swansonj
Guru
I read the paper on the plane last night on the way back from a meeting drafting a European safety standard. So, like you, a bit of a busman's holiday.
I am not bothered by the conflict of interest (mentioned by some people). The author gives their affiliation as the academic institution then declares that they are also CTO of the helmet company. That suggests either they are an academic who got involved in a spinoff company, or a product designer interested enough in how their products work to get involved in academia. Either way that sounds like something we should support. The interest was declared which is the important thing (own declaration of interest: I am currently auditing how well the conflict of interest policy works for the journal I am on the editorial board of, so it's a subject fresh on my mind).
There are two obvious big question marks about the work. One that you also honed in on is the selection criteria for the cases investigated. None are stated (editorial board hat on again: how did the refereeing process allow them to get away with that?). It seems highly likely the cases were chosen, subconsciously if not consciously, to be good test beds for the benefits of helmets. So the conclusion should perhaps be: helmets can have benefits in those situations where helmets are more likely to have benefits. Still a valid conclusion but not quite as universal, and not something anyone here would disagree with.
The other problem is that there's no validation of the performance of the finite element model of the helmet. You and I had an unfortunately antagonistic exchange recently about helmets cracking/splitting. (I still think we were actually agreed, even if we couldn't quite bring ourselves to say it, that a split helmet may well be, but is not necessarily, evidence of failure to perform as intended.) The model used in this paper certainly allows for the possibility of splitting (e.g. the shape of the stress-strain curves including the sheer one in figure A1). But what I don't know is whether they are realistically modelling real-life helmets - manufacturing defects, stress raisers, crack propagation, material inhomogeneities and all - or some highly idealised version. That is rather critical - if their model doesn't properly include the factor that we think may be the commonest reason for helmets failing to perform as expected in real life, well, reader, draw your own conclusion... They are also rather light on detail on how they model the evolution of the impact - if there was any info on the time step size, for instance, I missed it.
One could also point out that their assumption of a simple coefficient of friction does not allow for snagging, and they must have made some assumptions, which again I missed if they discussed them, about the performance of the straps.
HOWEVER, although all these factors argue for caution in interpreting the results, and although it is worrying they did not discuss these more themselves, I believe we should welcome papers like this. We have a situation where the epidemiological evidence is clear that helmets do not work. Yet most people approaching the subject have an intuitive sense that helmets should work. I don't think we can yet give a fully satisfactory explanation of why they don't. Part of that explanation would be about risk transfer etc, but part has to be about the helmet properties. The more we understand the actual helmet behaviour the better, because I don't believe we will ever stop the instinctive helmet advocates until we can say, not just that they are wrong in the faith they put in helmets, but why they are wrong.
Last edited:
