Effectiviteit van fietshelmen overschat door het gebruik van kansverdelingen

kade rechteroever

De effectiviteit van de fietshelm wordt vooral onderbouwd door zogenaamde case-control studies. In dit type studie inventariseert men het aantal fietsers met hoofdletsel, bijvoorbeeld op eerstehulpafdelingen van ziekenhuizen, met en zonder helm (de ‘case’) en vergelijkt dat met het aantal fietsers op straat met en zonder helm (de ‘controle’).

Is het aandeel helmdragers onder de slachtoffers met hoofdletsel lager dan op straat, dan leidt men daaruit een positief effect van de helm af. In de praktijk ontbreken evenwel gegevens over het helmgebruik op straat. Noodgedwongen nemen deze studies daarom als controle groep patiënten op eerste-hulpafdelingen die ander letsel dan aan het hoofd hebben.

Dat is alleen gerechtvaardigd wanneer het dragen van een fietshelm verder geen invloed heeft op de kans om niet-hoofdletsel op te lopen. Immers, wanneer er wel een invloed is van het dragen van een helm (bijvoorbeeld meer kans op letsel), is de controlegroep niet gelijk aan (en dus niet representatief voor ) de fietser op straat.

In dat geval zijn de conclusies niet van toepassing op de gewone fietser op straat. De vraag was lange tijd of aan deze stilzwijgende aanname in het geval van de fietshelm voldaan is.

Theo Zeegers heeft de aannames nader onderzocht en nu blijkt dat gemiddeld genomen fietsers met een helm op aanzienlijk vaker niet-hoofdletsel oplopen dan fietsers zonder helm. Hiermee heeft Zeegers aangetoond dat de controlegroep niet representatief is.

Daaruit volgt dat gerapporteerde positieve effecten uit helmonderzoeken niet valide zijn. In een drietal gevallen die expliciet in deze studie onderzocht zijn, was het verschil tussen de oorspronkelijke en gecorrigeerde cijfers zeer groot.

Zo bleek in heranalyse van een Nederlandse studie van Veiligheid.nl de kans voor een gehelmde fietser om niet-hoofdletsel op te lopen bijna driemaal zo hoog. Dit is waarschijnlijk het gevolg van het relatief hoge risico van wielrenners en mountainbikers op ongevallen. Hierdoor belanden ze vaker op de eerste hulp en dus ook in de controlegroep.

De Stichting Wetenschappelijk Onderzoek Verkeersveiligheid heeft zich jarenlang beroepen op casecontrolestudies om de effectiviteit van de fietshelm te onderbouwen. Nu al deze studies niet langer valide blijken te zijn en andere soort analyses wijzen op het afwezig zijn van een positief effect van de fietshelm, moeten claims over de effectiviteit van de fietshelm herzien worden.

Zeegers hoopt dat zijn onderzoek de overheid uitdaagt om snel aan te sturen op een betere onderzoeken naar de effectiviteit van helmgebruik en dat de overheid promotiecampagnes voor de helm tot die tijd stillegt.

Overestimation of the effectiveness of the bicycle helmet (pdf)

Theo Zeegers
Fietsberaad 09-2015
http://www.fietsberaad.nl/?repository=Effectiviteit+van+fietshelmen+overschat+door+het+gebruik+van+kansverdelingen

 

Helmet effectiveness research forced to go back to the drawing board

New research from Theo Zeegers in the Netherlands seems to have thrown the cat amongst the pigeons with regards to bicycle helmet research.

A Dutch cycling researcher and advocate Theo Zeegers has put forward a piece of research that may force us to conclude that we now have very little real life case evidence for the effectiveness of the bicycle helmet in preventing head injuries.

Bicycle helmet research has had two main threads of research, time series studies and case-control studies. Time series research[1] documents the rise and fall of head injuries and non-head injuries just before, during and after the introduction of bicycle helmet legislation (in Australia in 1991 for example) or increased helmet usage. The conclusions of these studies seem to be that there is very limited or no benefits from the increased usage of helmets. Arm and leg injuries follow the same rate as head injuries telling us that helmets are probably not the cause of any fall in head injuries rather it is likely that safer infrastructure, drink-drive legislation, safer vehicles etc. are responsible. This of course does have complications in that it doesn’t tell us for sure that an increase in helmet wearing itself may or may not be beneficial as the introduction of the law itself may have an effect on the behaviour cyclists populations.

Then we have case-control studies[2] which compare the number of helmeted and non-helmeted cyclists with a head injury during an accident, with the number of cyclists actually on the street with and without a helmet. If the number of head injured cyclists wearing a helmet (the case) is lower than the number of cyclists on the streets in general (the control) then it is claimed that the helmet is effective. However the number of cyclists on the streets that wear helmets is often not known and so the control group is often made up of a group of hospital patients with cycling injuries other than the head. This group then acts as the representative of the cyclist on the street (helmet wearing being more clearly known). However this group can only be truly representative if there is no correlation between helmet usage and the risk of getting a non-head injury.

Theo seems to have shown mathematically that this is in fact not the case and that the risk of getting a non-head injury is indeed correlated to helmet use. This then invalidates the use of this control group as a way of controlling for non-head injuries; which then in turn invalidates these case-control studies.

This was then checked against three real world studies and an over-estimation did occur; in all three cases the risk of getting a non-head injury was higher than without, and in two getting a head injury was higher than without a helmet. This does not necessarily mean that wearing a helmet is more dangerous (though some risk compensation studies have also proposed this possibility) rather it is more likely that some helmet wearing ‘populations’ (racing cyclists for example) partake in riskier cycling and so are more prone to accidents[3].

In any case what Theo’s study has shown is that evidence for bicycle helmet effectiveness that is brought to us through case-control hospital studies is flawed. Given that much of bicycle helmet legislation is based on the evidence of such studies this would also mean that mandatory bicycle helmet policy and legislation is also flawed.

I said that there were three threads to bicycle helmet research, but there is another interesting way to look at the effectiveness and that is to see this in the light of the benefits or costs to public health. When we deal with road safety we often think of Newtonian physics science, or heads hitting bonnets, of braking distances and forces in crashes and the effects on the human body. This is all well and good for many modes or transport, however it doesn’t really work for active mobility like cycling and walking. Passive modes of transport all have costs when it comes to public health such as a sedentary lifestyle and road crashes, cycling on the other hand has costs in the form of crashes but also benefits in that it takes effort to move which has a positive effect on our health.

What we do know is that the benefits outweigh the costs by around a whopping 20:1[4], once we factor this into the equation any reduction in cycling numbers brought about by helmet legislation, or indeed any road safety intervention, will almost invariably lead to a public health disbenefit. Piet de Jong[5] has calculated that even if bicycle helmets were 100% effective for all injuries in all crashes it would still only take a small reduction in cycling numbers to make bicycle helmets an ineffective public health measure.

Theo’s paper will be appearing in a peer reviewed journal soon, we will keep you posted. Until then it is available here and here.

Most bicycle helmet research can be found at this excellent website, often with critical reviews http://www.cyclehelmets.org

 

[1] http://injuryprevention.bmj.com/content/16/Suppl_1/A228.3.abstract

http://www.sciencedirect.com/science/article/pii/S0001457500000737

http://www.sciencedirect.com/science/article/pii/0001457596000164

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1410838/

http://www.ecf.com/news/what-happens-when-you-mandate-helmet-wearing-among-young-swedish-cyclists/

[2] http://www.ncbi.nlm.nih.gov/pubmed/2716781

http://www.ncbi.nlm.nih.gov/pubmed/8971066

www.ncbi.nlm.nih.gov/pubmed/10796827

[3] http://www.cyclehelmets.org/1261.html

[4] http://www.bmj.com/rapid-response/2011/10/31/health-benefits-utility-cycling-evidence-overlooked

[5] http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1368064


Ceri Woolsgrove
ECF 26-10-2015
http://www.ecf.com/news/helmet-effectiveness-research-forced-to-go-back-to-the-drawing-board/

Tags: