There has been much discussion about obesity and COVID-19 recently, with various claims being made about how obesity is the reason why the UK and the US has such high deaths rate. There are now headlines suggesting the Prime Minister is so convinced by these claims that he will “declare war” on obesity. This is entirely wrong, and it is wrong for a whole host of reasons. I will set out why in general terms rather than go over the more technical statistical problems with the claims, because I believe they are enough to show that the claims are wrong. Why does it matter? Because it distracts us from the real reasons why we have more deaths than places like Japan and Singapore and because it attempts to place blame on those who have died rather than those who should have prevented people from dying.

Let’s start with the most damning fact. “Weight” (in the sense of too much or too little) is the result of a highly complex interaction of genetics, metabolic rate, diet, gut bacteria, age, health history, anatomy, exercise and, possibly even (according to some epigenetic studies) the diets of our parents and grandparents. We understand very little of the interaction between these factors and are only now beginning to discover the role of things like bacteria in calorie absorption and the possible role of antibiotic overuse in obesity. No doubt all or some of these factors do play a role in whether we are infected by a disease and the subsequent progression of that infection but the idea that we can reduce all of these down to a single number that we can then use to say that the UK has more deaths than say Japan is frankly absurd.

We do not and cannot know how much anybody “should” weigh. There are a number of studies that show that many people characterised as overweight and obese are “metabolically normal”. At the same time, many people characterised as normal weight have significant metabolic health issues. We cannot therefore say that there is a weight that is ideal: it varies from person to person dependent on a myriad of factors and probably varies in time for individuals according to a whole range of factors, including age. In other words, a BMI that is healthiest (itself difficult to define) for one person may be much less healthy for another person. Supporters of BMI argue that these differences disappear with averages across a population. But as we learn more and more about humans, we see more and more that heath is about individuals. It is not just ridiculous but dangerous to say that a certain weight to height ratio is ideal for a population when it may not be ideal for a single individual.

Evidence suggests many Western countries have such high death rates is because of serious errors in care settings

That brings us to BMI. This is a simplistic measure with many acknowledged faults. It cannot distinguish between fat and muscle, between ectomorphs and endomorphs, nor between the young and the elderly. The latter point I will come back to. Further, the bands within BMIs are completely arbitrary. Weight and height (the components of BMI) are both on continuums. There are no step changes in either. Worse, the definitions themselves are arbitrary – it is not clear that there is a biological meaning to “obese”, let alone “overweight”. 

At the other extreme, BMIs for the elderly are meaningless. Past about 50, we all start to shrink in height. Various other processes also occur, resulting in less lean muscle and more fat, no matter what diet or exercise a person undertakes. Higher BMIs in the elderly are thus often the result of entirely different processes than higher BMIs in the young, and past 60, to some extent at least, if you measure higher BMIs you are simply measuring increasing age. 

We know even less about optimum BMIs for the elderly than we do for the young, because it is only recently that we had substantial populations of people over 80 to research. And that brings me to a key point about the elderly today. Somebody who is 85 in 2020 was born in 1935. They spent much of their childhood subject to food rationing. They are the people who are responsible for the increase in longevity in the UK and the US. Our current elderly are the longest lived people we have ever seen. To claim that they are “fat” and “unhealthy” is absurd. Many, if not most, have never been fat, and certainly not obese, for most of their lives. Lumping the elderly in with the young in terms of health problems caused by obesity is nonsensical. These are people who have shown they are the “healthiest” people we have: they have reached an advanced age without dying – is that not the ultimate measure of a healthy life?

We must also mention the so-called Obesity Paradox. A large number of studies have shown what the epidemiologists claim is a paradox – higher BMIs do not show any difference in terms of morbidity and longevity versus normal weight BMI. They do however correlate with better recovery from major health events such as heart attacks. The fact that they call it a paradox and not something that should make them think again about their theories never seems to occur to them. Only at the morbidly obese levels of BMI is there any increase in risk but similar increases in risks are found in the underweight too. 

We can now turn the claims about COVID-19 and the obese. Before we do so, we must look at the basics of epidemiology. Epidemiology works by looking at statistical differences between groups to see if something increases or decreases a risk. For serious illnesses epidemiology can use only observational studies – we cannot conduct randomised clinical trials on humans for diseases that might kill you. The problem is that teasing out the effects of one possible variable – obesity say –  from all the possible factors is extremely difficult. There are factors that may be extremely important – genetics for example – that at this point we simply cannot know, let alone control for. There may be tens or hundreds of other others we don’t know we don’t know. Thus any observational epidemiological study is fraught with problems. 

Epidemiology uses the risk ratio (amongst other names) to try and quantify this. It calculates how much less or more risky something is compared to the control. The classic study, and the one on which all epidemiology is based, is smoking. Sir Richard Doll was able to show that smoking increased the risk of lung cancer by 10-20 times. That was a huge increase in risk. When you see that sort of number you can be pretty sure that you have something with real, large effects. 

In 1994, the National Cancer Institute stated that:  

“In epidemiologic research, increases in risk of less than 100% are considered small and are usually difficult to interpret. Such increases may be due to chance, statistical bias, or the effects of confounding factors that are sometime not evident.”

Indeed, the gold standard is still viewed as at least a doubling or halving of risk, before any claim can be made that a risk is real. Even then, significant work must be done to verify the risk exists.

So what are the claims about COVID and obesity? First, we can say that using BMIs is the bluntest of blunt tools. It would be frankly amazing if there was some real causal link between COVID-19 and BMI. People are different height and different weights for a vast number of different and interacting reasons. My BMI of 31 is caused by completely different factors to your BMI of 31. We have different heights and weights, different physiologies, different genetics. Yet because we have the same BMI we have the same shared higher risk. 

Second, we can look at the claimed increase in risk. One study claims a 33% increase in the risk of dying. That is far too small for us to consider. It fails the doubling/100% test by some margin. We should simply ignore it. People find that difficult to accept but the simple truth is that the study shows that there is not an increase in risk. 33% is too small to be considered evidence that obesity makes you more likely to die. 

To further demonstrate this, we can look at the data. The table below is from the ICNARC report on COVID-19 in critical care 08 May 2020. It shows the percentage of COVID patients in ICU who were either discharged or died. The table also shows for comparison the average percentage of ICU patients who died from viral pneumonia in previous years:

	COVID-19		Viral Pneumonia
BMI	Discharged alive	Died in critical care	(2017-19) Died
<25	53.7%	46.3%	22.9%
25-<30	51.2%	48.8%	22.9%
30<40	55.7%	44.3%	19.7%
40+	56.0%	44.0%	15.0%

 

It is difficult to show any effect due to BMI there. The raw data shows that BMI plays no role in determining whether you live or die.

We can add a further table showing the outcomes of those who received respiratory support (the biggest health problem for serious cases of COVID-19) whilst in ICU:

	Advanced Support Discharged Alive	Basic Support Discharged Alive
<25	42%	81%
25-<30	39%	83%
30<40	45%	85%
40+	44%	80%

 

Again, this fails to show that higher BMIs have higher risks. 

So why is there a claim that being obese increase the risks? The study controls for all the real risks. If you ignore age, health issues, being male and whether or not you have a goldfish, then the risk of dying if you are obese is 33% higher. But if you control for all the real risks, the remaining risks are meaningless. We can describe this in simple terms. If we did not have the 18% of the population who are over 65, we would do away with a hugely disproportionate number of COVID deaths – around 88%. BY contrast, if we did not have the (say) 30% of old people who are obese we would do away with 30% of deaths. 

Thus even though the 33% increase exists statistically, it is not a real risk. It is absurd to claim that the UK and the US have high death rates because of obesity. It is simply not true. We have a risk ratio that even after it has been massaged by removing the real risks is still far too small to consider as real. Moreover, the chances that something as crude and largely meaningless as BMI could tell us anything about who gets COVID-19 and who dies from it are zero. 

This is more than just an academic exercise. There is real evidence that the reasons many Western countries have such high rates of death is that we made serious errors in hospitals, other health centres and care homes. We needed to protect those most vulnerable to the disease – the frail elderly – and instead we did the exact opposite. What countries with high numbers of deaths share is high rates of infections in care homes and high rates of nosocomial infections. It is the actions that South Korea, Singapore and Japan took to make sure there was no cycling of the infection between hospitals and care homes and back again that are at least partly responsible for their low death rates. These fallacious claims about obesity distract us from the real issues. We have made a horrible mess of this epidemic, and we will not learn the lessons if we blame something that is not relevant and instead blame the victims of our mismanagement.