bicycle helmet

Helmets aren’t saving lives but they might save yours

The helmet debate cropped up again in mainstream media with a glowing endorsement in their headlines such as:

Sydney Morning Herald:

Bike helmet review throws cold water on sceptics: they’ll likely save your life

Special Broadcasting Service:

Report proves bike helmets the difference between life and death

ABC Melbourne:

Do cycling helmets save lives? Researchers reject doubters and say fatal injuries greatly reduced

All from a meta-analysis study evaluating the efficacy of helmets in reducing fatalities, from medical literature. Its time for a response from a sceptic and we can start with the famously reproduced time series of Australian cycling fatalities around the introduction of Laws requiring mandatory use of bicycle helmets (lighter line).

decline of cycling deaths during mandatory bicycle helmet law introduction

At face value there is a substantial reduction in cyclist fatalities around the time the law was introduced, but this ignores the larger trends in road safety that were also occurring around the same period. Creating a longer view of trends and including the total road toll and pedestrian road tolls to the mix we can look for other trends.

decline of all transport mode deaths

Immediately it can be seen there were similarly abrupt drops in the road toll for pedestrians and even drivers, the sudden change in the overall road toll of approximately 1000 fatalities, when the change in pedestrian and cyclist tolls was only 250. This overall trend was disproportionally more effective for cyclists and pedestrians, but the other well publicised road safety campaign occurring at the time was the introduction of speed cameras shown with the red bar. Bicycle helmet laws came late to the party and claim all the credit when there were larger system wide effects on all modes of transport. The drops could be attributed to simply the increased public/media attention of road safety issues, to extract cycling specific interventions we can look at the mode share and share of fatalities the cycling comprises.

cycling mode share following cycling share of road toll fatalities

Using the Australian census data for method of travel to work collected since 1971 (if anyone has this data its contribution would be appreciated) we can estimate the change in cycling rates across the same period. Averaging the 5 years of road deaths around each census date the two are plotted on the same scale and we can see a reduction in cycling rates after the introduction of helmet laws but a lesser reduction in their share of the road toll. The ratio of these two is the relative risk of death compared to the overall transport population.

relative risk multiplier of cycling amongst all modes of road transport

Helmets fail to correlate with an improvement in the safety of cyclists, it is hard to draw any positive or negative correlation in this data to the safety of cyclists and the introduction of mandatory helmet laws. Helmets do reduce head injuries in the case of an accident, but their compulsory use in Australia cannot be attributed to a reduction in the deaths of cyclists unless we claim there were other factors which were increasing the risk to cyclists at the same time and bicycle helmets have offset this.

There are many well developed theories of risk compensation, and as motor vehicles get safer if the occupants maintain a static risk to themselves they increase the risk for other road users (http://www.john-adams.co.uk/2009/11/05/seat-belts-another-look-at-the-data/). This has been flipped on its head with research suggesting that drivers through their (unconscious) choices will expose a cyclist wearing a helmet to more risk (http://drianwalker.com/overtaking/). From the theory of risk compensation just promoting the safety of helmets would lead to further risk taking, promoting benefits higher than what is achievable would result in an increase in death and raises a substantial moral hazard.

It would appear from the longer timescale graph that the influential change in Australian road safety for cyclists was not helmets but the introduction of wider road safety campaigns affecting all road users, particularly speed enforcement.

decline of all transport mode deaths

Helmets are effective in reducing head injuries, this has considerable consensus, but extrapolating this through the complex system of road transport to draw a conclusion that this is saving lives lacks convincing data. Reversing the argument from road safety data and instead extrapolating to the effectiveness of helmets we could conclude that helmets aren’t protecting people. Both extrapolated conclusions create a paradox that is shown to be untrue.

Helmets reduce head injuries and fatalities in bicycle accidents.
but
Mandatory helmet laws have not reduced cyclist death rates.

These are not contradictory statements that need argument, they act upon each other through a wildly complex system of fallible humans each with their own cognitive biases that allow them to both be true simultaneously without paradox.

The selectively promoted 70% reductions in “serious head injuries” in accidents when comparing those wearing helmets with those who don’t has been distorted by the media once again in tangling it with fatalities and “saving lives”. Dr Jake Olivier has been careful to frame their research results around “serious head injuries” as that is what they have quality data for, attempts to associate helmet wearing with fatalities resulted in extremely wide error margins (http://event.icebergevents.com.au/uploads/contentFiles/files/2015-AIPN/Jake-Olivier.pdf). The handling of mathematical/statistical data by the media has always been comedic, but this has the possibility to give cyclists a false sense of security for the effectiveness of helmets, which if risk compensation is then applied will result in an increase in deaths and injury.

Jake Oliver has even had to comment on media coverage previously in their detailed blog, and puts a very balanced view forward when not distorted by poorly educated journalists:
https://injurystats.wordpress.com/2015/06/29/no-one-thinks-helmets-are-a-panacea-for-cycling-safety/

From their other posts is an agreement that the choice of mode share data is critical to these types of statistics presented here, and the Australian census is not perfect but is the only wide scale long term data available.

We can all agree helmets are one part of improving the safety of cycling, I’ll side with the groups that believe they should not be mandatory. But risk compensation places a moral hazard on the reporting of road safety efficacy, it must be accurate and accessible to all not just a headline grabbing for attention with misleading claims.

Advertisements

2 thoughts on “Helmets aren’t saving lives but they might save yours

  1. It isn’t a universal consensus that bicycle helmets prevent serious head injury. Spiegelhalter and Goldacre’s editorial in the British Medical Journal rather states that helmets don’t do that much to address it:

    “In any case, the current uncertainty about any benefits from helmet promotion or wearing is unlikely to be reduced by further research. Equally, we can be certain that helmets will continue to be debated, and at length. The enduring popularity of helmets as a proposed major intervention for increased road safety may lie in not their direct benefits — which seem too modest to capture compared with other strategies — but more with the cultural, psychological and political aspects of popular debate around risk.”

    http://www.badscience.net/2013/12/bicycle-helmets-and-the-law-a-perfect-teaching-case-for-epidemiology/

    A meta-analysis by Rune Elvik, rather similar to the one under discussion, and which this one was probably written explicitly to counter, was rather less than glowing about their benefits too.
    Elvik R. Accident Analysis & Prevention, 2011;43(3):1245-1251.

    Like

    1. These references are converging on the same problem of communication to the public and a lack of balance in whats presented. We can find some quality data sets for Australia (Victorian hospital admissions) which show a small reduction in head injuries adjusted for cycling rates and reporting methods, but that must be presented in the context of risk compensation, self selection, and other possible influences that puts in question the overall result. I’m not sure we can ever correct for all the underlying contributions simply because the numbers of occurrences are so small they are too noisy to extract any of the subtle effects.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s