BMI is a key health metric

This transcript has been lightly edited for clarity

How does BMI help us understand health risks?

Body mass index is one of the best studied health risks that we have in the Global Burden of Disease – through hundreds of cohort studies, some of them really quite large, like the CPRD study in the UK  that alone had about 30 million person-years of observation.

When you put all these studies together, there are 60 million plus person-years of observation that tell us about the relationship between body mass index and the risk of some 30-plus diseases, cardiovascular diseases, diabetes,  a long list of cancers, and some musculoskeletal disorders. And the relationship between BMI and a number of those disease outcomes is not only well studied but extremely strong.

So we have, at IHME, an approach called the burden of proof approach, which is a very cautious interpretation of the evidence. We star-rate the strength of relationships, and a number of the relationships between BMI and health outcomes, like BMI and gout or BMI and diabetes, are four- or five-star relationships. And then there’s a long list of three-star relationships that are very strong as well, such as colorectal cancer, ischemic heart disease, and many others.

And then there’s a list of other outcomes where the relationship seems present but perhaps not as strong as those really overwhelming powerful relationships. So we know a lot about the relationship. We know that the risk is lowest for a BMI somewhere between about 22 and 25 and then starts to go up pretty steadily. And it’s one of our better understood relationships overall in the Global Burden of Disease.

What do we know about BMI and race/ethnicity?

There’s a lot of interest and studying of the relationship between BMI and race/ethnicity – is it different, for example, in Asian populations? Is it different in populations of African descent? Is it different in different parts of perhaps Eastern Europe and the like?

And from what we know, and our general rule is do we see a statistically different relationship by race, ethnicity, or country of origin, that there are some outcomes – for example, BMI and postmenopausal breast cancer – that do appear to meet those criteria, that there’s a different relationship in Asian populations versus non-Asian populations. But when we look at the relationship for cardiovascular disease, for some of the major larger cancers, so far we don’t see a statistically different relationship.

There’s been quite a lot of intense study in some of the larger cohorts, in the United States for example, looking at outcomes in Black versus white populations. And while any given study, because there are so many of them, may show one relationship or another, the systematic review – not ours at IHME but a published review – suggests that there isn’t yet evidence that that relationship is different.

So we suspect there are some differences in Asian populations, and for the rest of the world so far, our current assessment is that the relationships aren’t yet proven different and are likely the same. But of course we’ll all benefit from larger cohorts, particularly in non-white populations.

Is BMI the best metric to study obesity?

In the field around studying obesity, there’s a lot of debate and controversy about what is the best measure of risk associated with obesity. What we have very strong evidence for – massively overwhelming evidence – is that this metric of body mass index, weight over height in meters squared, is very predictive of mortality. We don't really know the mechanisms. There are lots of theories about the mechanisms.

And what it has prompted is that it’s likely related to adiposity, but we’re not completely sure. But because it’s likely related to adiposity, people have proposed other measures like the waist-to-hip ratio, direct measurements of adiposity. And there are small studies – instead of millions of person-years of observation, we’re talking about 50,000 or less person-years of observation – that suggest that some of these other measures, like waist-to-hip, may be even better predictors of some outcomes than BMI.

We don’t yet have the data at the population level to come close to assessing whether that’s really true at a broad level when you only have a few thousand or tens of thousands of person-years compared to millions for BMI. It’ll take a lot more evidence to show that that’s statistically more powerful than BMI as a predictor of outcome.

Can you be healthy with a high BMI?

Going back to a very discussed paper by Katherine Flegal and colleagues from the National Center for Health Statistics – now, almost a decade and a half ago or more, was this notion that maybe people who are overweight or even in the first category of obese above a BMI of 30 might actually be at lower risk.

And at the heart of this controversy is this question of what’s called the washout period. So when you get sick, let’s say cancer before you’re diagnosed, you often lose weight. And in a number of other chronic diseases that is true as well. And so when you just go to the population and measure BMI, you may find out that people with a BMI that’s overweight, between 25 and 30 by those standards, may be at lower risk than those at even lower BMI, because some of the people at low BMI are already in sort of pre-cancerous or pre-chronic disease.

So many analysts use a washout period. They exclude the first five years or more of observation and then look at the relationship between BMI and outcome. When you do that, you get these very powerful curves where for BMI, the lowest risk is 22 to 25 and then it starts to go up. If you don’t do that correction, you can sometimes find that BMI risk is lowest in overweight or even slightly obese people.

And so that’s a super important part of making sense of the data, this washout period. The other part that people react to is you look at people who look very healthy, that because of musculature may turn out to have a BMI that’s overweight or even slightly obese. And, people think, well, if they’re very heavily muscled, they can’t be at increased risk.

And the answer is we don’t really know. If the mechanism is purely adiposity, then it would make sense that those particular individuals are not at higher risk. But since we tend to look in these studies at the baseline measurement and then follow people up for 10 or 20 years, it may turn out that people who are at baseline are very muscled with low adiposity.

And there are studies that suggest that they become more likely to be obese in the future as muscle converts to fat when they become perhaps less active. So we don’t quite really know what the fact is. All we know is empirically, BMI is predictive on average, and it may well be the case that there’s a subset within that group that are at lower risk because of lower adiposity.

But we don’t really know. We don’t have the evidence to say that clearly. We don’t have the cohort follow-up data that tell us that answer. 

Can GLP-1 drugs combat health risks of a high BMI?

We know that BMI is a huge risk. We know that it’s getting worse than the big risks in the world like tobacco, which is huge but going down. BMI is huge, but going up and it’s going up pretty much everywhere in the world. We have not a single population in the world where we have a statistically significant reduction in obesity or BMI that we've observed in the last 30 years.

So that’s a very telling statistic that this is a risk that grows. And up until recently, the solutions on offer were all about individuals adopting better diets, which would be great and beneficial for them, or more physical activity – also great and beneficial. And working together, that might reduce obesity, along with national policies around food or healthy food and food deserts and making sure people have access to healthy food, reducing subsidies on unhealthy food.

All those might work to improve obesity, but so far they haven’t yet. Now along has come a new strategy, one that’s very different. It’s not about changing people’s lifestyle and changing their food and physical activity, but it’s using this new class of drugs called GLP-1, drugs that actually seem to have quite a profound effect. You see reductions from some of the the now FDA-approved drugs of 15, 20, even 20 to 23% on average in body mass, and that translates into real health benefits.

So we’ve had randomized clinical trials published showing a reduction in cardiovascular outcomes, death, and major events that are quite substantial associated with these GLP-1 drugs. And they both reduce weight but prove the causal mechanism that that reduction in weight is leading to a reduction in death rates and major cardiovascular events. So put together, this opens up this new possibility that that broader use of GLP-1s might be part of the solution for rising obesity in the United States and in pretty much every country in the world where we’re monitoring the data.