Searching for statistics and studies

This sub-forum all discussions about this "lively" subject. All topics that are substantially about helmets will be moved here, if not placed here correctly in the first place.
User avatar
mjr
Posts: 11009
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Searching for statistics and studies

Postby mjr » 29 Sep 2017, 11:09am

Steady rider wrote:in 2013 police reported cycle accidents data shows 51% not wearing helmets, these including all accidents.

What's the source for that, please? In particular, is that really all collisions, or only all injury accidents?
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.

Steady rider
Posts: 2100
Joined: 4 Jan 2009, 4:31pm

Re: Searching for statistics and studies

Postby Steady rider » 29 Sep 2017, 12:02pm

DfT source, police reported accidents, 30 police forces out of 44.

AlaninWales
Posts: 1525
Joined: 26 Oct 2012, 1:47pm

Re: Searching for statistics and studies

Postby AlaninWales » 29 Sep 2017, 12:09pm

RobC wrote:This is a recent report using data from St Marys Hospital. The conclusion is "In a largely urban environment, the use of cycle helmets appears to be protective for certain types of serious intra and extracranial head injuries. This may help to inform future helmet design."

I've read it as a lay person - neither a doctor or a scientist - and it seems to be a compelling argument. I'm sure many of us agree that a helmet does provide a certain degree of protection for certain kinds of injury.

I don't actually wear a helmet - I don't believe the kind of cycling I do is any riskier than any other non-helmeted activity - but given that this study's conclusions are rather pro-helmet and will likely be used by the pro-helmet lobby as a result, I'm interested to know how to counter such arguments faced with research like this.

http://journals.plos.org/plosone/articl ... ne.0185367

How to counter it?

1. The study suffers from selection bias: It selects only cyclists ("A retrospective observational study of all cyclists older than 16 years admitted to a London Major Trauma Centre between 1st January 2011 and 31st December 2015 was completed.") and studies the head injuries they present, finds that some injuries could have been reduced or prevented by wearing a cycle helmet (yup, I might not have grazed my head [but may have instead suffered from a rotational injury to the neck or brain]) and decides this is a reason to recommend cycle helmets. To avoid selection bias, it should have studied all head injuries presenting and looked whether the same intervention would have reduced injury to the same (or more? Less?) extent with non-cyclists (e.g. motorist and pedestrian helmets).

2. As hinted above, it makes does not consider whether the intervention suggested (cycle helmets) has any mechanism of causing or worsening injury.

3. Unproven assumptions that where head injury occurs, this is the cause of death. Generally speaking, getting crushed by a motor vehicle causes pretty massive trauma elsewhere in addition to any head injury. The study states as (unreferenced - i.e. unproven) backround that "Head injuries are an important cause of mortality and morbidity in cyclists", examined all (and only) cyclists presenting to a Major Trauma Unit and then ignored the impact of the rest of the major trauma that they are likely to have suffered.

All in all, an extremely biased and unscientific approach. That it was accepted for publication, simply demonstrates that the assumption mentioned (and I repeat, not referenced) in the background is common amongst at least, the "peer reviewers" at PLOS.


Edit: To (miss) quote CS Lewis: "I wonder what they do teach them in these (medical) schools", it certainly isn't 'How to design a statistical study to establish the efficacy of a proposed intervention". Perhaps it is "How to design a statistical study to appear to support your pre-determined conclusions".

User avatar
mjr
Posts: 11009
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Searching for statistics and studies

Postby mjr » 29 Sep 2017, 1:48pm

Steady rider wrote:DfT source, police reported accidents, 30 police forces out of 44.

Please could you name the DfT source? The helmet field isn't in the published RRCGB tables that I've got and there isn't one called "police reported accidents". I'm not disbelieving you - I'd just really like to explore that source.

By the way, 51% of those in accidents helmet-wearing would be disproportionate (relative to 30something% of cyclists using helmets) and I suspect significantly so (although I've not done the maths to check).
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.

RobC
Posts: 87
Joined: 5 Feb 2008, 3:27pm

Re: Searching for statistics and studies

Postby RobC » 29 Sep 2017, 2:38pm

Thanks Alan, that's useful!

Steady rider
Posts: 2100
Joined: 4 Jan 2009, 4:31pm

Re: Searching for statistics and studies

Postby Steady rider » 29 Sep 2017, 7:14pm

I emailed the DfT, if you send a PM with your email address I will forward the data.

Steady rider
Posts: 2100
Joined: 4 Jan 2009, 4:31pm

Re: Searching for statistics and studies

Postby Steady rider » 29 Sep 2017, 7:38pm

Early days to assess the London paper, they say the results are similar to the NY paper. It looks like some difference in behaviour leading to the accident may be one aspect, 13 of the non wearers taking alcohol and 2 wearers, but no data on levels of alcohol, if I have got the figures correct see 'supporting information'.
^60% of non-wearers had max fax fractures v 40% for wearers, Spaite 1991 paper reported non-wearers had more severe impacts. It may take some time to assess the paper in more detail.

The study relates to an area with 3 million people and 129 serious head injuries over a 5 year period, 29 per year, one per 100,000 people per year roughly. I would have hoped that the wearing details could have been provided for the non-head injuries, 194 major trauma cases. This may have allowed for a comparison between the two groups without head or helmet considerations.

drossall
Posts: 4241
Joined: 5 Jan 2007, 10:01pm
Location: North Hertfordshire

Re: Searching for statistics and studies

Postby drossall » 29 Sep 2017, 8:51pm

RobC wrote:This is a recent report using data from St Marys Hospital. The conclusion is "In a largely urban environment, the use of cycle helmets appears to be protective for certain types of serious intra and extracranial head injuries. This may help to inform future helmet design."

I've read it as a lay person - neither a doctor or a scientist - and it seems to be a compelling argument. I'm sure many of us agree that a helmet does provide a certain degree of protection for certain kinds of injury.

I don't actually wear a helmet - I don't believe the kind of cycling I do is any riskier than any other non-helmeted activity - but given that this study's conclusions are rather pro-helmet and will likely be used by the pro-helmet lobby as a result, I'm interested to know how to counter such arguments faced with research like this.

http://journals.plos.org/plosone/articl ... ne.0185367

Surely the classic responses are that:
  • This is looking only at cyclists who arrived at hospital. Therefore, if helmet wearing (or, for that matter, non-wearing), were to prevent injuries in any instances, those would be excluded.
  • We have no idea whether the helmetted and unhelmetted cyclists had similar accidents. This was the error in Thompson and Rivara, which was later reanalysed to indicate that helmets were 75% effective at preventing limb injuries (because the accidents were not the same).
  • The article links helmets to riding in traffic and with lorries, which are so massively outside the design parameters of helmets as to be, well, I'm almost writing laughable, but it's not funny.
The numbers are quite large. I'd always understood that a single hospital doesn't tend to get statistically significant numbers of cycle casualties.

Steady rider
Posts: 2100
Joined: 4 Jan 2009, 4:31pm

Re: Searching for statistics and studies

Postby Steady rider » 30 Sep 2017, 12:10pm

Having briefly looked at the data, perhaps the benefits from helmet use may not be as great as it first appears.

Looking over some of the data, looks like for drinking alcohol, 13 non-wearers v 2 wearers, similar to other reports in proportions, 18.5% v 7.4%. Cyclists not wearing helmets appear to be similar to pedestrians in some ways.

For the severe GCS with a score of 8 or less, where helmet use is known, 11 cases, 6 of the 11 affected by alcohol, of the remaining 5 cases, 2 helmeted v 3 not helmeted, in proportion 2/27 = 7.4%, 3/70 = 4.2%. Excluding cyclists affected by alcohol and where helmet use was known, it appears that helmeted at a higher proportion of severe head injuries, 7.4% v 4.2%.

data needs checking etc.

The actual numbers are quite low and this should be a consideration.

Their data shows 19 cyclists were aged between 16 - 25 years, 2 wore helmets, 4 unknowns, 13 not worn.
This age group has been reported to incur about a 50% higher risk per hour cycled, see http://www.cycle-helmets.com/NZ-Cyclist ... uly-07.pdf similar porportions to the alcohol, 18.5% v 7.4%.
Three of the cyclists in this age group and not wearing helmets had also had been drinking.

Wanlock Dod
Posts: 292
Joined: 28 Sep 2016, 5:48pm

Re: Searching for statistics and studies

Postby Wanlock Dod » 30 Sep 2017, 10:26pm

It's not just cycling, I'm not sure that I found the paper by Thornhill, but I did find a response to one of their papers.