UKC

Interesting paper here: https://bmjopensem.bmj.com/content/5/1/e000437

No associations were found between level of performance in sport climbing, onset of a climbing-related injury, preferred style of climbing, education, gender and BMI

Seems like the BMI in their sample has a pretty tight range in the middle of healthy/normal.  Also it is a bit of a weird sample demographically with about 1/3 of them having PhDs.   The say it is a web survey which would mean people reporting their own BMI which leads to questions about the accuracy of the measurement.

They've got one 9a climber in the sample with a much lower BMI than the people climbing lower grades and they rule out that data point because it isn't significant since there's only one person.

They also don't split between bouldering and sport enough:  low BMI is probably more of an advantage in sport because of the endurance aspect.

Hmmm, not sure I agree with the no link - to me it said that there was no impact to climbing performance if you were within the healthy range of BMI. Although I imagine there's a lack of BMI 35 individuals climbing 8c to look at.

It would have been interesting to see what the BMI variations were like between less experienced climbers climbing say 7b and experienced climbers at the same grade - it might've given a view of where technique/tactics could fit in (Or it might have told us nothing useful! )

I have not bothered to read the link but as a 65 year old who is probably 2 .5 stones over my previous fighting weight that it sounds a load of balls.

Edit probably 1.5 stones.

Another possible conclusion (?):

No association was found between data supplied by self-selected web respondents and actual reality.

Actually (and more seriously) the main flaw here is that the data fails to show a correlation between BMI and the other factors. It does NOT show that there is positively NO correlation - Absence of evidence is not evidence of absence.

The problem with using BMI as indicator is that it doesn't differentiate between weight attributed to muscle or that to fat and it's entirely possible to have a well muscled physique and a relatively high BMI.
A better test would have been to use percentage body fat.  Bet they would have seen a difference then...

Pretty hard to climb well if you can't see your feet I would say 8-)

Yes, this is the precise issue with BMI as some people see it.  My likely "correct" healthy weight is still in the "overweight" range because I'm very heavily built.  If only I could actually reach that weight

The point of BMI really is that anywhere in the range can be healthy depending on you as an individual.

I put a stone and a half on over the past 3 months or so due to an injury stopping me running and comfort eating (yes, I know).  Fortunately I am now running again.  I can still climb but it's certainly knocked me down a grade or two and made my stamina worse (one overhanging route and I need a long rest!)

See also

https://latticetraining.com/2017/10/04/bmi-and-climbing/

These data show a very slight negative correlation between BMI and climbing, with most participants sitting in the "healthy" BMI range.

Pointless study by so called educated people. You'd need a sample group of a few hundred, then they'd need to train the same, climb the same routes etc.. for several years. Then you could/n't possibly see a pattern. But as said bmi as a measure of fitness and body fat level is useless. 

Excellent. Let me get back on my fried chicken diet and finally send that 8a.

Unfair, I think. This is several hundred participants, and fairly cheap to conduct. In the real world, you can't control for everything, like ensuring every climber has an identical climbing history. So what else can you do? Standard practice would be to try and measure everything that could be a confounder as well as possible and attempt to account (adjust) for these factors in the analysis. Quite hard to do. 

If John Dunne was in the sample maybe that skewed all the results.

A good quality study may be hard to do but that does not mean that it fine to publish unjustifiable conclusions based on poor quality data. That just makes a mockery of science. 

My BMI is 33. A couple of  years ago it was 29. There is no doubt, based on  a sample of 1, that this difference has a huge effect.

Maybe minor variations  in the normal  range don't make a difference.

exactly. That's why you can't draw any conclusions from their data. 

Find something else to research? 

Design some simple strength test, duration of dead hang, pull ups that can be done by anyone at any wall, compare strength to grade. Or balance; how many 1 leg squats can you do. Etc. 

Impossible. Climbing is too varied. 

A high bmi person could have good balance and push their way up routes that require balance..  tip toeing up etive slabs etc.. a gym queen could be thugging their way up short overhanging grit routes of the same grade.

Who is the better climber etc.. in reality because of muscle mass their bmi could be same. 

As i said. The study in its present form is worthless. Yes that's hard, but the person who designed must have known that at the outset. 

I don't think the conclusions are a mile off. I would perhaps word them differently ("no evidence for...") but there is a limitations section which clearly presents the appropriate caveats. It's published in an appropriate journal and associated impact (BMJOpen: reasonable peer review, low impact).

Does it add evidence to the research area? Yes.  Is it definitive? No.

Can it be used as part of a future meta-analysis? Yes, definitely. Should it not be published because it's not definitive? I don't think so.

What do you think makes climbing so unique, compared to, say, investigations into the effect of diet on health outcomes?

I invite you both to submit a comment on the paper in the most appropriate place:

https://bmjopensem.bmj.com/content/5/1/e000437.responses

I was going to surggest putting climbers in fat suits to see the differences in achievable grade from increasing BMI.  But this too is pointless because firstly it is pretty obvious what wil happen and secondly it only accounts for adding dead weigh not muscle which may have advantages. It really just highlights the uselessness of BMI in sport as it can't separate muscle from fat.

The reality is there is more than one body type that'll do well in climbing. The low muscle, super low weight model or the mid weight, higher muscle, low fat version. Both preform well, one leans towards indurance the other power.

In either case excess fat doesn't help as it comes down to power to weigh ratios. 

Thanks for that link Jon, it will be interesting to read the responses.  Obviously with more funding better studies could be undertaken, but I thought it was reasonable approach and quite interesting.

Ditto two stone.  I know I climb harder when I'm lighter.  I've just worked out my BMI and it's 29.

And I read the article.  There is very little variation - as far as I can see they're all within the normal range.  I would be interested to see what happens when you get into other categories, especially overweight and obese.  I would argue that the small sample range in terms of BMI doesn't give them a great deal of statistical power.  I'm trying to find the standard deviation or inter-quartile ranges for BMI in the UK so I can compare it to this study.

You could flip this on its head and say that those who climb regularly enough to consider themselves climbers to the extent that they'll complete a survey on it are probably reasonably fit and will therefore be within the normal range of BMI.  If they wanted to get more of a range, and control for climbing experience, they could've invited people coming into climbing walls for the first time to participate, asking them to record their BMI and then see how hard they can climb to start with.  I'm substantially overweight at the moment but when my shoulder wasn't bothering me I was projecting 6b on lead.  I've been climbing for most of my life and so have a better sense of footwork and movement than a complete novice climber.  I'm currently teaching two girls from my lab who are 3 stone lighter than me and jumped straight on the wall for the first time at 6a+.  

As for the high proportion of PhDs, the lead investigator clearly sent the link to all their climbing friends.  There's a huge crossover between scientists and climbers in my experience.  This is not surprising, but does in some way explain the lack of variation.  The study was circulated by word-of-mouth amongst a particular social circle.

Why the dislike , John?

You were always proud of your weight.

"I am two stone heavier than you lot, and still climb better! Wait till the diet kicks in."

I've just read the paper and I'm with you Rick.  It reads as if the author knows everything about stats and nothing about climbing.

Look at the references.  They're all medical papers.  Did the author not read any books about climbing, or talk to climbers?

As you say, it ignores the effect of BMI on climbers behaviour.  When we get lardy, we HAVE to drop our grade... Bleedin obvious.

As others say, i suspect there is a flaw in the initial questionnaire. Based on personal experience, injuries occur when I go to the wall more often because that is when I'm pushing my grade (i.e. increasing the loading my body) and I'm getting less recovery between sessions.  I think the questionnaire is faulty because of the author's pre-conceptions about what they were hoping to discover about BMI and how injuries occur.

It's a shame really - I'm currently nursing a rotator cuff injury and am attempting to work out if it's at least exacerbated by making it haul every ounce of 12* st up the wall too many times!  As someone who has struggled with my weight for most of my adult life, I would've liked the impetus of "if you don't lose it, you'll just keep injuring your shoulder" to make me get my sh*t together.  

*recent weight gain caused by enforced lack of training and associated comfort-eating, but my point mostly still stands.

The references are probably all medical because it's a medical journal. Not unreasonable - citation is a politicised practice, and what/who you cite has an effect on chances of publication. It's a limited study, and doesn't set out to use qualitative measures like interviewing climbers. Again, not unreasonable - you can't do everything. A more extended consideration of future research could have been helpful, including suggesting some qualitative investigation.  

Nah - just campus everything  (including slabs (you might need knee pads though)).

According to this 

https://ukdataservice.ac.uk/media/604458/mgreen.pdf

Median BMI in UK 2013 was 27 (men) and 26 (women) and standard deviation 4.5 (men) and 5.5 (women).   Overall I'm somewhat conflicted on this paper - from a quick read I think there's maybe some interesting data there but its not sufficient to justify the conclusions, particularly the level of certainty applied.   What we seem to have is a rather homogeneous self selecting group who are on average have lower BMI than the population and less variation.    And there appears to be some data to conclude that given a group of climbers with BMI profile like this their is not significant relationship between lower BMI and better climbing.  

I think the possible conclusion is quite interesting (and seems to match what Lattice found) - it seems to indicate that given a healthy BMI, having lower BMI does not necessarily correlate to improved climbing.  Understand that BMI is a very crude measure, particularly when muscle increases but this still runs counter to lot of peoples natural intuition.  There can be a tendency sometimes to argue to support intuition rather than to understand why that intuition may not be completely correct (see any number of discussions as to why climbing is easier for tall people even though there no evidence that this is true).

I'm not sure I can agree. I may be being pedantic (but good science is pedantic), but there are two fundamental problems here:

1. The research is presented as being about climbers in general rather than about this specific set. Therefore you have to consider the sampling methodology, i.e., to what extent is the sample set representative of the whole? What is the sampling error? With such a self-selected web based sample you have zero information about your sampling error - in other words you cannot (with any rigour) extrapolate an analysis of the sample to the whole.

2. Even with the restriction imposed above (that the analysis is relevant only to the sample set, not the whole), the conclusions drawn are not justified by the data, i.e., stating that there is no correlation between BMI and the other factors is plain wrong. All the author can really say is that the data does not support the premise that there is some correlation. This is not just playing with words - in science those two alternative conclusions are fundamentally different.

In scientific analysis, there is room for uncertainty (to reference your "mile off" point) - in fact uncertainty is mandatory. But this uncertainly is not derived from sloppiness (or deviation from expectation), it is derived from understanding the statistical significance of various error sources and ensuring that they are properly accounted for in the analysis. Yes, there is a section about limitations, but it doesn't cover the impact of indeterminate sampling error and certainly not on the subject of unjustifiable conclusions.

Unfortunately today there is pressure in scientific publishing to always 'prove' something. You rarely see papers which say "we wanted to see if X is true, but the data collected did not support such a conclusion". Papers like that rarely get published (or even submitted) and risk being deemed to have been a waste of time and money (and bad for the authors' CV). It's a big pity because it's important to understand the difference between (a) X probably not being true and (b) a poorly designed experiment. Publishing would have provided some differentiation.

I think it does because people misunderstand BMI, thinking that everyone's perfect weight is the precise middle of the range, and variations are just "vaguely acceptable".  In reality, someone can be at their ideal weight *anywhere* in the range, it depends very much on the makeup of the individual's body.  Some people might even be at their ideal weight outside it - I suspect this is the case for me being quite heavily built - while I'm not a rugby player I'm built like one, and I'd imagine a good many rugby players are at their ideal weight but would be "overweight" (but probably not "obese") by BMI alone.

What is of more relevance is what makes up the extra weight.  If it's a beer belly, no, that of course is no help for climbing whatsoever (though if you look at something like distance running it can be a little more complex than that, as someone with near-zero body fat is going to have to pay a *lot* more attention to nutrition along the way than someone with a bit - but then equally being actually fat is not helpful).  If it's pure upper-body muscle, it might well be helpful.

This is the sort of thing BioRxiv and similar are really useful for.  Negative data may not be the most riveting but still needs (and deserves to be) published.

For a start you'd need to measure grade performance on the same routes, we could all name two routes of the same grade that require vastly different physical attributes. 

Yes. You can research it, but there are many variables you need to control. 

Edit. 5a. Long hot summer on Rosa slabs and The Sloth... for example. 

I don't think anyone suggested it is unique. There have been some shockingly bad studies into effect of diet on health outcomes, especially when using self reporting.

Treatment of confounding variables e.g. with the effect of sugar, calorie intake, physical activity levels other lifestyle factors (which are all confounding each other) has often been pathetic. This leads to  the ridiculous, eternal conveyor belt of contradictory health pronouncements and some completely over the top and baseless assertions that damage the reputation of the scientific community.

If you do a multivariate analysis on even a relatively simple manufacturing process with a few key process variables your models can still break down and become meaningless if you aren't very careful (this doesn't stop them being used to present conclusions). If people think they can account for wide ranging influences in  studies relying on self reporting where nothing was controlled they are lying to themselves (and everyone else).

I agree with you. The negative views expressed here regarding this fairly modest contribution to the literature seem to suggest an extremely high bar of evidence required for publishing anything. This I dispute because, practically, nothing would get researched. No study is perfect. This study was unfunded, and made a reasonable attempt (imho) at trying to measure important variables, and with a cross-sectional design has inherent evidence limitations. To do it to the (admirable) level discussed here would cost millions of pounds. Who would *ever* fund that? Does that make a study that presents some information  where previously none existed "worthless"?

Edit for my ribbish tyoing.

I didn't say it should be published - I think negative results definitely should be published more often. My 'negativity' relates to the poor analysis and misrepresentation of the results.

In this particular case, yes. Its very small positive value in the data collected is outweighed (IMO) by the significantly incorrect conclusion. If the author had made an attempt to determine the sampling error and had then concluded that there was nothing statistically significant in the available data, I wouldn't have got caught up in this.

"It was cheap" can be an excuse for inconclusive results, but not for bad science.

Ok, fair enough

I remember my climbing got suddenly easier when I lost approx a stone and a half before it got linked to my un-diagnosed diabetes.

Edit: I reckon that there's possibly a fairly large 'grey area' as it were, though, where technique counts for more than a change in weight, with weight making more difference at the higher edge of it. 

PhDs were 115 out of 667 (1 in 5.8)

My personal experience of weight variation within normalise range is that each 3kg is worth a (sports) grade , if you include + as a separate grade so 10kg is difference between 6c and 7a+ with everything else being equal 

Not read the whole article, but I'm not surprised.

Climbing is about "you" against your limits, it doesn't matter if your heavy or not, or if your good or shit, it's about how hard you push you body, a fit skinny person can push their body over it's limits just as easy as a fat, unfit person.

That's why climbing is such a leveller, if you're climbing at your limit, it's the same for a VS leader as an E6 leader, there's not much difference in effort, perception or fear.

Dear all
Nice to see that the study is being debated.
there are quite a few themes covered in this thread, I´ll try to respond to them all. If I´ve missed out some, sorry about that!
 

Tom

University 5 years or more + phd = 1/3

Regular education level in Norway is 10% have university more than 5 years 1% have a phd

But, it´s a well known problem with self reported questinnaires that the well educated are more prone to reply than others. Plus that the more educated you are, the more likely you are to do sports at a high level. Mixed together this makes the sample a little less off the charts than what it seems in a first glance. This is not mentioned in the paper because it is well known an goes for all self reported questionnaires presented to the public.

The survey was open for all and was promoted by the national climbing federation and the national climbing magazine on their web sites.

On the narrow sample of BMI

It was, of course a selected population regarding BMI. If the study was aiming to assess main population BMI and achievements in sports it would have been a horrible study with this narrow sample of BMI, but as it is this is a study on climbers (or at least those who deem themselves to be climbers and visit climbing sites and responds to a survey about climbing…)

But, as it is stated in the limitations chapter, it is to few in the low level category! Although I doubt the conclusion would be different if there were more of the recreational climbers responding.

Why BMI?
Other meassures of body composition would, of course, have been better. BUT, as it is said elsewhere here, that would be expensive and time consuming. Still, the main problem would probably not be the costs of it, but the further skewing of the respondents. AS it is now the number of respondents is OK (could have been bigger). With all respondents have to show up for strength tests, measuring of body fat, registration of diet and so on the number of respondents would plummet and all would be citicens of one or two major cities instead of national.
BMI used on a population of a few hundred is OK, used for smaller saple sizes or for one person it is of no use.
Anyway, a quick rise in weight will undoubtly have an impact on performance! A slow rise and a steady training routine might not have the same impact? This study suggests that it may not have a great impact. For my own sake (if that mathers at all) I´ve differd in BMI from 21.5-23,9 and been doing the same level of climbing (8b/8b+ route 7C+/8A boulder). As this is a cross sectional study changes in BMI and performance over time is not assessable. I would love do such a study sometime though!

Dave cundy
As I´ve been a climbing instructor for about 25 years and authored a pile of papers for the climbing magazine and a book on climbing injuries (english version out this fall probably), running a facebookpage on climbinginjuries (https://www.facebook.com/klatreskader/ Most of it Norwegian so far) and have about a 100 first ascents up to 8c routes and 8A boulders I do think I know a thing or two about climbing (probably a lot more than I do on statistics).

As previous studies have suggested an association between BMI and injuries the results were quite the opposite of what was expected. Thing is that all the other studies have mixed up acute injuries and chronic injuries while this study only uses chronic injuries. Still, they´re all self reported so it might not be that they even have an injury, only some pain…

Bob kemp
Sitations in scientific papers is wise and usefull, sitations of “the right” people to be published would be either stupid or catastrophic!
And yes, bigger and better studies are needed.

Marek

I totally agree on the part of the selection bias! I tried to cover it in the «limitaions» section, but could probably have covered more there. Sorry if you feel it is not enough in there as it is.
I do, of course, not agree on the conclusion being unjustified, it might be written differently, but as it is there are no associations between performance or injuries and BMI in the analyses.
As I´m not a statistician and only relies on the help I´ve received from an epidemiologist and one statistician on these analyses it would be nice to know what statstical methods you would deem to be better for this analyses.

Cheers

Flawed methodology. 

The study simply says you cannot predict some factors based purely on BMI. 

A better long term study would be to take some climbers and alter their BMI over a period of time as there are simply too many variables that counteract the effect of BMI. 

Unless that’s what they’re saying. BMI has no effect because other factors are more pertinent. 

I am assuming they are trying to prove that changing your BMI won’t have an effect on your performance.

<SNIP>

...

Hi Gudmundg,

It is brilliant that you've found the time and motivation to reply here! Thanks.

As to my point...

Your conclusion states: "BMI is not associated with climbing-related chronic injury or level of performance in climbing" which is not quite the same as what you wrote above. Your conclusion positively states that there is no association (which is unjustified) whereas you statement above states that no association was found in the analysis (hence leaving open the significant question of whether the data was statistically meaningful). In the context of a scientific paper, those two forms of words are significantly different in their meaning. I not saying there's anything wrong with the mathematical analysis, my point is only about the semantic leap from the result (OK) to the conclusion (not OK). It is an issue of logic (and scientific rigour), not statistics.