The real facts about BMI

Your BMI is an indication of your relative weight, given your height. The standard way of calculating BMI was first developed in 1832, and is to divide your weight (in Kgs) by your height (in Metres) squared. I’m going to mix and match metric and imperial here, as I think most people tend to think in terms of both when it suits them.

So if you are 5ft 9ins (1.75m), and you weigh 13st 2lbs (83.6Kg) your BMI, according to the standard formula, is

83.6 / ((1.75) x (1.75))

which is

83.6 / 3.06

which is a BMI of 27.3.

And these happen to be the stats for an average man in Britain today (ONS data 2010). Which would place the average man firmly in the ‘overweight” banding according to the standard BMI classifications, which are:

Underweight:         BMI below 18.5
Normal weight:     BMI 18.5-25
Overweight:            BMI 25-29.9
Clinically Obese:    BMI 30-34.9
Severely Obese:     BMI 35-39.9
Morbidly Obese:    BMI 40+

There are however a number of concerns about BMI, and I want to explore them with you in order to give you a fuller understanding of this number, and how you should approach it yourself.

I’m going to focus on a new, improved BMI formula, known as the “Trethefen model”, and I’d encourage you to focus on this “new BMI” number – especially if you are particularly tall or short, as the first thing to say about the standard formula is that it appears to have sacrificed accuracy for simplicity.

If you are really tall or small, the standard formula for calculating your BMI doesn’t work quite so well, and it is worth checking out the new revised model, created using a formula developed by Nick Trefethen, Professor of Numerical Analysis, University of Oxford. Nick’s original paper is referenced here, and it provides a fascinating diversion into the maths of all this, as well as cheering up us taller folk who might be a bit depressed at our standard BMI measures!

For taller people (over 1m 85cm, or 6ft) this new, improved formula knocks at least a point off the standard BMI scale and for smaller folk (under 1m 55cm, or 5ft 1ins), it can add a point to your BMI. That might seem a bit unfair if you are smaller than average, but it’s better to confront the truth than hide behind a false number.

Here’s the grid to quickly check your BMI, using the Trethefen formula.

image 8

If you want to be accurate, you can click here to work it out specifically.

As I discussed in the book, for our purposes the dividing line at 30, between the “clinically obese” red area and the “overweight” orange area is where we want to concentrate.

Red, and you need to do some work to get out of Tribe 3. Green is the colour to aim for – particularly the emerald green (well the book is called “The Life of Riley”).

The second point of issue with BMI is whether or not it really tells us anything about the relative risks of being normal weight vs overweight.

Interestingly, some scientists query whether moving down from overweight to normal weight (i.e. getting your BMI below 25) actually has any further benefit. In fact, as Dr Malcolm Kendrick reported in a recent article in “The Independent” , looked at one way, the data suggests that the “best” weight to be for longevity is in the overweight 25-29 BMI range. Evidence from at least one large meta-data study in the US supports this view, and suggests that the risk of life-expectancy being shortened for the clinically obese is the same for those in the supposedly “normal” weight range, and with those who are “overweight” (i.e. BMI 25-29) having a statistically lower risk of mortality.

OK – this seems odd – a ‘get out of jail card” for folk who are carrying a few too many lbs. according to some experts. Why is this?

The arguments on this particular finding are still raging in academia, as one article in the magazine Nature explains. Scientists who dispute the findings cite both smoking (which lowers your weight & therefore BMI) and chronic illness, which can also lead to unhealthily low BMI readings as being part of the reason.

I think the most rational explanation for this statistical conundrum is by reference to a phenomenon known as the obesity paradox.

Mortality rates are in fact U shaped. People with very low BMIs are at as high a risk of early mortality (particularly older people), as people carrying too much weight. The logic seems to be that some diseases (heart disease, emphysema, Type 2 diabetes) are less risky for you if you are carrying a bit more weight. Researchers at the University of Alabama built a theoretical model to try and explain this, and the resulting findings look intuitively right to me – but as they say, there needs to be more work done on this. Below you can see what their model predicts the BMI mortality curves look like (they vary dependent upon age), and you can see where, if they are correct, those low BMI scores are driving really high mortality rates in older people (the yellow line for over 70s in particular stands out). In turn, these high numbers at the bottom end of the “normal” range for older cohorts explain why any “normal BMI” average might be higher than you’d expect.

image 9Interestingly this study also picks up on other research showing that the optimal BMI point (the lowest point on the curve, where the line has been artificially thickened a little bit) does increase as we age, from a low to mid 20s figure below age 60, rising to the mid to high 20s as we move into our 70s and 80s. This of course supports the theory that as you become older, and more susceptible to some of these chronic diseases, carrying a little more weight might prove beneficial (and is why, in the book, I flexed the ideal BMI target upwards for over 60s).

Looking at this as a layman however, one discrepancy here which would partially explain the “overweight is healthier than normal” finding might be because of the bandings themselves. It seems fairly clear to me that including the lower end of the “normal” band, from 18.5 to 20 BMI, into the “normal” figures is not helping comparisons, particularly for these older cohorts.

As I mention in the book, smaller, leaner women (around 5 ft 1 ins or under – one standard deviation below mean height for women) might just be able to squeeze into the 18.5-20 BMI range using the old formula and still be healthy, but for taller women and most men, having a BMI below 20 using the old formula would appear to be bordering on unhealthy.

I’ve done the numbers myself on how much more weight I’d need to lose to get to a BMI of 18.5 and it seems incredibly low to me. I would need to lose another 20Kgs/3 stone to hit a BMI of 18.5 and would likely have a body fat percentage of around 7% (similar to professional endurance athletes – I wish!) That seems quite severe, and not altogether healthy for an ordinary bloke like me.

In fact, checking my cycling stats, one of the world’s leading cyclists, Alberto Contador, is almost exactly average height, at 1m 76cm, and weighs 62 Kg, giving him a BMI of 20. Nearly all of the leading Tour de France contenders, guys like Chris Froome and Vincenzo Nibali have BMIs around 20. But these guys are professional cyclists, where every lb they don’t need to carry is critical. And therefore like all cyclists, they are unbelievably skinny. Alberto would need to lose another 5Kg to get down to a BMI of 18.5!

Some data released at the end of 2015 by Chris Froome sheds fascinating light into the world of elite cyclists and the importance of weight to them. Chris released a bunch of data on his physiology in order to disprove doping allegations, and in and amongst that data there were references to his weight at various points in his career. Chris is 1m 85cm tall (almost 6ft 1 inch), and in 2007 went for some testing in Switzerland when he was starting out on his career. At that point he weighed 75.6 Kg and had a measured body fat percentage of 16.9%. his standard BMI at that point was 22.1.

When Chris released his 2015 test data, his weight had dropped to 69.9Kg, with a body fat percentage of 9.8%. At this point, this double Tour de France winner had a standard BMI of 20.4. Only at the very point he had won the Tour de France, 5 months earlier, was his weight appreciably lower, at 67 Kg (no surprise after 3 weeks of cycling). Even at this point his standard BMI would have registered 19.6. By my calculations, at this point his body fat percentage was likely to be around 7% – simply incredibly low.

If a TdF winner can’t get below a BMI of 19.6 on the standardised scale at the very peak of his powers, when every ounce lost could be the difference between winning and losing, I’m not sure 18.5 can be considered normal for virtually any reasonably sized man.

You only have to look at them to realise just how skinny top cyclists are. I just can’t believe people who are even thinner than Chris or Alberto are not running some elevated health risk, so even though BMIs of between 18.5 and 20 are still classed as “normal”, might they be biassing comparisons between BMI bandings?

So why is 18.5 to 20 considered part of the “normal” range? Well for one thing smaller women, as I mentioned, can get this BMI score using the normal formula if they are lean, without being underweight. However, a 5ft 1ins woman who weighed 7st 5lbs would have a standard BMI of 19.2 but would see it increase to 20 using the Trethefen model – so maybe using the Trethefen model would eliminate the need to keep these very low BMI numbers within the normal range. And we are in any event probably only talking about a small percentage of the total female population here though, with around 15% or so being 5ft 1ins or under.

In addition, the bandings were agreed by the World Health Organisation back in 1997, with 18.5 to 25 set as the “normal weight” range. If this has been set as a global standard, is it perhaps the case that this lower segment has been included because of specific 3rd world issues relating to much smaller heights overall in many 3rd world countries and weight/nourishment concerns etc? I don’t know, but I can see clearly why it is unhelpful from a western perspective.

At a population level, western, advanced economy BMIs have clearly shifted over time – data from the US suggests male BMIs have increased by over 3 in the last 50 years, from under 25 to an average of 28. For women, the numbers from the states look even worse, rising from 24 to 29 as an average. OK – some of that is us putting on more fat, but perhaps some of it might just be a healthy gain.

We are generally bigger in the West than we were 100 or even 50 years ago. For a start, we are breeding bigger, taller people in general (just look at the Dutch!) We have also had better nutrition as we grow for a number of generations, so it’s no surprise that we are taller, and more muscular than our forebears, and tallness increases the standard BMI measure for the same level of physicality, so that will have some effect. In fact when you look at old movies it is easy to see how short and skinny even the healthiest of actors looked compared to today.

So whilst 18.5 might have been the low end of normal BMI 50 to 100 years ago in the west, I’m not sure it is now except for that one subset of smaller women – and most of that would disappear if we used the Trethefen formula. The other factor to think about here is that even if you take the average across the line from 18.5 to 25, the number of people actually in that 18.5-20 segment are tiny compared to those above it – if we weighted risk by size of population many of the comparative effects we are talking about would disappear.

So the comparisons between normal and overweight bands might be skewed. Today, for most people (taller women and nearly all men) having a BMI of 18.5-20 might mean you are at elevated risk of some diseases, and averaging those risks out might mean that the 18.5-25 band overall looks worse than the 25-29.9 band.

If you started “normal” a bit higher, at say a BMI of 20, then the comparative benefits of “normal” vs “overweight” would probably reassert themselves, and some studies are starting to do this.

Perhaps the best approach to this was taken by a team of researchers at City University in New York who went even further – and reanalysed overweight and obese data, comparing mortality rates only to data from people with a 23-25 BMI (recognised as the sweet spot for optimal life expectancy) which showed both overweight and obese BMIs were associated with much higher mortality risk.

And some of this theorising is borne out by NHS research. The NHS NOO meta-study from 2010 suggested “… in both sexes, mortality was lowest in the upper end of the ‘healthy’ range (BMI 22.5–25 kg/m2 )…”, with a BMI of 22.5 being the perfect point – towards the high end of the “normal” range. This meta study also noticed an uptick in mortality for BMIs below 22.5 – hence my suspicions about “overweight” being better than “normal” might just be due to the inclusion of more of these quite low BMIs.

So, I wouldn’t panic if you had a BMI in the overweight (25-29.9) range, but I think on balance the evidence suggests shifting it down to the top end of the healthy band would be beneficial. There is much evidence to support other health benefits being derived from lowering your BMI to below 25 in terms of avoiding type 2 diabetes etc to make this still a sensible target.

In the end, let’s stick with the art of the possible. Given we are starting with a UK average BMI of over 27, anything below this and above 20 is probably in the “sweet spot” as far as risk reduction is concerned, and the closer to the middle you can get the better, with that 22-25 goal I talk about in the book still to aim for if you can, with 22-23 if you are under 60, and 22-25 as an over 60s goal.

If we use 22.5 as a sensible target to aim for for those under 60, this simple table gives you a clear idea of the ideal weight for your height, using the Trethefen model:

image 10



Don’t fret about being a few lbs or a Kg or two either side of these numbers – they really are just a guide.

I am assuming most people who read the book are focussed on getting their weight down from a BMI above 30, but if you do have a BMI well below 22.5 I’m not suggesting you go out and eat cream cakes all day (although some doctors have suggested that!)

However, very low BMIs, below 18.5, are definitely associated with increased risk of illness and early mortality so if you are in that category I would suggest seeing your doctor just to check you are OK, and if you are a man skinnier than Alberto Contador or Chris Froome, with a BMI below 20 (or a taller than average woman with the same sort of BMI), that’s worth checking too in my view just to be safe.

One final criticism which undermines BMI is that it is just a proxy for how much fat you are carrying, not a direct measure, and therefore it can be affected by the ratio of muscle to fat you have.

Professional athletes (rugby players or rowers spring to mind) who are the same height and weight as their sedentary spectators will enjoy the same BMI – but have radically different physiques! As Nick Trethefen points out in his paper, the effects of adding or losing a bit of muscle don’t really alter your BMI number much, although replacing fat with muscle, whilst keeping your overall weight constant, is on balance clearly a good thing!

In fact, a US study in June 2008 examined 13,601 people and showed that only 21% of men and 31% of women had BMI-defined obesity (BMI > 30). Yet using body fat percentage data instead to define obesity, it was found in 50% of the men and 62% of the women (according to WHO standard thresholds). That’s a huge gap.

In the same study BMI values in the intermediate BMI range of 20–30 were found to be associated with a wide range of body fat percentages. For men with a BMI of 25, about 1 in 5 had a body fat percentage below 20% and 1 in 10 had body fat percentage above 30%. So the same BMI number might be hiding two significantly different fat stories and therefore you might not be out of the woods if your BMI is OK, if you are still carrying excess fat.

One recent US study found that 30% of those with a “normal” BMI did carry some markers for weight-related ill-health, such as poor blood pressure and/or glucose, cholesterol and triglyceride levels, with 70% rated healthy. For people in the overweight category (BMI of 25-29.9) the results were almost 50%/50%. Of those people with a BMI over 30 (i.e. clinically obese) 75% were carrying those ill-health markers, but fully 25% weren’t.

The research findings were being used to claim BMI was therefore a faulty indicator of health, which I thought was a poor conclusion to draw. A direct correlation between higher BMI and increased measures of ill-health seems pretty useful to me, even if it isn’t a perfect match. Also, I’d have wanted to delve a bit deeper into the research myself as I’m not sure if they filtered out fit, well-muscled obese folk (gym bunnies or rugby players for example, who would know they had a low body fat whilst being heavy for their size and would probably test as healthy) and also if they did any stratification by age (as we know, obesity gradually worsens health over time, so younger obese people will certainly show fewer ill-health markers). I’m also fairly sure they were using a standard BMI scale, not the Trethefen formula, so some healthy tall people in the sample may have been incorrectly categorised as obese even if they weren’t. So the numbers might shift a bit if these factors were brought into play – but in any event the study seemed to me to prove two things.

First, you are clearly more likely to have ill-health markers if you are obese vs normal weight (75% vs 30%), so it does appear to be actually a good indicator of health, not useless at all. Second, it does show that BMI on its own isn’t as accurate as it could be, with a lot of false positives (High BMI but healthy) and false negatives (low BMI but not healthy). It’s probably the false negatives I’d be more concerned about. If you have a high BMI but are healthy, you’ll suffer no harm from bringing your weight down a bit, but it’s more of a worry if people in the “normal” band think they have a clean bill of health when they too might be suffering from weight-related health issues.

Which leads me to what I consider to be a much better approach, which is to focus on your body fat percentage instead. This can easily be measured by buying a set of special scales. Prices range nowadays from £10 upwards – although I’d suspect a slightly better set will cost a little more, and would encourage you to buy one of the higher end sets if you can. I use a Tanita, and they range from around £30 to £70, depending upon functionality. Whichever scales you buy, make sure they use bioelectrical impedance analysis to work out your body fat. The Tanita ones I use seem pretty reliable, and the body fat percentage figures they give me match those I am given when I occasionally go for a professional private health check.

I’ve been using mine for over 5 years, and I measure my weight and body fat percentage every day.

I don’t obsess about it, if it goes up or down, I just take it and stick the results into a spreadsheet. I now have over four years worth of data. I know – geek.

The body fat percentage change is for me a much better indicator of how healthy you are. WHO data suggests that a body fat percentage over 25% for men and 35% for women is the equivalent of the BMI>30 number in terms of defining “obesity”. I’m not sure these numbers are too helpful though – as the US study above suggested over 50% of men and women would be defined then as obese.

I prefer the table below, which a) varies by age, and b) adds in an “overfat” category before obese. So for a typical 40-59 year old woman, 34% to 40% is overfat and 40%+ obese, and for a man it’s 22% to 27% overfat, and 28%+ obese.

image 11

And here’s a good example of the body fat percentage and BMI trends I have had on the “Life of Riley” habits:

image 12.png

In fact, looking at the two lines side by side (this is over the 16 months since I started), the decline in my body fat percentage (red) looks far more impressive than the shift in my BMI (blue).

In the end it is the amount of fat you are carrying which is the acid test, so being able to measure that, and hopefully see it reduce over time, is the goal. Tribe 3 could better be described as people with body fat in the red zone above.

It’s also interesting that the body fat charts have an age-related shift in them. Given that we know having a little more fat might be helpful in coping with late-onset chronic illness, this makes sense.

Is there one single number to aim for? Well, given we know being too lean carries some elevated mortality risk, if the graphic above does capture the right boundaries between athletic, fit, healthy etc, I’d say being on the boundary between fit and healthy was the sweet spot, and you can calculate that for your age quite simply:

For men over 30, it would be

16% + 0.1% x (your age – 30)

So if like me you were 56, the number would be

16% + 0.1% x (56-30)

=16% + 0.1% x 26

=16% + 2.6%


and for women over 30, a similar calculation using the formula

28% + 0.1% x (your age – 30)

So a 60 year old woman would generate

28% + 0.1% x (60-30)

= 28% + 3%


Again, just an interesting number to aim for, and one which should give you a good idea if you are on track or not. The numbers produced by this formula would put you comfortably at the top end of the ‘normal” body fat range.

Some people will loathe weighing themselves every day. For others it will feel perfectly normal. Do whatever suits you, but if you only want to weigh yourself once a week, you must do it on the same day, given the cyclical nature of the “Life of Riley” habits.

I think Friday mornings are probably the best time, after you have completed a four day cycle of sticking to the plan. Try and save the data – both weight and fat % – either in a diary or, preferably, in a spreadsheet if you have a pc to hand.