(no subject)

[ajva]

Oh fuck *off*.

Everybody knows BMI is a shit measure anyway. Why does an entire profession, scientifically trained, ignore the evidence of the scientific method? Knowing that BMI is shit because it measures only weight with respect to height rather than anything else like bodyfat or muscle, still they persevere with it because they can't be bothered to do anything else.

Are they going to turn away athletes like Martin Johnson or Matthew Pinsent?

Cunts. Morons. I despair.

Comments

[ciphergoth.livejournal.com]

There's seriously no correlation between weight and adiposity for a given height? That is astonishing and shocking. I'd like to read more about that if you can point me in the right direction - thanks!

[ajva.livejournal.com]

If you want me to dig up some links then I'll get onto it, but surely it's well known that muscle is more dense than fat, that it's heavier per unit volume?

So surely from there just a moment's reflection will tell you that two similarly-boned large people who look exactly the same size, but have different body composition, will be different weights, the more muscular one being heavier, but the more "adipose" one being unhealthier, though lighter? Would you not accept that this would screw up the correlation you're suggesting?

The word "unhealthier" in the above, of course, makes the assumption that more muscle and less fat is a healthier state to be in, which may well itself not have been proven - but either way I think it does away with the weight/fat correlation theory.

[ciphergoth.livejournal.com]

I'll concede right away what you seem to be arguing for, which is that the correlation won't be 100% because there are other factors that affect weight than adiposity. But I can't believe the correlation is statistically insignificant either, because adiposity does affect weight, and I can't see why the other factors would exactly compensate. I've known people whose weight and adiposity visibly changed in tandem, in both directions. It would take some really amazing evidence to convince me that the correlation is zero.

[ajva.livejournal.com]

I find it really difficult to know how to respond to this. It seems as if there is no possible way to proceed to try to win you over on this issue. :o)

Body composition is not routinely measured, so it is difficult to get hard statistics that I could use to further my argument. You say this:

But I can't believe the correlation is statistically insignificant either, because adiposity does affect weight, and I can't see why the other factors would exactly compensate.

But where is your evidence for other factors not "compensating", as you put it? It is based on your own anecdotal judgement:

I've known people whose weight and adiposity visibly changed in tandem, in both directions.

If there be, as you instinctively believe, some sort of positive correlation between adiposity and weight, then surely it is a correlation relating somehow to the absolute levels of fat tissue, rather than the proportional. This is all you could realistically deduce by sight, surely? Do you have X-ray Fat Vision? But the health aspect depends on the proportionality, which is in no way measured by weight, and therefore is in no way measured by BMI.

There are plenty of slim people who show up as normal on BMI measures, but have a high proportion of bodyfat; these are truly "fat", in health terms*, even if the dress/trouser size says otherwise.



* warning: Assuming a useful correlation between bodyfat percentage and healthfulness that, as I said above, may not be proven.

[ciphergoth.livejournal.com]

It's not routinely measured, but it is sometimes measured. I googled and found this:

http://www.unu.edu/unupress/food2/UID10E/uid10e05.htm

which, interestingly, suggests that the correlation is good enough to be useful for Europeans, but less useful for non-Europeans.

[ergotia.livejournal.com]

You are wasting your time with P on this one.

[djm4]

But where is your evidence for other factors not "compensating", as you put it? It is based on your own anecdotal judgement:

I can't speak for ciphergoth, but I can't think of a single other factor that would compensate in the way you'd need it to to make the correlation zero. For that, you'd need a factor that tended to make people with a higher adiposity lighter, and I don't know of one.

Just to be clear here, I'm talking about a correlation between adiposity and weight as the only two measured variables about a person. I'm not suggesting that that correlation is medically useful, nor am I suggesting that it's a 100% correlation. It won't be, because there are many other factors affecting weight. However, none of those other factors (as far as I know) produces a negative correlation between adiposity and weight, so all they do is increase the scatter about the points on the adiposity/weight graph. Which lowers the correlation, certainly, but not to zero.

But the health aspect depends on the proportionality...

Well, yes. That's what I thought you were talking about when you said that the correlation wasn't a useful measure. I agreed with that, but you seem to be saying something different here; you seem to be saying that the correlation isn't statistically significant, which is something that I doubt very much.

Note, though, that 'the correlation between adiposity and weight (or BMI for that matter) is statistically significant' is a purely mathematical judgement as far as I'm concerned. It leaves totally open the question of whether adiposity should be considered for referrals/treatment. It also leaves open the question of - given that we're talking about beople being denied access to medical help - it's justifiable to use a measure that's less than 100% correlated with the thing that we're trying to measure.

There is, I'm pretty sure, a statistically significant correlation between white-skinned people and those entitled the treatment at an NHS hospital (if you take to population of the world as a whole, I'm much less sure about the population of the UK). I would still be horrified if a doctor were to start refusing to refer someone to a hostpital based purely on the colour of their skin.

[djm4]

There is, I'm pretty sure, a statistically significant correlation between white-skinned people and those entitled the treatment at an NHS hospital (if you take to population of the world as a whole, I'm much less sure about the population of the UK). I would still be horrified if a doctor were to start refusing to refer someone to a hostpital based purely on the colour of their skin.

Footnote - that's purely supposed to be an example if where I think it would be wrong to use even a statistically significant correlation. It wasn't meant to suggest - and I wouldn't suggest - that it's in the same league as anything anyone's suggested here.

[ciphergoth.livejournal.com]

and it's true, as ergotia indicates, that she and I have discussed this before without coming to 100% agreement.

[adjectivegail]

I can start you off - The Obesity Myth by Paul Campos. He belabours the point somewhat, but he does quote lots and lots of scientific studies, and cites them at the back of the book so you can go look them up yourself if you don't believe him.

The gist of it:
- There is no correlation between obesity and morbidity, unless you take into consideration other factors such as smoking, diabetes, cancer, etc.
- In fact, a healthy, active person with a BMI over 25 has a lower risk of mortality than an equivalent person whose BMI is under 25.
- Fat people who try to get thin in order to avoid certain health problems (e.g. hypertension, heart problems, diabetes, etc.) are more likely to wind up with exactly these health problems than either a fat person who takes up regular exercise (e.g. 30 minutes walking a few times a week) or even than a fat person who doesn't change their lifestyle.
- Because a lot of people believe that if they're thinner they'll be healthier, they try to be thinner by restricting what they eat. In fact very few people manage to succeed at this, and those who do actually reduce their life expectancy. And those who yo-yo diet make it even worse. In fact, mortality rates were most reduced in subjects who took up regular, moderate exercise - and hardly any of them got any thinner.

And it goes on and on and on.

I really hope these Trusts get their asses sued off. Even if I do realise that the likelihood of that happening is close to nil. Grrr.

[ciphergoth.livejournal.com]

Um. I know that if I read the book I'd be frustratingly half-convinced - I know it's not my field, and I'm sure he presents what seems like a very convincing case, but I just bet that if I met someone equally knowledgable and determined to present the opposite case they'd convince me the other way. I really don't know how to go about figuring out the truth on this one.

However, I am pretty convinced that there's a correlation between weight and adiposity, whether or not reducing either is good for you.

[djm4]

That's not what ajva said. She said no useful correlation existed (my italics). I think, if you're going to dispute this, you should at least have a working definition of what you mean by 'useful'. Your original comment suggests that you think BMI can be used in a useful way (well, OK, you said 'sane', but it seems to be in the same area) to decide whether or not to grant someone treatment on the NHS.

That's what I think ajva, asrana (and I, for that matter) take issue with. Not that a correlation between adiposity and weight exists (as you point out, it doesn't have to be 100% for it to be there) but that it's at all useful in deciding who's a priority for treatment.

[ciphergoth.livejournal.com]

It's stupid to use it as a gate for treatment because, quite apart from all the other problems with such a gate, they have more accurate measures at their disposal.

I think I'm arguing that BMI is still a useful proxy for adiposity, and the best at our disposal since we don't have the equipment for better measures of adiposity at home. Some references suggest you classify your frame into one of three types, and use that when interpreting BMI; I'd guess that makes it a better proxy. It still has problems, but I'd need some convincing that it's too misleading to be useful.

Of course, that still leaves the question of whether losing weight/fat is good for your health, which is one I don't know the answer to.

Aside from the 'is BMI a useful proxy' issue:

[adjectivemarcus.livejournal.com]

It's not being used as a gate for treatment.
It's being used as a gate for referrals.

It's the PCTs (Primary Health Care Trusts, ie local doctors surgeries) saying they won't refer BMI30+ patients to hospital for such treatment, irrespective of the hospitals' referral criteria.

The hospital would judge if the patient was in a physical condition that made an operation too risky, decide where on the list to place them etc based on their own clinical findings.

[djm4]

Of course, that still leaves the question of whether losing weight/fat is good for your health, which is one I don't know the answer to.

I think that's because, as a general question applied to the population at large, it makes about as much sense as asking whether wearing a particular strength of spectacles would improve people's eyesight.

[ciphergoth.livejournal.com]

Yeah, obviously losing weight wouldn't be good for my health, for example. I'm just trying to reference in shorthand a whole bunch of questions that are part of the wider debate about what recommendations should be made with regards weight, adiposity, diet and exercise.

In longhand I guess it looks like: is there really a significant subset of the population who would enjoy better health if they reduced their calorie intake, and if so what measure most accurately indicates if you're one of them? Of those who exercise little or not at all, would some benefit from exercise far more than others and if so which ones? And that's still not a totally precise way of putting it.