Cyclists in Cyprus call for repeal of compulsory helmet law

Rich_cb wrote:

It's a tiny sample yes but that's why any evidence is hard to obtain. Case control studies are the best evidence we have. They show a benefit.

Berkson's paradox applies to any size of sample and is one of the reasons that the case control studies need to be treated with scepticism and certainly not used as a pretext to put PPE ahead of actual road safety initiatives that reduce danger.

Case control studies are high quality evidence and they are widely accepted as being particularly useful when studying rare disease/outcomes which makes them an ideal fit for investigating the efficacy or otherwise of bicycle helmets.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1706071

A small sample size is never as good as a larger sample size but you can still draw valid conclusions with a well designed case control study.

Rich_cb wrote:

Case control studies are high quality evidence and they are widely accepted as being particularly useful when studying rare disease/outcomes which makes them an ideal fit for investigating the efficacy or otherwise of bicycle helmets. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1706071 A small sample size is never as good as a larger sample size but you can still draw valid conclusions with a well designed case control study.

I'm not trying to rubbish case control studies but they need to be examined in context. They work really well for diseases as people contracting the disease and those being admitted to hospital with the disease are the same and so it makes sense to extrapolate.

They also work well for cyclists who are involved in accidents/collisions. Both the helmeted and non helmeted will be attending hospital for their injuries.

If you look at otherwise identical cyclists with identical non head injuries, eg a broken arm and broken ribs, and then compare the rate of head injuries for the helmeted Vs non helmeted you would be able to calculate the relative risk of head injury in the helmeted Vs the non helmeted.

It's obviously not perfect but it's a well established technique and considered a high quality form of evidence.

Rich_cb wrote:

They also work well for cyclists who are involved in accidents/collisions. Both the helmeted and non helmeted will be attending hospital for their injuries. If you look at otherwise identical cyclists with identical non head injuries, eg a broken arm and broken ribs, and then compare the rate of head injuries for the helmeted Vs non helmeted you would be able to calculate the relative risk of head injury in the helmeted Vs the non helmeted. It's obviously not perfect but it's a well established technique and considered a high quality form of evidence.

The problem is with extrapolating the unfortunately injured cyclists to the rest of the cycling population. When you only sample from injured cyclists, then it's entirely possible that they may be very unlucky, have poorly maintained bicycles, take a lot of risks or ride in areas prone to poor driving. More importantly, the figures will ignore the cyclists that don't have incidents and the ones with minor incidents that don't require hospitalisation (both naked and helmetted).

It seems likely that only selecting from hospitalised cyclists is going to bias the result as there's no easy way to determine that those cyclists are representative of the entire cycling population.

But, as you say, it's probably the best tool we have apart from population level studies which disturbingly, don't show much benefit to helmets assuming that helmets are to reduce head injuries and not just reduce the numbers cycling.

I'm not really following your point.

If we want to know how well a helmet will protect you in a significant collision/accident then we have to look at cyclists who have been in significant collisions/accidents.

It doesn't matter how well they maintain their bike etc. All that matters is that they were in a significant collision/accident.

If a helmeted cyclist crashes because he hit black ice and a non helmeted cyclist has an identical crash because they did not maintain their bike they can be compared for the purpose of assessing how much protection the helmet afforded the wearer. The actual cause of the crash isn't important.

Population level studies are far less reliable than case control studies for assessing the efficacy of helmets as the populations are not matched as they are in a case control study so you cannot be sure that differences are not down to changing population characteristics rather than changes in helmet wearing.

Rich_cb wrote:

I'm not really following your point. If we want to know how well a helmet will protect you in a significant collision/accident then we have to look at cyclists who have been in significant collisions/accidents. It doesn't matter how well they maintain their bike etc. All that matters is that they were in a significant collision/accident. If a helmeted cyclist crashes because he hit black ice and a non helmeted cyclist has an identical crash because they did not maintain their bike they can be compared for the purpose of assessing how much protection the helmet afforded the wearer. The actual cause of the crash isn't important. Population level studies are far less reliable than case control studies for assessing the efficacy of helmets as the populations are not matched as they are in a case control study so you cannot be sure that differences are not down to changing population characteristics rather than changes in helmet wearing.

It's to do with possible misleading conclusions from just using hospital admissions. I'm specifically thinking of Berkson's paradox: https://en.wikipedia.org/wiki/Berkson%27s_paradox

Quote:

Berkson's original illustration involves a retrospective study examining a risk factor for a disease in a statistical sample from a hospital in-patient population. Because samples are taken from a hospital in-patient population, rather than from the general public, this can result in a spurious negative association between the disease and the risk factor. For example, if the risk factor is diabetes and the disease is cholecystitis, a hospital patient without diabetes is more likely to have cholecystitis than a member of the general population, since the patient must have had some non-diabetes (possibly cholecystitis-causing) reason to enter the hospital in the first place. That result will be obtained regardless of whether there is any association between diabetes and cholecystitis in the general population.

There's some more detail on specific instances here: https://rss.onlinelibrary.wiley.com/doi/10.1111/1740-9713.01413

Quote:

Like other threats to causal inference (e.g. a third factor explaining a link between the exposure and outcome), once you become aware of the collider bias mechanism, you see it lurking everywhere. This includes studies of hospitalised patients, for example. While hospitalised patients are an obvious source of Covid-19 data, they are not a random sample of the general population, but instead represent individuals who are effectively selected from that population on the grounds of being older, frail with health issues, smokers, and – of course – those suffering from Covid-19.

To see how this could lead to a misleading link between smoking and protection against Covid-19, imagine – for simplicity – that patients are admitted to hospital for just two reasons: smoking-related illness or Covid-19 (Figure 2b). Covid-19 tests on these hospitalised individuals are likely to show lower infection rates among smokers than among non-smokers, because the former are also hospitalised for smoking-related illness, not necessarily Covid-19. This could explain the reports from several studies claiming that smoking might protect against Covid-19 infection.

That's not really applicable in this instance.

We are assessing the efficacy of helmets in preventing significant injuries from accidents/collisions when cycling.

To assess this we need a population of people who have been significantly injured whilst cycling.

That population will almost inevitably present at a hospital.

The uninjured population are irrelevant to this study.

The only groups who will be missed are those with fatal head injuries who die before reaching hospital and those helemted riders whose helmets prevent an isolated head injury and who therefore don't present to hospital. In both cases the effect will be to reduce the apparent efficacy of helmets.

Rich_cb wrote:

That's not really applicable in this instance. We are assessing the efficacy of helmets in preventing significant injuries from accidents/collisions when cycling. To assess this we need a population of people who have been significantly injured whilst cycling. That population will almost inevitably present at a hospital. The uninjured population are irrelevant to this study. The only groups who will be missed are those with fatal head injuries who die before reaching hospital and those helemted riders whose helmets prevent an isolated head injury and who therefore don't present to hospital. In both cases the effect will be to reduce the apparent efficacy of helmets.

That seems overly simplistic to me. For a start, those studies don't even consider the cases where helmets have prevented hospital admissions - that seems a glaring oversight when that's the topic under discussion.

If the case control studies are accurate, I'm puzzled as to why they show so much discrepancy between them - why wouldn't they be consistent?

Personally, I'm far more interested in remaining within the uninjured population - that's where the focus on road safety should be.

Amen!  Help yourself, but if others are positively uninterested in doing anything at the level of the more effective interventions but are just hassling you about you taking the least effective ones...

The cases were helmets prevented a hospital admission will be missed so, as I said previously, the studies will likely underestimate the efficacy of helmets.

The difference in results could occur through simple statistical variance or through differences in the local population/environment. Eg a hospital near Whistler may see a very different injury mix to one near Whitstable.

I agree that preventing the accident/collision from occuring in the first place is the best course of action but the same could be said for all manner of research. That doesn't reduce the value of said research.

Rich_cb wrote:

The cases were helmets prevented a hospital admission will be missed so, as I said previously, the studies will likely underestimate the efficacy of helmets. The difference in results could occur through simple statistical variance or through differences in the local population/environment. Eg a hospital near Whistler may see a very different injury mix to one near Whitstable. I agree that preventing the accident/collision from occuring in the first place is the best course of action but the same could be said for all manner of research. That doesn't reduce the value of said research.

I've pointed out specific issues with hospital admission case studies and I don't consider them sufficiently accurate to drive policy and yet, that does appear to be what's happening in Cyprus. What we need is wider studies that aren't focussed on hospital admissions, but unfortunately it seems that politicians are more interested in cheap, quick, victim blaming "fixes" rather than actually taking road danger seriously.

If you want to know if helmets actually work at preventing head injuries then case control studies are the gold standard. There are no viable alternatives.

Is it the only thing to consider when devising policy? No, but I've never said it was.

Rich_cb wrote:

If you want to know if helmets actually work at preventing head injuries then case control studies are the gold standard. There are no viable alternatives. Is it the only thing to consider when devising policy? No, but I've never said it was.

I think the issue is that trying to push the benefits of helmets when only looking at hospital admissions is missing the bigger picture of people's behaviour. Ultimately, it's irrelevant if helmets provide a certain amount of protection (especially when we can't even agree what that figure is) when the population level studies have indicated that the negative effects seem to be larger. I'd guess that full plate armour would be extremely good at full body protection, but it's risible to suggest that we should mandate it to improve road safety.

You may consider that case control studies are the gold standard (and I'd agree with you for medical trials etc), but they are misleading people to think that mandatory bike helmets are a good idea and as I've stated, there's good reasons to think that they include bias when extrapolating from hospital admissions to all cyclists.

The only bias they could introduce is underestimating the efficacy of helmets at preventing injury.

I agree that helmet efficacy is only one part of the debate around mandatory helmet laws (which FWIW I oppose) but there is a lot of good evidence for the effectiveness of helmets at preventing and it makes no sense to dismiss that evidence on the basis that mandatory helmet laws are harmful.

Rich_cb wrote:

The only bias they could introduce is underestimating the efficacy of helmets at preventing injury. I agree that helmet efficacy is only one part of the debate around mandatory helmet laws (which FWIW I oppose) but there is a lot of good evidence for the effectiveness of helmets at preventing and it makes no sense to dismiss that evidence on the basis that mandatory helmet laws are harmful.

I'm not dismissing the evidence, but suggesting that it's not rigorous nor conclusive when it's only looking at part of the picture. The bias could easily go both ways as the naked cyclists that don't present at hospitals aren't being counted either. I do believe that bike helmets provide some level of protection - I just think that level is hugely exaggerated for commercial and political reasons and in the real world, helmet mandates decrease safety.

In terms of helmet efficacy what part of the picture is being missed?

Cyclists who aren't involved in collisions/accidents are entirely irrelevant when it comes to assessing the efficacy of helmets at preventing injuries in the event of collisions accidents.

The only meaningful data being missed is those cyclists for whom a helmet prevented a hospital admission and those cyclists who died pre hospital of head injuries that would not have been fatal with a helmet on. Those omissions will reduce the measured efficacy helmets so won't change the conclusions.

Rich_cb wrote:

If you want to know if helmets actually work at preventing head injuries then case control studies are the gold standard. There are no viable alternatives. Is it the only thing to consider when devising policy? No, but I've never said it was.

But population level data (that has no/minimal correlation between helmet usage and injury rates between different countries or different time periods) puts a likely range of effectiveness of helmets.

That control group studies get drastically different results suggests that the groups aren't representative (easiest explaination). (e.g. Some early studies had issues where it was proven that original data showed helmets were more effective against leg injuries than head injuries... turns out white middle/upper class riders who could afford helmets fell over cycling in parks recreationally while black poor riders who couldn't afford them were hit by cars when riding on roads for transport...)

Basically it is impossible to get precise and accurate data on how effective helmets are because ethics forbid studies that can actually measure it reliably (can't require people ride without helmets to provide control + study groups not subject to selection bias) and injury rates are low enough that we would need huge study groups over long time periods to get enough incidents for statistical significance (~11 deaths per billion km cycled according to https://discovery.ucl.ac.uk/id/eprint/10053381/1/Mindell_Cause%20of%20de...). (Of course cycle helmets aren't alone in this - a huge proportion of control group research tells us the effects of X on students, not the population as a whole because the researchers have access to uni students...)

Of course the far more important point is the data suggesting that mandating cycling helmets is bad for public health overall - Even if helmets prevent 100% of head injuries, the expected reduction in cycling rates causes more harm to average health because it reduces regular exercise when some people decide they won't cycle rather than using a helmet (even just for rental bike schemes that are no longer practical because people won't carry helmets just in case). I expect the law only applies on roads/cycle paths (ignoring the much higher risk for mountain biking because helmets are heavily used and enforcement would be difficult) and even then will be focused on major routes (where head injuries from falls are lower (smooth maintained tarmac) than minor routes (muddy, potholed tracks with overhanging branches, tree limbs etc)

The leg injuries error is simply inadequate matching of cases to controls.

I agree that a randomised controlled trial is ethically impossible and cohort studies would be prohibitively expensive. Which leaves case control studies as the best we've got.

Population studies have the same problem as a badly matched case control series. Dissimilar groups make for unreliable data.

In the UK, deaths from cycling have fallen quite rapidly over the last few decades whilst helmet wearing has increased.

Unfortunately traffic, health care, population density etc etc have all changed massively over the same period making any attempt to weedle out the effect of increased helmet wearing virtually impossible.

Rich_cb wrote:

He's refused to even read papers in the past because they referenced TR&T in the introduction!

So would I.

Tell me you don't understand research without telling me you don't understand research.

Rich_cb wrote:

High risk activities, eg time trialling, would be almost unaffected by a law change as helmets are almost ubiquitous in this area already.

Helmets were made compulsory in UK time trials - at least, events run by CTT - at the 2021 AGM. They were already compulsory in all events run by BC, including on closed circuits.

The number of people riding TTs without a helmet before the rule change was already vanishingly small. I can think of just 1 individual out of hundreds I have seen, either in person or via photos or discussions online.

What is 'high risk' and on what basis do you class UK road TTs as 'high risk'?

Rich_cb wrote:

Relatively low risk activities, with lower existing rates of helmet wearing, like cycling to the shops would be far more likely to be reduced by mandation. The risk profile of the cycling population therefore changes with high risk activities making up a higher proportion of cycling overall

How do you assess any risk profile change when TTs are such a tiny proportion of bicycle journeys?

And can you factor in that TT participation has dropped significantly since the covid lockdowns, possibly as much as 50%? In a typical year pre-2020 would have ridden 10~15 TTs alongside 250 or more commute and leisure rides. Since March 2020 I've ridden 1 time trial. I'd argue that I feel more at risk riding to work than during a 10 mile TT.

I think you've misunderstood my post.

Rich_cb wrote:

I assume you're referring to that old chestnut of 'risk compensation/homeostasis'? The Wikipedia page on the subject has a great quote. " It commands about as much credence as the flat earth hypothesis" I couldn't put it better myself.

I'm sure you've seen comments on here about people who won't ride their bike without a helmet or you know people with that attitude, that is risk compensation, they put on the helmet and then they take part in a "risky" activity that they would not do without the helmet.

Rich_cb wrote:

In order to prove that helmet are beneficial you must compare two similar populations, one helmeted, one not and then compare injury rates. We know that the cycling population, that is people who cycle, changes dramatically after compulsory helmet laws. Fewer cyclists in general but the reduction is more concentrated I some groups than others so there groups become overrepresented. As such, a comparison of injury rates pre and post law is misleading. A randomised control trial where cyclists were randomly assigned to helmet/no helmet groups and then knocked off their bicycles in various ways would likely provide that proof but would struggle to get past the ethics committee. Given that, the best, feasible, option is a case control study. Many of these have been done and whilst some, like TR&T, have been correctly criticised for methodology failures others have produced robust, reliable results that show a clear benefit from wearing helmets. Should helmets be compulsory? Definitely not. Are they beneficial? According to the best evidence available, they are. Example of case control study: https://www.researchgate.net/publication/51159929_Bicycle_helmet_wearing...

I agree with the overall picture that bike helmets do provide some protection against head injuries, but I do think their benefits are oversold and thus we end up with situations like this where politicians think that mandating helmets is a good idea.

Case control studies do have their limits. That study you link to has an interesting way to get around the issue of self selection (by being involved in an incident that needs medical attention) by comparing the cyclists with head injuries to those without head injuries. One issue with that is that it also has the effect of separating the smaller incidents from the more serious ones (e.g. slipping on some ice may lead to just a fractured wrist but being hit by a car may result in head injuries and a fractured wrist).

The main problem with the case control studies is the limited selection - it's not sampling against all cyclists, but just those cyclists involved in incidents. That can easily introduce bias and also means that they miss out the social effects of cyclists wearing helmets.

What makes me suspicious about a lot of these studies is that the results don't seem to be replicated when helmets are mandated. Unfortunately the entire (stated) point of mandating bike helmets is to reduce cyclist head injuries, but that doesn't seem to be replicated when mandates are introduced. To make matters worse, the behaviour of police around helmet mandates is often to selectively enforce the laws against the disadvantaged and minorities.

I think there is an important distinction between helmet efficacy on their own and when combined with mandatory helmet wearing, whihc skews behaviour. I don't think there are many people on here saying that it should be made mandatory to wear them in the UK.

The bit of science you can't argue with is that they do help cushion the blow IF you have an accident (even if it's faster than they are designed for). So in most instances they will reduce brain damage, so perhaps where it wouldn't have been survivable, you could live, albeit with serious concussion.

The bit where there is grey areas is the behaviour change a helmet can cause (either from car drivers, or superman syndrome for the rider), which may be where the data gets very questionable either way, especially when combined with helmet laws causing additional behaviour changes, like less riders etc.

But please don't be confused about the efficacy of having some impact protection if your head is heading towards tarmac, as thinking they are completely useless because burtthebike says so, is just as dangerous as assuming they are making you safer and therefore taking more risks. Just treat them the same as you would insurance, i.e. you still don't want to have an accident!

AndyRed3d wrote:

... Just treat them the same as you would insurance, i.e. you still don't want to have an accident!

I think that's fair enough as long as that includes "... which you will often find doesn't cover what you are expecting".

Many cyclists have a sensible (eg pretty conservative) view of the protective properties of helmets. Their manufacturers are certainly cautious in their claims! I bet most non-regular cyclists(eg. most people) have either never actually thought about this or are misinformed / wildly optimistic about their possible effects.

Anyone outside of Zwolle (or Sark etc.) proposing legislating on the subject is anti-cycling. They almost certainly haven't realised they are but it's true. It's choosing to focus on the least effective safety intervention - PPE - and not fix systemic issues or even modify the behaviour of those causing danger but instead their potential victims. (They could do both - but are they doing so...?)

Actually - it may "make the roads safer". (A little - most people drive already...) Insofar as it's an extension of what has been happening most places for decades (discourage all non-motorised road users from much of public space). Just don't look at health issues like pollution, inactivity, the effects of social exclusion etc...

chrisonatrike][quote=AndyRed3d wrote:

Many cyclists have a sensible (eg pretty conservative) view of the protective properties of helmets. .

Strangely enough, that was the subject of my MSc dissertation, and I can tell you that most cyclists overestimate the protective effect of a helmet, some ridiculously so.

Well this is a chinstroker indeed. I do like my helmet, it's very comfortable but when the time comes to replace it I don't know what to do . I think I might wear a pair of underpants on my head with two or three crumpets stuffed inside instead. Just to save money of course.

AndyRed3d wrote:

.....as thinking they are completely useless because burtthebike says so,....

I have never said so; please correct your text.