Saving the Unicorn

Recently, Jersey voted overwhelmingly to make cycling helmets compulsory for under 14s. It turns out this was based on a report from the Transport Research Laboratory. Let’s take a look at it, then.

Prologue

I’m going to ignore the material pertaining to the efficacy of helmets in individual incidents. Partly because it’s a can of worms that warrants an extensive article of its own; partly because, as we shall see, it turns out to not really be very important.

I should also add that I’ve not read all of the cited references. There are 68 of them and I simply don’t have the time. There are a couple that stand out when reading the report, which I shall endeavour to read at a later date, but what I’ve written here is based on the TRL report only. I don’t consider this unfair: the report should stand on its own merit, and will doubtless have been treated accordingly when the legislation was passed.

Let’s jump into the report at section 4.

Helmet wearing rates

Section 4.1 (p27) discusses helmet wearing rates in states and countries where helmet use falls into one of three categories: optional, compulsory, and enforced.

It notes that in the Australian provinces of Queensland and Victoria, compulsion caused helmet wearing rates to jump from 31 to 75% and from 16 to 52% respectively, with Queensland’s rate continuing to rise to match that of Victoria.

It notes that in Alberta, Canada (which has compulsion for under 18s), rates for under 13s rose from 75 to 92%, and from 30 to 63% for 13-17 year olds; also, that Nova Scotia (with universal compulsion) has helmet wearing rates of 77% in youths and 71% in adults, Ontario (under 18 compulsion) has corresponding figures of 47% and 39%, and Saskatchewan (no compulsion) has figures of 33% and 25%.

It is worth pointing out that in the context of helmet wearing rates, the TRL report makes no comment on the cycling population figures before and after legislation. Take any two mutually exclusive sets, and if you reduce the population of one, the rate of membership of the other increases. In other words, a rise in the helmet wearing rate among cyclists can be achieved by more cyclists wearing helmets, or by fewer non-helmet wearing cyclists cycling. In reality there will be a combination of both, but TRL do not discuss this here.

If, however, we optimistically assume that there is zero discouragement from cycling, we note that of the non-helmet wearing population, around half obey the new law.

The pessimistic assumption, of course, would be that none obey the new law, and half cease cycling; the reality would be somewhere in between. I do not buy TRL’s rather odd counter-argument to the population drop effect, that—

It is also possible to argue that mandating helmets could be perceived by the public as an act by authorities to improve cycling safety in general, thus overcoming what is known to be a major barrier towards cycling.

—because anyone who feels cycling is made safe by wearing a helmet has the option to do so, and anyone who feels that the cycling environment poses a risk against which they consider a helmet insufficient protection is only going to have that feeling reinforced if authorities acknowledge that problem by mandating helmets instead of addressing it.

(Amusingly/tragically, TRL follow their curious hypothesis with, “Evidence is more important than intuitive-sounding counter-arguments, whichever way they are aligned with regard to the debate.” – but I digress.)

Now, bear in mind those rates and let’s skip to the data from Jersey.

Take a look at section 6.1 (p36).

The current helmet wearing rate for under 17s in Jersey, with no legislation in place, is 84%.

Compare that to the rates above and it’s astonishing. It’s higher than most places with compulsion, and by no small margin. And, as noted later in the report, it’s rising.

This figure gives credence to one of TRL’s conclusions, which is that compulsion is unlikely to have a major effect on cycling participation on the island.

However, in the context of its other conclusion – that legislation can be expected to provide benefit – it raises a key question: Since helmet wearing rates are already well above those which might normally be expected under compulsion, what will compulsion achieve? The absolute best it can achieve is to encourage 16% of children to wear helmets, but figures from elsewhere suggest that at most this will be 8%, and that’s before we consider any reduction of the cycling population.

But the data from Jersey gets even more interesting.

The unicorn

The TRL guidance considered casualty data from Jersey. Curiously, there is only one year’s worth of data given in the report, for 2013, but it’s notable that cyclist casualties on Jersey were exceptionally high in 2013. Most years have seen markedly fewer casualties.

Have a look through figure 6-3 on page 39, which lists all cycling casualties by age and gender.

Now, remember that Jersey’s law applies to under 14s.

That table lists just two casualties for under 14s: two boys aged 11 and 13, both slightly injured.

Head and brain injuries are not slight injuries. And slight injuries are, other than being markers for incidents, not really of interest in the context of public health.

So – and here is where we realise can ignore the discussion about efficacy – where is the problem that helmet compulsion is supposed to address? The report contains no evidence that it exists. None.

Remember, the legislation will only have an effect where:

• the individual is under 14, and
• is in the 16% not already using a helmet, and
• is in the 50-100% of that 16% who will start using a helmet rather than not cycle, and
• is in the 50% at most of that 50-100% of that 16% who – given the pre- and post-compulsion data cited in the report – will actually obey the law, and
• is involved in an incident in which, without a helmet, they would sustain a head injury, and
  
• the incident is of a nature where the helmet makes a substantive difference to the outcome

All of these criteria must be fulfilled by at least one individual if there is to be any gross benefit at all (ie disregarding any disbenefit). The last point heads into a very speculative area, where section 3 of the report (regarding efficacy) becomes relevant, but the available data – the data cited in the report – shows that before we even get there, the set of potential beneficiaries is zero.

Of course, despite the lack of data, over the long term the chances are it’s not zero – but it’s without doubt very, very small; and once the rest of the points are accounted for, the probabilities of any benefit dwindle to negligibility. If there occurs an incident in which a helmet makes a substantive difference, there is between a 4 and 8% chance that the legislation will have caused a helmet to be worn where it otherwise would not.

Chance of any positive effect: a low probability, times a low and unknowable probability, times roughly 6%. And then multiply it by the health benefit the helmet actually gives; at which point we’re into the can of worms that is section 3 of the report.

And remember, this is purely gross benefit we’re talking about. Not net benefit.

A solution to what?

When forming their guidance for this legislation, the TRL considered a set of data which contained two people to whom it would apply, and for neither of whom compliance with that legislation would have made any difference.

Of course, any inference that the legislation is totally futile is a little undermined by there only being a single year’s data on offer; but then, the exact same can be said of TRL’s opposite inference, that legislation is worthwhile.

Significant time and money has presumably been invested in the report. It’s understandable that it covers matters such as helmet efficacy, but it’s hard to see past the rather more prominent fact that based on the available data the legislation is incredibly unlikely to achieve any positive health benefit.

Helmet wearing rates are already exceptionally high and rising; casualty rates of the target population are already exceptionally low. Jersey is achieving better voluntary helmet use than legislation normally achieves.

This is the “if it saves one life” line of thought distilled to absurdity: the data show no lives to be saved, nor even any injuries to prevent. All of the statistics point to there being no effect even if helmet use runs to 100%, and even if no-one is discouraged from cycling.

TRL’s advice is that “legislation requiring the wearing of cycle helmets in Jersey can be expected to have a beneficial effect on the injury rates of those impacted by the legislation“.

Yet the data in the document do not bear this out. Not in the slightest.

So… why was this legislation introduced?

Conspiracy theorists, it’s over to you.

Footnote

So, who is dying from head injuries on Jersey?