Highway Code revisions: helmet discussion

For all discussions about this "lively" subject. All topics that are substantially about helmet usage will be moved here.
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: helmet discussion

Post by Jdsk »

Steady rider wrote: 7 Dec 2021, 3:52pm3 of the 28 head injury studies in the 2017 review, Olivier J, Creighton P. Bicycle injuries and helmet use: A systematic review and meta-analysis. International Journal of Epidemiology 2017;46:278-292
reported a negative outcome
https://academic.oup.com/ije/article/46/1/278/2617198

Yes. it's a systematic review. How many reported a positive outcome?

Thanks

Jonathan
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: Consultation complete and Government response published

Post by Jdsk »

Vorpal wrote: 7 Dec 2021, 2:23pm
Zulu Eleven wrote: 6 Dec 2021, 4:56pm
Pete Owens wrote: 6 Dec 2021, 4:49pm But those authors have a long notorious history of pushing absurdly fanciful claims on the effectiveness of helmets, based on studies with extremely small and systematically biased samples. Their helmet advocacy long preceded their study.
Yet, magically, still no valid challenge to the Cochrane review based upon either methodological failings or contrary results. It’s almost as if the anti-helmet brigade have so far entirely failed to garner a valid challenge based on the flaws they claim exist, or present it through the established and robust comment and editorial process
https://www.cyclehelmets.org/1068.html provides a summary of the criticisms of one of the Thompson, Rivara & Thompson paper included in Cochrane review.

https://www.cyclehelmets.org/1243.html discusses the Cochrane review itself.

Adams and Hillman wrote this published dissent of the Cochrane review https://injuryprevention.bmj.com/content/7/2/89

There is also a meta-analysis by Elvik that discusses the Cochrane review https://toi.brage.unit.no/toi-xmlui/bit ... sequence=2

Plenty more where these came from.
Thankyou for a rational approach to studying a published systematic review. There's far too much special pleading in this thread and this is a pleasant contrast.

Thompson et al was 1999. Eivik was 2011, and was able to include some more studies. Olivier and Creighton was 2017 and has even more.
https://academic.oup.com/ije/article/46/1/278/2617198

On the SR front we should use the most recent unless there's a special reason not to. Limitations of methodology might of course persist.

(This post is on the question of injuries and not on the question of mandation.)

Jonathan
mattheus
Posts: 5030
Joined: 29 Dec 2008, 12:57pm
Location: Western Europe

Re: Highway Code revisions: helmet discussion

Post by mattheus »

There is another interesting statement in JD's source:
https://www.bmj.com/content/346/bmj.f2674
A third possible explanation for our results is that the effectiveness of helmets is greater for mild and moderate head injuries than for the severe head injuries captured by hospital admission data.
This fits my expirences perfectly. I had a nasty looking gash on my forehead when I went to Accident&E after being SMIDSYed. It is pretty hard to dispute that a helmet would have prevented that injury! (Whilst it wouldn't have saved my dental damage, nor the ankle and shoulder injjuries, but anyway ... ) As a kid, I had a few lesser blood-gushing incidents - no trip to hospitla required, luckily.

But statistics don't show an improvement in brain injuries - and these are where the focus of hospital treatment will be. I wasn't even concussed, so they didn't keep me in after initial triage.

I'm not worried about a few bumps, cuts and scrapes.On my head, or my knees, elbows, hips etc etc ... I've had worse driving my car! If I thought a lid would make my brain *significantly* safer, I might well be persuaded to wear one.
Steady rider
Posts: 2749
Joined: 4 Jan 2009, 4:31pm

Re: Highway Code revisions: helmet discussion

Post by Steady rider »

https://www.researchgate.net/publicatio ... le_helmets

All the included studies were based on injury comparisons, where behaviour could have been very important regarding injuries. The method employed appears similar to the Seattle study that reportedly had a 400% overestimation for helmets.

In addition there is the issue of accident involvement, where helmet wearers have a higher rate of falls, that give a lower proportion of serious head injuries.
https://www.researchgate.net/publicatio ... jury_rates

The evidence does not appear to be sufficiently sound to tell people they 'should' wear a helmet.
User avatar
mjr
Posts: 20297
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Highway Code revisions: Consultation complete and Government response published

Post by mjr »

Jdsk wrote: 7 Dec 2021, 12:43pm But this pales into insignificance when compared with quoting your own work without declaration. Or cherrypicking your own work when other studies and systematic reviews are available.
Would those systematic reviews be the ones whose authors cherrypicked their own work? I think my irony meter just broke.
MJR, mostly pedalling 3-speed roadsters. KL+West Norfolk BUG incl social easy rides http://www.klwnbug.co.uk
All the above is CC-By-SA and no other implied copyright license to Cycle magazine.
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: Consultation complete and Government response published

Post by Jdsk »

mjr wrote: 7 Dec 2021, 4:30pm
Jdsk wrote: 7 Dec 2021, 12:43pm But this pales into insignificance when compared with quoting your own work without declaration. Or cherrypicking your own work when other studies and systematic reviews are available.
Would those systematic reviews be the ones whose authors cherrypicked their own work? I think my irony meter just broke.
Including your own work with that of others is the opposite of cherrypicking. Cherrypicking is about deliberate exclusion.

But I don't know if any primary studies by Olivier or Creighton made it into the 40 that were in the meta-analysis. Anyone, please?

Jonathan
User avatar
mjr
Posts: 20297
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Highway Code revisions: Consultation complete and Government response published

Post by mjr »

Jdsk wrote: 7 Dec 2021, 4:34pm
mjr wrote: 7 Dec 2021, 4:30pm
Jdsk wrote: 7 Dec 2021, 12:43pm But this pales into insignificance when compared with quoting your own work without declaration. Or cherrypicking your own work when other studies and systematic reviews are available.
Would those systematic reviews be the ones whose authors cherrypicked their own work? I think my irony meter just broke.
Including your own work with that of others is the opposite of cherrypicking. Cherrypicking is about deliberate exclusion.
I'd bet the review authors picked search terms that included all their past work in the field. Did either of the two you mentioned exclude any of their own studies?
MJR, mostly pedalling 3-speed roadsters. KL+West Norfolk BUG incl social easy rides http://www.klwnbug.co.uk
All the above is CC-By-SA and no other implied copyright license to Cycle magazine.
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: Consultation complete and Government response published

Post by Jdsk »

mjr wrote: 7 Dec 2021, 4:37pm
Jdsk wrote: 7 Dec 2021, 4:34pm
mjr wrote: 7 Dec 2021, 4:30pm
Would those systematic reviews be the ones whose authors cherrypicked their own work? I think my irony meter just broke.
Including your own work with that of others is the opposite of cherrypicking. Cherrypicking is about deliberate exclusion.
I'd bet the review authors picked search terms that included all their past work in the field. Did either of the two you mentioned exclude any of their own studies?
I don't think that authors of reviews should exclude their own primary studies. They should make it absolutely clear if there are any, and provide enough data to make it as easy as possible for others to rerun the analysis with and without those studies.

And by the time you get to 40 included studies and publish the analysis it's usually pretty obvious if too much weight is being given to any study, whether it's an outlier or not.

Jonathan
Steady rider
Posts: 2749
Joined: 4 Jan 2009, 4:31pm

Re: Highway Code revisions: helmet discussion

Post by Steady rider »

The 400% figure was quoted from Zeegers 2015 work. The method used in the 40 studies seems to be the same as the Seattle study, so how can the 40 studies be considered reliable?

Data from New Zealand covers both injuries and time cycling per year. It shows a major increase in the accident rate per million hours cycled from 23 to 63 ('other' - mainly falls is assumed). More than 90% wear helmets in NZ. The data can be checked as it is available on line.
drossall
Posts: 6106
Joined: 5 Jan 2007, 10:01pm
Location: North Hertfordshire

Re: Highway Code revisions: helmet discussion

Post by drossall »

Can I say that I've tried hard all my life to follow reports of research and to make decisions on my own helmet wearing that are based on evidence? I've got a science degree, so some ability to follow papers, but I'm only a physicist :D But I'm finding it hard to take anything that I can use from this discussion. It's difficult to see a summary view. There are plainly two sides here, but they aren't clearly engaging with each other or going anywhere. So I'm going nowhere.

As things stand, I've been reluctant to wear a helmet. I've been very aware of the deficiencies of Thompson and Rivara, and therefore surprised that they were invited to prepare the Cochrane Review, and even more so that the review so much emphasised their own work and derivatives. There have definitely been some studies that were more impressive, especially hospital studies that used other injuries to control for the kind of error that T&R evidently made, and that found positive results. But strongly positive results are difficult to accept, precisely because there doesn't seem to be much corresponding real-world population-level evidence, except where the effect was probably to reduce levels of exposure. So, in a way, the more positive the results, the less likely they are to be true.

From the physics point of view, I'm really aware that most of the massive difference between falling off at 12mph and being hit by a heavy car at even 20mph, which on the face of it makes it nearly impossible for a device designed for the former consistently to make a substantial difference to the latter. Again, therefore, the bigger the claims, the less likely they are.

I've just looked at Olivier et al cited here but, in a world where so many studies are criticised on methodological grounds, I didn't see an effort to establish what were reliable methodologies and to filter accordingly. Nor is there any obvious accounting for later critiques. It feels a bit like quantity over quality (without of course having read any of the primary research, so not really an attempt to make what would anyway be unqualified comment!)

I'm impressed by Goldacre (who appears to be more truly disinterested, and yet better qualified, a commentator than anyone), as well as what Chris Boardman says about making cycling look dangerous - and that almost any other measure would have more effect. So I've tended to follow Cycling UK's statement as a broad position. But it has not escaped me, for example, that that refers to neck injuries, which Olivier et al describe as rare, which would if true mean that at least some update were needed to that statement.

I'm very much in agreement with mattheus here, as well:
mattheus wrote: 7 Dec 2021, 4:14pmI'm not worried about a few bumps, cuts and scrapes.On my head, or my knees, elbows, hips etc etc ... I've had worse driving my car! If I thought a lid would make my brain *significantly* safer, I might well be persuaded to wear one.
The critical point here being that it is perfectly possible for helmets to reduce scalp injuries and yet increase brain ones so, whilst I'm not asserting that to be the case, it's pretty important to be clear what we're measuring! So any study worth including in a review is going to be explicit about that.

So where from here? Do we invite both sides to state cases in clearer ways? Is there any way to reach a new assessment? Should I be rethinking? (More convincing answers please than "Of course (not)!"
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: helmet discussion

Post by Jdsk »

drossall wrote: 7 Dec 2021, 6:59pmIs there any way to reach a new assessment? Should I be rethinking? (More convincing answers please than "Of course (not)!"
Yes. Quick thoughts on how to make progress:

1 Agree ground rules for the discussion. These are always there somewhere in scientific method, sometimes explicit and sometimes not. In this case it's basically to use evidence-based methodology. (NB this is not a given in many areas of science because it just isn't needed, often because you can do experiments and remove confounders.)

2 Separate the different questions. For example: do helmets affect injury if an impact occurs? Does wearing a helmet affect behaviour and risk? What are the effects of mandated wearing of helmets?

3 Accept the current limitations of knowledge. Discuss how they might be addressed.

How does that sit with Goldacre & Spiegelhalter? They'd simply agree. I recommend reading that paper again through these three lenses.
https://www.badscience.net/2013/12/bicy ... demiology/

Jonathan
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: helmet discussion

Post by Jdsk »

drossall wrote: 7 Dec 2021, 6:59pmI've just looked at Olivier et al cited here but, in a world where so many studies are criticised on methodological grounds, I didn't see an effort to establish what were reliable methodologies and to filter accordingly. Nor is there any obvious accounting for later critiques. It feels a bit like quantity over quality (without of course having read any of the primary research, so not really an attempt to make what would anyway be unqualified comment!)
They have a whole section on "Limitations". As should any report using evidence-based methodology.

It might not be obvious but their systematic review tries to follow some painfully evolved guidelines. That's an important part of evidence-based methodology. This won't overcome limitations of design in the primary studies but it should help to identify them.

But when there aren't primary studies with the highest levels of evidence in the evidence hierarchy (as is the case here) that leaves a massive problem for the separate issue of policy. The least that can be done is to clearly and repeatedly identify the limitations of current knowledge.

Jonathan
drossall
Posts: 6106
Joined: 5 Jan 2007, 10:01pm
Location: North Hertfordshire

Re: Highway Code revisions: helmet discussion

Post by drossall »

Yes, I think those are different questions. In my recollection, though, the sequence was that doubts appeared first when research started to show negative results. I think the first I saw quoted was in 1986 in the Journal of Product Liability, but that could be memory playing tricks. Risk compensation then came in as an explanation for the negative results, not as a reason to gainsay positive ones. So I find the focus on arguing that positive results would be counterbalanced by risk compensation difficult.
Jdsk
Posts: 24478
Joined: 5 Mar 2019, 5:42pm

Re: Highway Code revisions: helmet discussion

Post by Jdsk »

drossall wrote: 7 Dec 2021, 7:21pm Yes, I think those are different questions. In my recollection, though, the sequence was that doubts appeared first when research started to show negative results. I think the first I saw quoted was in 1986 in the Journal of Product Liability, but that could be memory playing tricks. Risk compensation then came in as an explanation for the negative results, not as a reason to gainsay positive ones. So I find the focus on arguing that positive results would be counterbalanced by risk compensation difficult.
I'd have given a completely different answer if asked to explain the history or the current nature of the debate!

I was only trying to suggest the least worst way forward...

: - )

Jonathan
drossall
Posts: 6106
Joined: 5 Jan 2007, 10:01pm
Location: North Hertfordshire

Re: Highway Code revisions: helmet discussion

Post by drossall »

:oops: You see, that's me complicating things now. Not sure I got myself nearer to an updated view by that, sorry. :oops:
Post Reply