Helmet design flaw may put cyclists at risk

[BeSeenOnABike.com]

BICYCLE helmet standards have a design flaw that could leave cyclists
vulnerable to serious head injuries, say researchers in Belgium. The
current standards fail to protect one of the most vulnerable parts of
the human head - the temple. Yet the researchers have shown that the
temple is a common impact site.

They studied head injuries in 86 cyclists who had been involved in
accidents. They found that 57 per cent of them had suffered impacts to
the side of the head, and a further 27 per cent had suffered impacts to
the front.

Helmets do protect against some of these injuries, but most current
helmet designs leave the temple unprotected.

http://www.newscientist.com/article.ns?id=mg17924070.400

Chris Street
www.BeSeenOnABike.com

 

[Tony Raven]

Tony Raven wrote:
>> Its a 2003 NS article and the paper, for those with Science Direct
>> access is at http://dx.doi.org/10.1016/S0001-4575(03)00062-9
>>
>
> An interesting excerpt:
>
> "The wearing of a bicycle helmet was documented in only three victims.
> All three had a combination of severe injuries: skull fractures,
> subdural haematoma, subarachnoid haemorrhage, contusions and brain
> swelling in all three, an intracerebral haematoma in one and an
> extradural haematoma in one. All three patients died. Their ages were
> 50, 65 and 71. In one patient, a bystander reported that the helmet had
> shifted backwards at the moment of the impact on the forehead."
>


Must stop replying to my own posts but having searched the paper I've
finally found the sentence that for 83 individuals studied "Mortality
was 33.3%". Which means statistically you would have expected two of
the three helmeted cyclists to have survived whereas all three died. So
what's the probability that it was chance that all three rather than one
died?

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham

 

[Tim Woodall]

On Sat, 07 Jan 2006 11:20:01 +0000,
Tony Raven <[email protected]> wrote:
> Tony Raven wrote:
>
> Must stop replying to my own posts but having searched the paper I've
> finally found the sentence that for 83 individuals studied "Mortality
> was 33.3%". Which means statistically you would have expected two of
> the three helmeted cyclists to have survived whereas all three died. So
> what's the probability that it was chance that all three rather than one
> died?
>
Hope I've got this right

none died - 8/27
one died - 4/9
two died - 2/9
three died - 1/27

(at least it adds up)

Tim.

--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t,"
and there was light.

http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/

 

[Simon Brooke]

in message <[email protected]>, Tony Raven
('[email protected]') wrote:

> which reminds me of
>
> "Only four riders died of fatal injury to head alone and one of
> these was the only rider known to be wearing a safety helmet."
> Sage MD. 1985. NZ Med J: 25
> Dec 1985 Vol 98 No 793

<whistling mode="innocently">
Would someone out there happen to have access to the full text of this
article?
</whistling>

--
[email protected] (Simon Brooke) http://www.jasmine.org.uk/~simon/

;; All in all you're just another hick in the mall
-- Drink C'lloid

 

[Tony Raven]

Simon Brooke wrote:
> in message <[email protected]>, Tony Raven
> ('[email protected]') wrote:
>
>> which reminds me of
>>
>> "Only four riders died of fatal injury to head alone and one of
>> these was the only rider known to be wearing a safety helmet."
>> Sage MD. 1985. NZ Med J: 25
>> Dec 1985 Vol 98 No 793
>
> <whistling mode="innocently">
> Would someone out there happen to have access to the full text of this
> article?
> </whistling>
>

I don't but I know zis Guy who does ;-)

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham

 

[Tony Raven]

Tim Woodall wrote:
>
> Hope I've got this right
>
> none died - 8/27
> one died - 4/9
> two died - 2/9
> three died - 1/27
>
> (at least it adds up)
>

Thanks, its been a while since I've done those calcs and I would have
had to spend some time refreshing my memory to get it right. Pretty
interesting result is it not? There is a 96% probability that all the
helmet wearers dying was not chance.

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham

 

[Tim Woodall]

On Sat, 07 Jan 2006 14:53:42 +0000,
Tony Raven <[email protected]> wrote:
> Tim Woodall wrote:
>>
>> Hope I've got this right
>>
>> none died - 8/27
>> one died - 4/9
>> two died - 2/9
>> three died - 1/27
>>
>> (at least it adds up)
>>
>
> Thanks, its been a while since I've done those calcs and I would have
> had to spend some time refreshing my memory to get it right. Pretty
> interesting result is it not? There is a 96% probability that all the
> helmet wearers dying was not chance.
>
I'm not sure what sort of a conclusion you can draw from this (you need
a statistician Certainly, if the only time helmet wearing was
recorded was when the victim was obviously dead to police at the scene
and wearing a helmet then you would expect 100% fatality rate.

What I found most interesting was the fact that these were then only
recorded cases of helmet wearing. Give the number of helmet saved my
life anecdotes etc, I would have expected to see the results skewed but
in the opposite direction to my above paragraph - i.e. I would expect
the police to record "Wearing a helmet" when the helmet wearer was
obviously alive at the scene (may die later) and "Not wearing a helmet"
where the helmet wearer was obviously dead at the scene. The remaining
two classes I would have expected to be "Helmet wearing not recorded".

This could still be true if very few cyclists are obviously dead at the
scene (regardless of helmet wearing) but usually die hours to days later
in hospital. Additionally, the obviously dead at the scene cyclists
could well tend to be the HGV casualties. I don't suppose anybody would
think a helmet would protect anyone in the (all too common) left turn
crush type accidents and so again, the presence or absence of a helmet
may not be recorded as it's obviously irrelevant.

Tim.

--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t,"
and there was light.

http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/

 

[Tony Raven]

Tim Woodall wrote:

>
> I'm not sure what sort of a conclusion you can draw from this

I draw the conclusion that you are making it up as you go along.


> Certainly, if the only time helmet wearing was
> recorded was when the victim was obviously dead to police at the scene
> and wearing a helmet then you would expect 100% fatality rate.
>
> What I found most interesting was the fact that these were then only
> recorded cases of helmet wearing. Give the number of helmet saved my
> life anecdotes etc, I would have expected to see the results skewed but
> in the opposite direction to my above paragraph - i.e. I would expect
> the police to record "Wearing a helmet" when the helmet wearer was
> obviously alive at the scene (may die later) and "Not wearing a helmet"
> where the helmet wearer was obviously dead at the scene. The remaining
> two classes I would have expected to be "Helmet wearing not recorded".
>
> This could still be true if very few cyclists are obviously dead at the
> scene (regardless of helmet wearing) but usually die hours to days later
> in hospital. Additionally, the obviously dead at the scene cyclists
> could well tend to be the HGV casualties. I don't suppose anybody would
> think a helmet would protect anyone in the (all too common) left turn
> crush type accidents and so again, the presence or absence of a helmet
> may not be recorded as it's obviously irrelevant.



Perhaps you should read the paper first.

It was not just police records of helmet wearing and it was not the
obviously dead at the scene we are talking about. These are patients
who were admitted to hospital for neurological intervention. You do not
intervene for people that die at the scene or in the ambulance.

Helmet wearing was documented from police records and questionnaires
sent to all patients or their relatives. It was documented that three
were definitely wearing helmets, 56 were definitely not wearing a helmet
and the remaining 27 are unknown because the questionnaire was not returned.

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham

 

[Simon Brooke]

in message <[email protected]>, Tim
Woodall ('[email protected]') wrote:

> What I found most interesting was the fact that these were then only
> recorded cases of helmet wearing. Give the number of helmet saved my
> life anecdotes etc, I would have expected to see the results skewed but
> in the opposite direction to my above paragraph - i.e. I would expect
> the police to record "Wearing a helmet" when the helmet wearer was
> obviously alive at the scene (may die later) and "Not wearing a helmet"
> where the helmet wearer was obviously dead at the scene.

It was in Belgium, where cycling is more common, and, I believe, helmet
wearing is only common amongst racing cyclists.

--
[email protected] (Simon Brooke) http://www.jasmine.org.uk/~simon/
Windows 95:
You, you, you! You make a grown man cry...
M. Jagger/K. Richards

 

[Tim Woodall]

On Sat, 07 Jan 2006 17:37:30 +0000,
Tony Raven <[email protected]> wrote:
> Tim Woodall wrote:
>
>>
>> I'm not sure what sort of a conclusion you can draw from this
>
> I draw the conclusion that you are making it up as you go along.
>
> Perhaps you should read the paper first.
>
Of course I am. I don't have science direct access and so, no I haven't
read the paper. All I had is the statistics you posted to which I
calculated the probabilities.

You then posted that there was a 96% probability that this was not
chance. There was no way anybody could draw that from my figures because
the information used to calculate them was insufficient for drawing any
conclusions other than the probability of getting N from 3 where the
probability of each success is 1/3

> It was not just police records of helmet wearing and it was not the
> obviously dead at the scene we are talking about. These are patients
> who were admitted to hospital for neurological intervention. You do not
> intervene for people that die at the scene or in the ambulance.
>
> Helmet wearing was documented from police records and questionnaires
> sent to all patients or their relatives. It was documented that three
> were definitely wearing helmets, 56 were definitely not wearing a helmet
> and the remaining 27 are unknown because the questionnaire was not returned.
>
Even with this you still can't draw any conclusions. _I_ think this is
significant but 27 is too close to 1/3 for any confidence. _I_ would
guess that non-responses are more likely for fatalities, either because
it isn't known by relatives whether a helmet was worn or the relatives
are too distressed to reply. The relatives of dead helmetted cyclists
may be more likely to respond because they are not thinking "what if
s/he was wearing a helmet".

Of those 27, there needed to be 6 surviving helmet wearers for this to
give the expected results assuming no benefit/loss for helmet wearing.
This would give a helmet wearing rate of about 10%

Tim.

--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t,"
and there was light.

http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/

 

[Mark Thompson]

> I don't but I know zis Guy who does ;-)

His website says he's just got the abstract, and the paper dates from 1985
(they started putting things up on the web for free (with 6 months delay)
in January 2002.

On the bright side, I've now got a pdf of "Bicycle-related head injury" as
uni have forgotten to cancel my athens login dooberry

 

[Tony Raven]

Tim Woodall wrote:
>
> Of course I am. I don't have science direct access and so, no I haven't
> read the paper. All I had is the statistics you posted to which I
> calculated the probabilities.
>

I'm not talking about the calculation you had done - I was talking about
your comments on the methodology of the study

>>
> Even with this you still can't draw any conclusions. _I_ think this is
> significant but 27 is too close to 1/3 for any confidence. _I_ would
> guess that non-responses are more likely for fatalities, either because
> it isn't known by relatives whether a helmet was worn or the relatives
> are too distressed to reply. The relatives of dead helmetted cyclists
> may be more likely to respond because they are not thinking "what if
> s/he was wearing a helmet".

It shouldn't matter what the unknowns were. If one third of the whole
population dies, then you would expect one third of a sub population to
have died all things being equal, not 100%. You actually require
relatives of dead helmeted cyclists to be less likely to respond to
support your argument by the way.

>
> Of those 27, there needed to be 6 surviving helmet wearers for this to
> give the expected results assuming no benefit/loss for helmet wearing.
> This would give a helmet wearing rate of about 10%
>

But there is no reason to believe that the population distribution of
the unknowns is significantly different from the knows. Of the 27 you
would expect 3/59ths or about 1 or 2 to be helmet wearers. Looking at
the helmet wearing percentages of surrounding countries - 0.1%
Netherlands, 2% Germany, 2.4% France, 3% Denmark you would only expect
two or three helmeted cyclists in the whole sample of 86. In fact based
on those figures, unless all the unknowns were unhelmeted and the three
were the only ones helmeted, helmet wearers would seem to be
disproportionately represented in this cohort. This would suggest that
helmet wearers were not only more likely to have a head injury accident
requiring neurological intervention, they would also be much more likely
to die as a consequence of the injury.

--
Tony

"Man is a credulous animal, and must believe something; in the absence
of good grounds for belief, he will be satisfied with bad ones."
- Bertrand Russell

 

[Tim Woodall]

On Sat, 07 Jan 2006 21:58:47 +0000,
Tony Raven <[email protected]> wrote:
>
> I'm not talking about the calculation you had done - I was talking about
> your comments on the methodology of the study
>

Maybe it's my writing but I hadn't thought I had made any comments about
the methodology of the study.

As we are obviously reading completely different meaning into the same
posts I'll just stop here.


Tim.

--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t,"
and there was light.

http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/

 

[James Annan]

Tony Raven wrote:

> Tim Woodall wrote:
>
>>
>> Hope I've got this right
>>
>> none died - 8/27
>> one died - 4/9
>> two died - 2/9
>> three died - 1/27
>>
>> (at least it adds up)
>>
>
> Thanks, its been a while since I've done those calcs and I would have
> had to spend some time refreshing my memory to get it right. Pretty
> interesting result is it not? There is a 96% probability that all the
> helmet wearers dying was not chance.
>

Rubbish. You simply can't draw that sort of inference from those numbers.

James
--
James Annan
see web pages for email
http://www.ne.jp/asahi/julesandjames/home/
http://julesandjames.blogspot.com/

 

[Simon Brooke]

in message <[email protected]>, James Annan
('[email protected]') wrote:

> Tony Raven wrote:
>
>> Tim Woodall wrote:
>>
>>> Hope I've got this right
>>>
>>> none died - 8/27
>>> one died - 4/9
>>> two died - 2/9
>>> three died - 1/27
>>>
>>> (at least it adds up)
>>
>> Thanks, its been a while since I've done those calcs and I would have
>> had to spend some time refreshing my memory to get it right. Pretty
>> interesting result is it not? There is a 96% probability that all the
>> helmet wearers dying was not chance.
>
> Rubbish. You simply can't draw that sort of inference from those
> numbers.

Agree. Population way too small. This is noise, not signal. There's
enough dodgy use of statistics on the other side of the argument without
us indulging in it as well.

--
[email protected] (Simon Brooke) http://www.jasmine.org.uk/~simon/
;;Drivers in the UK kill more people every single year than
;; Al Qaeda have ever killed in any single year.

 

[Tony Raven]

Simon Brooke wrote:
>
> Agree. Population way too small. This is noise, not signal. There's
> enough dodgy use of statistics on the other side of the argument without
> us indulging in it as well.
>

So given a population of 86 in which 33% died, what is the probability
that in a group of 3 people, all three dying is down to statistical
fluctuation? Yes the numbers are small and complicated by quantisation
(you can't have part person deaths) but noise is just another name for
statistical fluctuations.

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham

 

[James Annan]

Simon Brooke wrote:
> in message <[email protected]>, James Annan
> ('[email protected]') wrote:
>
>
>>Tony Raven wrote:
>>
>>
>>>Tim Woodall wrote:
>>>
>>>
>>>>Hope I've got this right
>>>>
>>>>none died - 8/27
>>>>one died - 4/9
>>>>two died - 2/9
>>>>three died - 1/27
>>>>
>>>>(at least it adds up)
>>>
>>>Thanks, its been a while since I've done those calcs and I would have
>>>had to spend some time refreshing my memory to get it right. Pretty
>>>interesting result is it not? There is a 96% probability that all the
>>>helmet wearers dying was not chance.
>>
>>Rubbish. You simply can't draw that sort of inference from those
>>numbers.
>
>
> Agree. Population way too small. This is noise, not signal. There's
> enough dodgy use of statistics on the other side of the argument without
> us indulging in it as well.
>

It's worse than that - there just isn't any way to draw that type of
inference out of those sort of numbers. Certainly I'd agree the numbers
do nothing to demonstrate that helmets are helpful.

A related question that may help to illustrate the point:

A certain medical test gives a negative result for 90% of healthy
patients (positive for the other 10%), and a positive test for 90% of
ill patients (neg for the other 10%).

Someone takes the test, and the result is positive. What is the
probability that he suffers from the disease?

I've heard that doctors frequently get this wrong, BTW.

James
--
James Annan
see web pages for email
http://www.ne.jp/asahi/julesandjames/home/
http://julesandjames.blogspot.com/

 

[David Martin]

James Annan wrote:
> It's worse than that - there just isn't any way to draw that type of
> inference out of those sort of numbers. Certainly I'd agree the numbers
> do nothing to demonstrate that helmets are helpful.
>
> A related question that may help to illustrate the point:
>
> A certain medical test gives a negative result for 90% of healthy
> patients (positive for the other 10%), and a positive test for 90% of
> ill patients (neg for the other 10%).
>
> Someone takes the test, and the result is positive. What is the
> probability that he suffers from the disease?
>
> I've heard that doctors frequently get this wrong, BTW.

The words 'Bayes rule' flit through my brain and I can't quite remember
it. Something like (checks google)

p(x|A) = p(A)*p(A|x)/p(x)

Which is the probability of condition x given evidence A is the
probability of getting evidence A multiplied by the prior probability
of having A given x divided by the distribution of x.

In which case not knowing the distribution of illness in the population
it is impossible to say (using this method). Were you missing a part of
the question or was it deliberately set as one where not all the data
is available?

...d



>
> James
> --
> James Annan
> see web pages for email
> http://www.ne.jp/asahi/julesandjames/home/
> http://julesandjames.blogspot.com/

 

[Richard]

James Annan wrote:
> A certain medical test gives a negative result for 90% of healthy
> patients (positive for the other 10%), and a positive test for 90% of
> ill patients (neg for the other 10%).
>
> Someone takes the test, and the result is positive. What is the
> probability that he suffers from the disease?

You can't tell based on those numbers. You also need to know the
proportion of the population who actually have the disease in question.

> I've heard that doctors frequently get this wrong, BTW.

Well, I'm a doctor, and...oh, you mean /that/ sort. I have bad
handwriting, does that count?

R.

 

[James Annan]

Richard wrote:

> James Annan wrote:
>
>> A certain medical test gives a negative result for 90% of healthy
>> patients (positive for the other 10%), and a positive test for 90% of
>> ill patients (neg for the other 10%).
>>
>> Someone takes the test, and the result is positive. What is the
>> probability that he suffers from the disease?
>
>
> You can't tell based on those numbers. You also need to know the
> proportion of the population who actually have the disease in question.

Exactly my point.


James
--
James Annan
see web pages for email
http://www.ne.jp/asahi/julesandjames/home/
http://julesandjames.blogspot.com/