Index Home About Blog
Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Albert Green MD MPH) 
Date: Jul 28 1995
Newsgroups: rec.bicycles.misc

Not this again.

Bastian Wimmer (image04@pop03.ny.us.ibm.net) wrote:

: and fancies. Fact 1: In the United States, over 900 
: cyclists are fatally injured every year. Fact 2: 
: Almost 80% of these die of head injuries. Fact 3: 
: Three fourths of these deaths could be avoided with 
: bicycling helmets. End of argument. Let's debate 
: these facts, not each other.

#3 isn't a fact.  There are lots of badly done and biased articles in 
various publications which make various claims to this effect, or its 
opposite.  There is ALSO good quality, well-done medical research in 
reputable, peer-reviewed journals.  Several methods have been used, and 
all converge at about the same numbers in *reputable* scientific work (not 
advocacy publications!).

Bottom line: helmets probably can prevent about 1/3 of deaths and 2/3 of
serious head injuries (bad enough to require neurosurgery or to leave
permanent disability such as seizures or movement disorders). 

We have no idea what fraction of lesser injuries they might prevent,
because no reasonable denominator can be established for exposure.

Baseline risk for death by head injury among serious cyclists is 
somewhere in the one to two per thousand range.  Helmets make a 
significant reduction in a risk which is not very great to begin with; 
i.e., a large relative risk reduction but a small absolute risk reduction.

Do I wear one?  Yes.  It's cheap, easy to do, and even a low risk of 
ending up sitting in a nursing home drooling on myself is something I'd 
prefer to reduce.  Do the data justify haranguing those who choose 
otherwise?  Not IMO.  So chill, already, before I end up having to treat 
the entire newsgroup for ulcers!! ;-)

--
Lee Green MD MPH
Dept Family Practice
University of Michigan
greenla@umich.edu


Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Albert Green MD MPH) 
Date: Jul 31 1995
Newsgroups: rec.bicycles.misc

Thomas H. Kunich (tomk@netcom.com) wrote:
: In article <3v9akp$kbe@lastactionhero.rs.itd.umich.edu>,
: Lee Albert Green MD MPH <greenla@umich.edu> wrote:

: >Bottom line: helmets probably can prevent about 1/3 of deaths and 2/3 of
: >serious head injuries (bad enough to require neurosurgery or to leave
: >permanent disability such as seizures or movement disorders). 

: Studies always seem to misss the reality, don't they? Helmet laws
: for motorcyclists and bicyclists have succeeded in reducing fatalities
: through the same avenue -- decreased ridership.

I'm sorry, but this is simply false.  The studies in question were 
case-controls.  That is, the answer they returned is likelihood of injury 
PER HOUR EXPOSED, not per population which may or may not ride bicycles.

Whether mandatory helmet laws reduce ridership is another subject.  My 
suspicion is that they probably do, but mostly by reducing the number of 
the least skilled and least committed riders.  Of course this group is 
the most likely to have accidents.  But this is still a different 
question from whether and how much helmet use reduces an *individual's* 
risk of injury.  The data demonstrate that they do, to the modest extent 
previously posted.

At the risk of being redundant (but because so many seem to flame before 
reading), PLEASE read my post carefully.  I did not, do not, and will not 
advocate helmet laws.  I am commenting on the data about protection 
afforded to *individuals* who choose to wear helmets, NOT about helmet 
laws.  Helmet laws are a value judgment; helmet effectiveness is data.  
Some who evaluate that data conclude that the modest but definite risk 
reduction justifies requiring people to wear them, others feel that that 
is too great an intrusion on personal liberty for such a modest risk 
reduction and consequently limited impact on the public.  That value 
judgment is strongly made one way or the other by many people, who tend 
to fall prey to the temptation to fudge the data to make their case 
stronger.  Please do not do so.


Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Albert Green MD MPH) 
Date: Aug 03 1995
Newsgroups: rec.bicycles.misc

Thomas H. Kunich (tomk@netcom.com) wrote:
: In article <3vh9c3$9m1@lastactionhero.rs.itd.umich.edu>,
: Lee Albert Green MD MPH <greenla@umich.edu> wrote:

: >I'm sorry, but this is simply false.  The studies in question were 
: >case-controls.  That is, the answer they returned is likelihood of injury 
: >PER HOUR EXPOSED, not per population which may or may not ride bicycles.

: Dr. Green, we seem to have this argument every couple of months. How do
: these studies estimate riders, mileage and TIME EXPOSED in a reliable
: manner? Everything I've read shows from pretty crude 'guestimations'.

The studies use several methods.  There are case-controls, in which 
riders are matched on various characteristics and injury severities 
compared.  These studies only see the people who get hurt of course, 
which is why the only reliable data is on severe injuries and deaths, not 
on minor injuries.  Cohort studies are what you refer to, which use what 
you term "guestimations".  Your term is pejorative, however, and shows no 
understanding of the sophistication of this sort of research.  The 
researchers are NOT making wild-ass guesses about exposure; they use 
multiple methods and check their reliability against one another quite 
carefully.

The most telling point though is that the studies replicate reliably, and 
well-done studies of all methods used so far in reputable peer-reviewed 
journals all converge on the same result.  That's the benchmark of 
reliability in science.

: Even worse, I've heard one statistition say that a _single_ event was
: data enough to extrapolate!

"Worse"?  It may seem so if you've not studied statistics, but there are 
indeed situations in which that statement is true.  Very few of them, but 
they exist.  However, that's completely irrelevant here.  The studies 
in question are not based on "one" or even a few injuries.  They're based 
on hundreds or thousands of injuries.

: >Whether mandatory helmet laws reduce ridership is another subject.  My 
: >suspicion is that they probably do, but mostly by reducing the number of 
: >the least skilled and least committed riders.

: The very fact that you recognize that should also clue you in to the
: further fact that a _lot_ of riders could stop riding without there
: being significant changes in the statistics.

Untrue, and again a misunderstanding of statistics.  Whether most or a 
minority of riders in the population wear helmets makes *no difference 
whatever* in the RELATIVE risk.  An individual's odds of death with 
helmet are 2/3 of those without, of serious brain injury 1/3 of without.  
The ABSOLUTE risk depends on how many people get into accidents where 
they might be injured, i.e. the prevalence.  If you have a population of 
klutzes the absolute risk might be 6/1000 per year, vs. a typical 
population of club cyclists which will be under 1/1000 per year.  How 
many people helmets can save is a function of relative risk reduction and 
prevalence.  Helmets prevent 1/3 of deaths, hence in the population of 
klutzes they will prevent 2 deaths per 1000 per year if everyone wears 
them, or 1 per 1000 per year if half the people wear them.

: Child use of bicycles in California has dropped so drastically that
: it is noticeable to almost any bike rider. Moreover, articles quoting
: bogus studies have appeared in the newpaper touting great reductions
: in head injuries due to the helmet laws for children.

I'm sure they have.  But 1) helmet laws are a value judgment, and I'm 
discussing helmet effectiveness not what you choose to do based on that 
data (laws or no), and 2) newspapers are not my idea of the peer-reviewed 
responsible medical literature.

: Now add to this that most of the "low riders" (that catagory of custom
: bicycle builders that haven't been chased off -- yet) don't wear helmets
: and that any time now I expect to see some study saying that 97% or all
: children with head injuries on bicycles were riding sans helmets.

But that is exactly the kind of study which would appear in the advocacy 
press, NOT in the peer-reviewed literature.  It doesn't tell me anything 
to know what fraction of injured didn't wear helmets.  It tells me 
something to compare the degree of head injuries among helmeted and 
unhelmeted riders who crash.  If you don't understand this crucial 
denominator issue, and why one study is bogus and the other reliable, 
Streiner et al.'s "PDQ Epidemiology" is a very readable introduction to 
the subject.

: Believe me, I understand that as a physician you should be concerned about
: the physical well being of your patients. I only argue this with you
: because I also think that it is up to the rest of us to strike the
: balance with reality.

The "rest of us"?  I did not surrender my hold on reality when I took up 
the practice of medicine, Mr. Kunich.  I don't tell my patients what to 
do, I tell them the facts (helmets are modestly effective, more than the 
strident anti's want to admit and less than the strident pro's want to 
claim), and let them make their own decision.

I would suggest that if you really want to strike a balance with reality, 
you begin with some basic statistics and epidemiology, and with a clearer 
distinction between the data and the value judgments people make based on 
what those data mean to them.  "2 per thousand per year" is data.  "That's 
too many, we should have a law" vs. "that's not many, keep your laws off 
my head" is value judgment.


Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Albert Green MD MPH) 
Date: Aug 05 1995
Newsgroups: rec.bicycles.misc

Distribution: 

R.D. Auchterlounie (rda@eng.cam.ac.uk) wrote:

: >compared.  These studies only see the people who get hurt of course, 
:                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
: In which case they cannot possibly measure helmet efficacy.

: They provide data on the benefits of helmets 
: _in a serious injury accident_ and that is _all_ they show.
: They cannot possibly show reduction in risk per hour of cycling
: because they haven't measured it.

One more time: the studies in question are case-controls and cohort 
studies.  Yes, the case-controls provide data on the benefits of helmets in 
accidents significant enough to bring patients to the ER.  Yes, that is 
all they show and yes, that is all I've claimed they show.  Are some 
people in accidents which would result in a head injury, but no other 
injury, and avoid the head injury because they're wearing helmets, hence 
never get counted as a helmet "save"?  Yes.  How many?  We don't know for 
sure, but survey studies indicate not many in the 
death-and-serious-injury category.  And that is what I've been posting 
about.  Again and again I have emphasized that we do not have data for 
less serious accidents.

: In probability terms you want probability of death (or serious head
: injury) per hour (or km) conditional on wearing and not wearing helmets.
: BUT if you only look at accident victims you have measured those 
: probabilities conditional on having a serious injury accident.
: >helmet are 2/3 of those without, of serious brain injury 1/3 of without.  

No, the conditional probability is what would be ideal to have.  What we
can get is the relative risk, or odds ratio.  It does not tell you what
the conditional probabilities are, but does tell you what the ratio of the
probabilities is.  That can be applied to the prevalence to *estimate* the
absolute risk (or conditional probabilities per hour of exposure). 

: This conclusion cannot be drawn solely from studies of accident 
: victims (as you stated above). The correct statement is:

:   "An individual's odds of death
:    --- following an accident involving serious injury --- with helmet 
:    are 2/3 of those without, of serious brain injury 1/3 of without."

Which, if you will read my previous postings, you will discover is 
exactly what I said.  You've simply managed to rephrase in probabilistic 
terms what is known in epidemiological terms as "relative risk".  Which 
is distinct from "absolute risk", as I've pointed out repeatedly.

Again, however, case-controls from the ER are not the only data we have.  
Comparison studies between areas can estimate the fraction of the local 
population wearing helmets (research assistants standing on streetcorners 
tally samples, same as is done for seat-belt estimates).  Areas of high 
helmet use have lower incidences of serious neurological trauma than 
those of low helmet use, and the numbers work out to about the same as 
from the case-controls.  However, the incidences of broken collarbones 
etc. don't differ between the areas, so the "crash rates" seem to be similar.

The similarity of these results to the ER-based ones suggest what the ER 
docs themselves observe: crashes serious enough to kill or maim due to 
head injury are usually accompanied by other injuries significant enough 
to merit treatment even if the head injury is prevented.  In the probability 
terms you've used, we can indeed estimate the difference between P(head 
injury in serious crash|helmet) and P(head injury in serious crash|no 
helmet), because the P(presentation to ER|helmet in crash) and 
P(presentation to ER|no helmet in crash) appear to differ insignificantly 
for those crashes severe enough to cause the deadly or disabling injuries 
I've specifically limited my comments to.

Again please be clear on what's being stated here: an individual's odds 
of death or serious brain injury if wearing a helmet are 1/3 and 2/3 
respectively of whatever those odds are if s/he is not wearing a helmet.  
That's stated as a relative risk.  You can restate it as conditional 
probabilities if you wish; the literature uses RRs though and I prefer 
them as they're intuitively more understandable to most people (though 
I'm a neo-Bayesian myself).

--
Lee Green MD MPH
Dept Family Practice
University of Michigan
greenla@umich.edu

Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Green MD MPH) 
Date: Aug 29 1995
Newsgroups: rec.bicycles.misc,rec.bicycles.off-road,rec.bicycles.tech

In article <LUSKY.95Aug28133637@sol.hc.ti.com>, lusky@sol.hc.ti.com (Steve
Lusky) wrote:

> Educating people
> how to prevent crashes, including how to react better when in an
> emergency situation, would do more for lowering injury rates of
> bicyclists than just wearing a helmet

No, it won't.  Here's why:

We would ALL like to reduce the rate, not just the severity of injuries
in, crashes.  In a perfect world... Now for the letdown: it doesn't work. 
Sure, schools hold "safety towns" and lots of organizations pamphleteer
and proselytize for safety.  They mean well, and are enthusiastic.

Ask one of them, any one, to show you any results.  Whether it's bike
safety, fire safety, water safety, childhood poisonings, whatever.  They
can't.  They never will be able to.  Attempts at public behavior change
are, in general, depressingly dismal failures.

Sure, if you can get people to join a bike club and become experienced
club cyclists, their accident rate goes way down.  Few do, however.

What works?  Years of safety campaigns made not one bit of difference in
childhood poisoning rates; childproof caps on medications (esp aspirin,
the #1 culprit for two generations), solvents, etc cut childhood poisoning
rates by over 80% in less than three years.  Safe driving campaigns didn't
make any difference in motor vehicle fatalities; seat belt laws and
improvements in auto crashworthiness did.  Fire safety campaigns do
nothing; smoke detectors save lives.

The bottom line is that interventions which don't depend on mass voluntary
behavior change on the public's part work, and "education" doesn't.

The matter then becomes one of a value judgment.  Since you can't design a
more crashproof bicycle, you have to crashproof the rider.  The only way
we have to do that right now is helmets, but helmet laws infringe personal
liberties.  I personally don't support them, because bike fatalities just
aren't a high-prevalence problem (though you won't see me riding without
my helmet on personally!).

The neurosurgeon has a different view, but with all due respect to my
colleagues it's a biased view: neurosurgeons see the worst-case victims,
no matter how few of them there are.  We all react to what we see.


Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Green MD MPH) 
Date: Sep 01 1995
Newsgroups: rec.bicycles.misc,rec.bicycles.off-road,rec.bicycles.tech

In article <DE6JLv.3vG@freenet.carleton.ca>, ab833@FreeNet.Carleton.CA
(Avery Burdett) wrote:

> > Ask one of them, any one, to show you any results.  Whether it's bike
> > safety, fire safety, water safety, childhood poisonings, whatever.  They
> > can't.  They never will be able to.  Attempts at public behavior change
> > are, in general, depressingly dismal failures.

> For an MD you really post a lot of rubbish.

If you have a substantive argument it can stand on its own, without
attacking the other person.  If you just want to be a rude boy, go to one
of the alt groups.

> The obvious example where
> public attitudes have changed is in regards to drinking and driving, or
> perhaps your not old enough to remember when it was OK to party all night
> and then drive home.

This "obvious example" is not akin to attempts to improve the skills of
cyclists, nor the storage of household poisons or other examples I cited. 
Public attitudes have changed about drinking and driving, but the changes
in death rates have come about exactly and only where the *laws* have
changed.  Public attitude has brought about change by bringing about laws
mandating serious penalties for drunk driving.  Which is precisely the
point.  If we had laws prohibiting cycling alone until someone had a year
of riding with a club, we'd have safer cyclists.  The inescapable fact
remains however that attempts to demonstrate better outcomes as a result
of *public-education safety campaigns* (in contrast to laws with real
teeth in them) reliably fail to do so.  They're just plain ineffective.

Subject: Re: Great Helmet War
From: greenla@umich.edu (Lee Green MD MPH) 
Date: Sep 06 1995
Newsgroups: rec.bicycles.misc,rec.bicycles.off-road,rec.bicycles.tech

In article <DEGC5r.8zr@boi.hp.com>, Cliff Weisgerber <cliff@boi.hp.com> wrote:

> Wrong! Here's why:

I'm sorry, but it's not wrong.  The rationale you present as to why is
creative, but the numbers don't bear it out.

> A helmet will do less to prevent injury than the avoidance of an
> accident in the first place

Of course.  No one doubts that reducing accidents is a good thing.  The
question was whether the bike safety campaigns accomplish that goal.  Such
campaigns, and ones like them for fire safety, water safety, household
poisoning reduction, etc., have long been promoted by well-meaning people
using just the sort of rationale you've presented.  Unfortunately, such
programs are seldom subjected to program assessment: measuring whether
they actually achieve their goals of reduced accidents, rather than just
giving the people who put them on a warm fuzzy feeling that they've done a
good deed.

In the cases where program assessments have been done, such programs have
been consistently found ineffective at changing behavior.  They attract
participation mostly from the people already least inclined toward
risk-taking behavior, and the changes seen in people's behavior revert
within weeks.  It was this research, done as far back as the mid-1970s,
which finally led policymakers to give up on the National Safety Council's
poison education program and require childproof caps on aspirin bottles. 
Poisonings declined rapidly and immediately.

> Funny how the idea of accident reduction does *not* appeal to one
> such as yourself who makes money by treating the injured.

Warning: well-deserved flame follows.

That's about the snottiest and wrongest thing you could say to someone who
does what I do for a living.  I've had to wash bits of a child's brain off
my hands, go out in the waiting room, and tell the father that his
daughter did not make it.  When you come with me and face that situation,
and feel what that feels like in the pit of your stomach, and try to
provide some support for that father, then you can comment on what I want
to see happen, you son of a bitch.

Flame off.

I would very much like to find an *effective* way to reduce accidents. 
But the approaches which have been tried in the past have not been
effective.  Until someone finds a way to reduce accidents which actually
works, not just makes the people doing it feel good, we're stuck with
trying to reduce the lethality of the accidents which happen.

Perhaps requiring every cyclist to take something akin to the Effective
Cycling course would be effective, but I doubt the political will to make
such a requirement exists.  It is very unlikely that a bicycle operator
license with genuine training requirement could ever be enacted, as that
would entail an infringement of liberty most would not accept.  In the
meantime, I see little point in continuing to hand out "bike safety"
pamphlets known to be ineffective.  Let's put our work into finding
something that actually does some good instead.


From: cbm@athena.mit.edu (Charles Bruce Musgrave)
Newsgroups: rec.bicycles.racing
Subject: Re: Campag Delta Brakes
Date: 4 Mar 1996 16:06:24 GMT
Distribution: world

In article <tomkDnMpuE.HB6@netcom.com>, tomk@netcom.com (Thomas H. Kunich) writes:
|> In article <4h6vd2$er5@senator-bedfellow.mit.edu>,
|> Charles Bruce Musgrave <cbm@athena.mit.edu> wrote:
|> >
|> >You keep making these kinds of general statements, but including 
|> >little real data or explanation to back them up.  In many cases
|> >a helmet can obviously help reduce head injuries.  For example,
|> >in cases where the speed differentials aren't high, or where
|> >most of the speed differential is contributed by the cyclist
|> >(Teeing a car that makes a left turn in front of you, e.g.).
|> 
|> I see that you've really been doing your homework before making
|> public statements. Strange how manditory helmet laws in both
|> New Zealand and Australia don't seem to have reduced the rates
|> of serious or fatal head injuries. And if they don't prevent that
|> then why wear one?

Maybe you should look a little harder at the stats before making
statements yourself.  What was the rate of complience before and
after the mandatory laws?  What is the year to year fluctuation
in serious and fatal head injuries before and after the mandatory
helmet law?  What is the difference in rate of serious head injury 
between helmeted and unhelmeted cyclists?   


|> >If you know something that we don't about the deformation of
|> >materials, how impact force is or is not deflected through
|> >transmission through a solid, or partially adsorbed by 
|> >material elastic, and plastic deformation, please share.
|> >
|> >Otherwise, your comments, like the one above will continue to 
|> >look like the product of someone speaking from a fanatical,
|> >and ignorant position, or from someone whose head met a car
|> >with unfortunate trajectory conditions.
|> 
|> Exactly what do you know about the "deformation of materials"?

A little.  I have a B.S. in materials science and engineering from
Berkeley, a PhD. in materials science from Caltech. A postdoc in
chemE and materials science at MIT and a professorship in materials
science and engineering at Stanford.  Believe it or not, I've
picked up a few things on how materials deform.  

|> What is the basis of your belief that a hat made of 1" of foamed
|> plastic can have a significant effect at the velocities and forces
|> found in the average serious accident?

First, the pressure defocuses as it is transmitted through the material.
That is the forces of impact are spread over a larger area after
being transmitted through the material.  Look at the biomechanics of
a running shoe which uses an even thinner piece of foam to spread
out the forces of impact.  The ratio of the areas depends mostly
on the Poisson's ratio for the material which is very favorable for
foam.

Second, the shockwave is disappated as it is transmitted through 
the material, in addition to being defocused.  This is obvious
from even a casual knowledge of the phonon dispersion curves for
materials like foam. The disappation is high because the sound 
velocity in foam is low, the void volume high, and the dispersion 
curve bends over severely and early, 

Third, the energy to plastically deform the material further
reduces the transmitted forces, in addition to that energy 
disappainted through elastic deformations.  Energy is required
to crush the material which is no longer available to crush
one's head.  

If you read closely, I never stated that a helmet would protect
you from death in every accident.  However, in many cases what might be
a routine accident with a helmet, could be quite serious without
one and the extent of head injuries is reduced by wearing a helmet 
in the vast majority of cases.

|> 
|> Strange how those who haven't studied a problem are still qualified
|> to recognize "fanatical and ignorant" positions.

Unfortunately, your fanaticism and ignorance is all too easy to 
recognize.


How about a challenge, Tom?  You and I both are to have a brick
dropped onto our heads from a height of 4 feet above us.  
You not wearing a helmet, and me wearing mine.  It is apparent
from your posts that this won't be the first time you have 
accepted such a challenge, but I'd like this one videotaped.

Charles

Index Home About Blog