the healthiest that we can be.

And with that, I’m off for my run.

Join the Conversation


  1. Thank you for this. I cringed when I saw that study result, because I knew people would take it as an excuse to say, “See? Being overweight isn’t bad for you.” As you did so nicely, one needs to look at all the variables when assessing the outcome of this particular study.

  2. I’m actually a little confused about the terms of these sorts of studies — specifically the phrase “increased risk of mortality.” We all die, right, so everyone has a 100% risk of mortality (at least to this non-science person). Is what they mean an increased risk of earlier than average mortality? or an increased risk of death from particular things?

    These are perhaps stupid questions.

    Enjoy your run.

  3. Thanks. I also wish we had a better tool than the BMI for most folks. And I think you are right about using a smaller control group. Just an individual perspective: I consider my self “very healthy”, but I have a chronic illness (that I mostly try to ignore). I personally feel healthier when my BMI is 22-23 than when it is under 20. It’s partly because that means my disease is not very active, and partly because I know I have some “cushion” if it does get more active. I think this is a more complex question than this study considered.

  4. Very insightful post. I work in the life insurance industry and we see firsthand how obesity affects mortality, and yes, the heavier you are, the more risk you carry. BMI may not be the best measurement when people are on the cusp of being overweight, but it is just as good as any for life threatening obesity.

    1. What the study also doesn’t speak to is correlation vs causation. People with high BMIs are routinely shamed by doctors (thus making them less likely to visit all doctors) misdiagnosed bc some doctors refuse to treat (beyond a recommendation to lose weight), and have no reasonable access to medical equipment for their size. (This goes for even cheap med supplies like BP cuffs, wheelchairs, and gowns, but it doesn’t excuse hospitals from having accommodations in more expensive life saving equipment like CT scans. A fat person with brain cancer still has a need for treatment.) Not to mention that when a fat person does take steps to improve their health (including healthy food choices, movement, and not hating themselves), they’re often not supported in those efforts if these things don’t cause weight loss, which makes the healthy behaviors seem like they have no value. All these things will obviously affect health outcomes, and have nothing medically to do with weight.

      And even if it could prove that obesity on its own causes bad health outcomes, it doesn’t say how to “cure” it (there is no reliable way to do that). Extremely tall people often have worse health outcomes, but whether it’s caused directly by height or not, the bigger question is how to lessen risk, not how to change the patient.

      Signed, a fat person who walk-toured a huge city yesterday, and will do it again today and tomorrow.

  5. You make several good points here. There is also the issue identified by Alex Hutchinson here: He notes that a previous NEJM paper was very similar to the JAMA paper except that instead of trying to adjust for the effect of smoking statistically, they reported two sets of results, on for those who had never smoked and another for the rest. Unsurprisingly, the optimal BMI for those who have never smoked is appreciably lower than that reported in JAMA, and in fact much the same as previous received wisdom.

    The trouble is that smoking is such a powerful confounding factor, so intertwined with so many aspects of health, that it is not possible to make it disappear with purely statistical techniques, even with 2.88 million subjects. That is not to say that the JAMA finding is not important; Hutchinson puts the matter well:

    “The debate between these two approaches can be summarized as generalizability versus validity. The fact is that many people in society smoke or have smoked in the past, and many people have serious conditions like cancer or heart disease — so if you want data that you can accurately generalize to the population at large, you need to keep these people in the analysis. If you want data that’s valid and not skewed by factors like smoking, the other side argues, you do have to exclude them.”

Leave a comment

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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s

%d bloggers like this: