Lightening the load of heavy weight research

There’s a new study out on weight and mortality risk this week.  What is it saying?

It depends on who you ask.

If you ask the press, they’ll say this:

Or this:

Yuck!  That sounds just dreadful.  Why are they saying this, what does this mean, and is it true?

First, let me fill in some back story.  In 2013, prominent epidemiologist Katherine Flegal and co-authors published a paper examining relationships between body weight and all-cause mortality (risks of death from all causes).  What they found was a lower mortality risk in the so-called overweight BMI category of 25-30, and not-increased risk in the so-called obesity I BMI category of 30-35.  Their results ran contrary to conventional wisdom (so much for conventional wisdom…).  They also unleashed a furious and very rude backlash among prominent and heretofore relatively well-behaved public health  and obesity researchers.  Here are a few reactions:

“It’s a horrific message to put out at this particular time. We shouldn’t take it for granted that we can cancel the gym, that we can eat ourselves to death with black forest gateaux.”
UK National Obesity Forum

“You’d hate to have the message get out there that it’s good to be overweight. The reality is that people who are overweight very often become obese and that’s clearly not good.”
Mercedes Carnethon, Northwestern Univ. School of Medicine

Since the Flegal et. al. 2013 article, some researchers who disagree with those findings have been trying to explain how being “overweight” (I use the quotes because I’m referring to the BMI category of 25–30 here, not any description of a person’s body) can lower your mortality risk.  Andrew Stokes, a population health researcher at Boston University, has been working on trying to tease out what’s going on with weight changes over time and mortality (death by any cause).  In a bunch of recent papers he and his coauthors have looked not just at BMIs and death rates, but at maximum BMI of individuals and possible relationships between that max, trends in their BMIs over time, and death rates.  (side note: my friend Dan and I are working on an article addressing Stokes’ work, which is in progress.  I’ll certainly blog about our work when we have results).

This newest paper looks at population data from three very big longitudinal studies and concludes that we can explain the so-called “obesity paradox” (that BMI 25–30 confers lower mortality risk rather than increased mortality risk) by looking at maximum BMI.  Those with maximum BMI of 25 or greater had increased mortality risk compared to those with maximum BMI of <25.

Ah– so being fatter really is bad for you.  Whew; public health and medicine don’t need to change all that signage after all.

Well, maybe they do.  Looking at this article, I found some complicated and interesting results (which I’ve seen in other such articles, but aren’t splashed across the headlines.)

Interesting result one:  being “underweight” (BMI <18.5) carries a much greater mortality risk than being “overweight” (BMI 25–30).  For a lot of age/sex categories, it carries a much great mortality risk than being “type I obese” (BMI 30–35).  For instance, for non-smoking men< 70 years old, the mortality risk was almost the same for <18.5 BMI as for >35 BMI (2.89 and 3.19 respectively).  That is, people at the far ends of the weight spectrum measured both had much increased mortality risk.  Again, we are talking about maximum BMI here (just to be precise).

Interesting result two:  the mortality risks from a particular max BMI shift as the population ages.  The details are pretty complicated, but here’s an example:  if my max BMI is say, 31, then these results show how my mortality risks may go up and down as I age.  This is interesting and important for patients and health care providers.  Given some max BMI, the medical advice might be different depending on the age of the patient (and other features of her medical history).  Of course, many medical practitioners act on this already by paying special attention to many features other than BMI in caring for their patients.

Interesting result three:  the results are based on three very large samples (about 225,000 people) of white people– they made up more than 91% of the sample.  We already know that BMI distributions vary across racial categories, so these results (if they turn out to be correct), would not apply in a simple way to other groups.

Interesting result four: In the article, the authors point out that their targeted group (BMI 25–30) is pretty diverse with respect to body fat percentage and waist circumference.  They’re also going to be pretty diverse with respect to their eating and physical activity practices (like every other BMI group).  The authors think that they can use max BMI to identify who in the BMI 25–30 group is at increased risk.  But to what end?  It’s not like medical practice has any currently effective procedures for bringing about and sustaining weight loss over time (except maybe some forms of gastric bypass, which aren’t indicated for the population targeted in the article).  So, what is an appropriate response to this information from patients and providers, other than more moral panic?

For me, my response to this article is to dig into the details, talk to my colleague Dan about our article, and attend to my health-as-I-define-it in the best ways I know how.  I’m not convinced these folks are right.  And I’m not convinced that we even agree on how their being right might reasonably translate into anything medically useful or practical.  However, we all know that science, medicine and health care are super-complicated, so while we’re waiting for the fog to clear, let’s just do nice things for ourselves.  So I’m headed out for a bike ride now!