Yet another study was published this week that challenged the concept of ‘metabolically healthy obesity’. MHO is a phenomenon that has been recognised for over a decade now, with more and more studies coming out that suggest not all fat people are unhealthy. How many exactly varies with study, but a 2011 review found that the larger cohort studies tended to put the figure at about 30-40% based on BMI. Interestingly, when ‘obesity’ is defined based on body fat percentage, the rate is much higher - possibly around 50%. Similarly, around 25% of not-fat people (BMI <25) are classified as ‘unhealthy’. I’ll come back to that.
So what is ‘metabolically healthy obesity’? Well, in the scientific literature, ‘obesity’ is usually defined by BMI categorisation, i.e. having a BMI of 30 or higher. In some studies, they use other measures, such as body fat %, and occasionally, they lump ‘overweight’ and ‘obese’ participants together in their analysis (i.e. BMI over 25). More important though, is how they define metabolically healthy. You better make sure you’re sitting comfortably – this gets a bit complicated. And if you want to skip the science bit, just jump on down to the next paragraph. The most frequently examined metabolic health markers are: abdominal adiposity, insulin sensitivity, triglycerides, HDL cholesterol (the good stuff), blood pressure, and inflammation. But how many of these factors are included, and how each is measured, varies by study. Some studies define ‘healthy’ as having zero risk factors, others as up to 1 or 2. If you’re looking at health across all weights, high waist circumference might be included as one risk factor. Alternatively, you could assume that ‘obese’ people automatically have high waist circumference and look at the number of additional risk factors. Some use different tests or different blood markers to assess a risk factor, or the same test but different cut-off points. Like I said, it all gets very complicated. But the vast majority of studies define MHO as having a BMI of 30 or more, and, at the very least, impaired insulin sensitivity. A 2014 review identified 15 cohort studies (a group of people followed over time) and 5 cross-sectional studies (measured at a single point in time), and for each one, describes what was measured and what they found. It’s freely available if you want to check it out.
The evidence from studies of metabolically healthy obesity is pretty clear - individuals who are ‘metabolically healthy’ do better than those who are not, regardless of their weight. So MHO generally fare better than thin people with a poor metabolic profile. Most find no difference between obese and non-obese metabolically healthy. But as the entire medical community goes into paroxysms of terror that maybe fat isn’t so awful in and of itself, one question that has been raised is whether MHO is just a temporary state, a stepping stone on the path of metabolic deterioration. This isn’t an unreasonable question, and it is this that the study published this week, by a team at University College London, tried to answer.
The team used data from the 1991-1993 Whitehall II Cohort – 8637 British civil servants (government employees). They limited their analysis to those for whom data on BMI and metabolic status were available at baseline and after 20 years, which was 2521 individuals. ‘Obesity’ was defined by BMI, and metabolic health by presence of fewer than 2 of the following risk factors:
- Low levels of good cholesterol (HDL < 1.03mM for men, 1.29mM for women);
- Elevated blood pressure (130/85 mm Hg or higher) or use of any anti-hypertensive medication;
- Raised fasting plasma glucose (5.6 mM or higher) or use of anti-diabetic medication;
- High blood triglycerides (1.7 mM or higher);
- Evidence of insulin resistance (HOMA-IR > 2.87)
In other words, you couldn’t get a metabolically healthy label if you had more than one of these, so this was a reasonably stringent definition of MHO. Stringent is good in this case – it means you’re not including people who are already fairly sick in your so-called ‘healthy’ group. So what did they find?
Well, at its simplest level, the data showed that people who were metabolically healthy at the start of the study were more likely to develop new risk factors if they were obese at the start than if they were non-obese. And it is this finding that has been widely reported. And correctly, which makes a nice change. Well done Health Press. But the wider implications of this finding are a little bit more nuanced.
Let’s look at numbers first of all. Out of their 2521 participants, 181 were ‘obese’ – just under 8%. The rest were classed as ‘non-obese’. Presumably, this includes ‘normal weight’, ‘overweight’, and ‘underweight’ individuals – this isn’t clear from the paper. So the first thing to note is that this population isn’t typical of the UK population at the time. The earliest Health Survey for England data, collected in 1994, estimated that about 15% of the British population was ‘obese’ by BMI categorisation. While this might limit how representative the findings are, it isn’t in itself necessarily a huge problem. On then other hand, another third of us were classified as ‘overweight’ at that time – a number that hasn’t changed much in the 20 years that followed. And this is the first potential problem in interpreting the findings of this analysis. Health outcomes for ‘overweight’, ‘normal weight’, and ‘underweight’ are not identical. Not by a long shot. Underweight people generally have the worst health outcomes, but since there are not that many of them in population samples, this isn’t likely to have a huge impact. On the other hand, ‘overweight’ people usually have better long-term outcomes than do ‘normal weight’ people. If ‘overweight’ were added to ‘normal weight’ in the ‘non-obese’ group in this study, the outcomes might have been better than would have been expected with only ‘normal weight’ as the comparison group.
Second, the main outcome reported in this analysis (and the press) was the likelihood of becoming ‘unhealthy obese’. So it compares how likely ‘healthy obese’ are to become ‘unhealthy obese’ compared with ‘healthy non-obese’. And the numbers that made it into all the press releases were that at 5 years, nearly 4 times as many ‘non-healthy non-obese’ had become ‘non-healthy obese’ as those who started out both healthy and ‘non-obese’. In other words, they were more likely to get fat. It doesn’t tell us much about any changes in their metabolic health. Compared with this number, ‘healthy obese’ were 11 times more likely than ‘healthy non-obese’ to become ‘unhealthy obese’. So we’re not really comparing the rate at which their metabolic health is deteriorating – we’re confounding it with a measure of how likely those healthy thin people (or at least not-fat people) are to become unhealthy AND fat. At 20 years, the difference is a little smaller 3 times and 7 times more likely (rather than 4 and 11), but we’re still looking at the wrong thing.
Since we know that metabolic risk factors are more important for outcomes than is BMI category (the entire reason for conducting this analysis in the first place was to determine if the known protective status of healthy obesity was clinically relevant or whether ‘healthy obesity’ was just a short-lived intermediate state), we don’t really need to factor in changes in weight. Some people may be naturally slim but still lead an unhealthy lifestyle and have poor metabolic health. What we really need to look at is whether ‘healthy obese’ are more likely to become ‘metabolically unhealthy’ than are ‘healthy non-obese’. Still with me?
And the answer is: yes, they are. But the ratios are nowhere near as high as the numbers reported. At every time point, it seems that fat people who start out with no more than one metabolic risk factor are about twice as likely to develop more risk factors than not-fat people who start out with no more than one risk factor. So yes, fat people are more likely to get sicker. Why?
Well if you’re a mainstream commentator, you will say that all those extra fat cells are the cause of the problem. This is possible. Fat cells are biological entities – they release chemicals that are known to impact on health, including insulin sensitivity. But it’s not as simple as that. Fat people are also more likely to diet. What are the odds that people who start out fat have dieted over the subsequent 20 years? Probably numerous times. And we know that the most likely long-term outcome of intentional weight loss attempts is actually weight gain. And while the evidence on the long-term effect of weight cycling (yo-yo dieting) is mixed, there is certainly some suggestion that it is associated with systemic inflammation (a metabolic risk factor). Inflammatory conditions are association with an increased risk of, among other things, heart disease, hypertension, and diabetes. Want to know what else causes systemic inflammation? Stigma. Which is increasingly being recognised as a fundamental cause of structural health inequalities. What are the odds that those fat people were exposed to direct and indirect fat-shaming over that 20 year period?
So even given the likely occurrence of dieting attempts and the practically inevitable occurrence of finding themselves in a stigmatising environment, we’re still only looking at about double the risk. Is there anything else that might be important? Well, yes. While it’s difficult to account for the effects of dieting in longitudinal studies, unless you ask people to keep a record at the time (which wasn’t done in this cohort), and practically impossible to control for weight stigma, since you’d be hard pressed to find a similar population who hadn’t been exposed to it that you could compare them to, the three most easily measured and high impact confounding factors are socioeconomic status (SES), smoking, and physical fitness. The Whitehall II Cohort was developed primarily to evaluate the effects of the social determinants of health, in particular SES. This data set has some of the soundest available information about SES out there, and the data from it informed much of what we know about the all-encompassing effect of SES on health – far, far greater than personal behaviours. And yet the authors didn’t adjust their model for it, despite adjusting for age, gender, and ethnicity. They also had information about smoking status (as well as a bunch of other health behaviours) and didn’t take that into account either. And last but not least, the cohort were also asked about their physical activity levels, which is the next best thing if you can’t actually test their fitness levels directly. The way physical activity was measured in this cohort changed over the 20-period (becoming more detailed and reflecting the growing awareness of its importance in determining health outcomes), but no attempt appears to have been made to account for it. Given that we know that physical fitness is one of the key factors differentiating between outcomes for different BMI groups, not to take it into account is unconscionable. For example, a 2014 review of MHO found that of the 7 cohort studies included that adjusted for exercise levels, none of them reported any difference in all-cause mortality between metabolically healthy ‘obese’ and metabolically healthy ‘normal weight’ and only one found a difference for cardiovascular mortality. And that study used a rather odd definition of metabolic health, which basically boiled down to not fat and not diabetic. Half of the studies that didn’t control for exercise found a difference (although half didn’t). You can find an up-to-date systematic review and meta-analysis of the fat vs fit data here (note, this is NOT the same as metabolically healthy obesity – one is about fitness levels and the other is about insulin resistance, lipids and so on). If you can’t be bothered to read a whole paper on it and just want a quick visual representation, this is the one that changed my life:
So yes, there’s a lot of stuff that could be going on here that we don’t know about. But perhaps the biggest problem with extrapolating too much from this analysis was the absolutely paltry number of ‘obese’ participants involved. Only 181 ‘obese’ people were in the sample at baseline. Of these, 36% were categorised as metabolically healthy – fitting in with what other studies have found. That represents 66 people. So the media farct we have seen this week, the headlines, the editorials, the conversations over the water cooler, they are about a study that found half of 66 ‘obese’ people who had one or fewer metabolic risk factors in 1993, had developed at least one more risk factor 20 years later. Um, OK. Also, half didn’t, even with all that other crap going on. And nearly 30% of ‘non-obese’ shared the same fate. But, anyway.
Which brings me to the title of this blog post. Not long after this study came out, I saw the following tweet from @GetActive GetFit.
The link is to a Forbes article about the study, with the headline ‘Healthy obesity is mainly a myth, study finds’. Well, I’ve dealt with the study. So now let’s have a look at the tweet. It always astonishes me how quickly people are to discount HAES(R), clearly while having no understanding of what it actually says and doesn’t say. Health At Every Size(R) is, at root, a non-weight focussed approach to health. It says, if you would like to improve your health, whatever your size, engage in health-giving behaviours, and measure your progress using health metrics. Weight loss, per se, is not a guarantee of health, or a measure of improving health. You want to improve your blood pressure? Move more, sleep better, stress less, and measure your success with a sphygmomanometer, not a scale. It demands the same evidence-based health care that is expected by thin people be applied to all people, rather than where a thin person is treated for what ails them and a fat person is put on a diet. This is not sound medicine. HAES does NOT say that all fat people are healthy. It does NOT say that anybody can be healthy at any size. It does NOT say that weight loss is evil. It simply does not focus on weight at all. Take care of your body. Do healthy things. Let your weight settle where it will. Is that really so hard to get behind? @GetActive GetFit, according to their Twitter profile, is about “Helping Others Get Active & Get Fit … Tips, Strategies, Equipment, Clothing & More.” I’m not seeing anything in there that would preclude a HAES approach.
And then there was the “your”. Enjoy YOUR HAES. Yo, fat pigs, joke’s on you, LOL.
So let me be clear. HAES is NOT about fat people. It is about health. About equal access to the tools and structures of health and wellbeing, regardless of age, gender, race, employment status, or yes, size. HAES is for fat people and thin people. Everyone is harmed by the current focus on body weight, as thin people are not treated for conditions that doctors think they can’t have, fat people are treated for conditions they don’t have, and eating disorders rise exponentially in ever younger children. Under fives, ffs. How is this helping anybody?
So even if the UCL study showed that every single fat person became a walking time bomb with more health problems than you could shake a stick at, HAES would still be for all of us. We would still deserve respectful and appropriate, ethical, evidence-based healthcare. And all of the evidence we have says that given our genetics and our environment, both of which are largely beyond our control, our best shot at optimising our health outcomes comes from adopting healthy behaviours.
And the thing about healthy habits, they work for all people. Whether you’re fat or thin, currently healthy or unhealthy, there is no real down-side from engaging in healthy behaviours. There is some indication that putting MHO people on a diet actually worsens their metabolic health. But encouraging to move, to not smoke, those are all pluses. Remember those unhealthy thin people from the start of the blog? Guess what would improve their health? Yeah, healthy habits. In the MHO literature we often see arguments about resource management – make sure we target our interventions at the fat people who need it. This is missing the point. There are 64 million people in the UK. Addressing public health interventions at ‘obese’ people would ‘inappropriately’ target approximately 19 million MHO and miss over 16 million metabolically unhealthy ‘normal weight’. Multiply those numbers by 5 and a bit for the US. Here’s an idea. Why not just encourage everybody to take care of themselves? Health is not measured by a 12-inch-square contraption in the corner of your bathroom.
But the thing that struck me most about that tweet was the sense of triumphant glee. Talking to a friend the other day she was bemoaning the way that the diet industry continues to distort the truth and manipulate the data to maintain the status quo. And I have to say, in some ways, that’s not the worst part of the anti-obesity rhetoric for me. Those people are doing it out of self-interest. For money, for power. For more money. It’s reprehensible but I get it. What I struggle with is the sheer hatred that is directed at fat people by huge numbers of supposedly ‘normal’ people, for no reason other than their size. That the idea that they maybe shouldn’t just go off and kill themselves and do everyone a favour. The study that started all this is an interesting study. It took a useful look at an interesting data set and asked a valid question. It is not perfect. We can talk about it like scientists, like professionals, like intelligent adults. Or we can put on our party hats and say Haha. Fat People. Nenenenene. Your choice.
*Statistical note: prevalence ratios in the paper are adjusted for age, gender, and ethnicity. This graph is based on actual case numbers as reported in Table 1 of the published paper.