Nigel Mills

http://www.perg.bham.ac.uk/pdf/IRCOBI03.pdf Dr. Nigel Mills is an acknowledged expert on the mechanical properties of helmets and their materials, appearing as an expert witness in numerous court cases. He is a helmet advocate. An example of the sort of work he does can be found here. The paper is in part an update of the authors' earlier work on helmet construction and in part a kick back at some comments made regarding helmet efficacy and the possible reasons for the failure of helmet laws to deliver measurable benefit. As is often the case in helmet advocate vs. sceptic exchanges, some of the criticisms are aimed at a caricature of the real argument, presenting a rather easier target than perhaps might be the case for the argument actually advanced. The technical credibility of the paper is inclined to be undermined by this obviously polemical and somewhat personalised undertone, which may account for its not having been published in any peer-reviewed journals despite the authors' academic credibility.

Premises
The authors begin with the following statement: THIS PAPER DEALS WITH HELMET DESIGN and performance issues, but not the interpretation of trends in bicycle accident statistics, which are reviewed by Thompson et al (2003). The 2003 Cochrane review is cited as the authoritative (indeed, the sole) source for data on the efficacy of helmets, despite the published criticisms which form part of that review. Put briefly, the reviewers' own work forms the bulk of the selected papers, despite their claims of efficacy being consistently higher than most other research; valid published research possibly contradicting their findings is excluded; indeed only one study methodology is considered; contradictory opinions are summarily dismissed without evidential basis. For a fuller critique of the 2003 Cochrane review see this commentary or Curnow (2005)[1]. As the authors acknowledge, there is no shortage of pro-helmet research to choose from, so this particular selection might been taken as indicative of bias. The paper continues:

In spite of the large number of papers written by medics, showing the efficiency of bicycle helmets in reducing head injuries, there are several campaigners against the compulsory use of cycle helmets, who argue that helmets are ineffective.

There are two problems with this statement:
 * The evidence in favour of helmets is almost entirely of one type: observational studies. Recent experience regarding the supposed link between hormone replacement therapy and reduced levels of coronary heart disease in women has highlighted the way that a common flaw which is duplicated in all studies, in this case self-selection bias (a criticism equally applicable to helmet studies) means that it is perfectly possible for a large number of studies all to be wrong[2], [3], to the point of inviting the question of whether epidemiology is sustainable at all.
 * Campaigners do not deny that helmets are effective - they dispute the degree of efficacy against serious and fatal injuries, and they note that the major impact of helmet laws has been to deter cycling, not to reduce cyclists' head injuries in any measurable degree; this much is documented fact (e.g. Robinson, 2004[4]).

The dispute between the authors and the sceptics appears to exist in part because they are arguing from different premises, and the paper appears to mix ideological and technical arguments.

Criticisms of Detractors
Mills & Gilchrist focus mainly, here, on the work of Avery Burdett of the Ontario Coalition for Better Cycling (OCBC), and his helmet faq. This document was one of the first to systematically set out the sceptic stall, and has been much linked and discussed. The authors rightly choose to test a number of the possible reasons advanced by Burdett to account for the observed failure of helmet laws in practice, but they personalise the issue rather more than seems appropriate in a scientific paper. What is not made clear is that just as it is possible for all the observational studies to be wrong, sharing as they do a common inherent flaw, so it is equally possible for any or all the possible suggested mechanisms for failure of helmet laws to be wrong, without changing the fundamental fact that the helmet laws have failed. In addressing only the suggested mechanisms, without suggesting any of their own, the authors move the state of knowledge forward only slightly.

Rotational Forces
The authors address the issue of rotational forces: Curnow (2003) argued that the design of helmets reflects a discredited theory of brain injury (that injuries are caused by peak linear acceleration). He is correct in that the standards (EN 1078, Snell, CPSC) do not have oblique impact tests, in which the headform rotational acceleration is measured. However the premise, that the majority of bicyclists’ head injuries are due to rotational acceleration, is not proven. He argues that epidemiological studies do not distinguish between skull fractures and injuries caused by angular acceleration; consequently they do not demonstrate the efficiency of helmets in preventing head injury. Franklin (2000) also criticised the lack of oblique impact tests in helmet standards, writing ‘there does not appear to be research evidence that cycle helmets are effective in mitigating angular impacts. and A BMJ booklet (1999) states: ‘Cycle helmets are designed to protect the head during a low speed impact, e.g. 20 km/h, such as would occur in a fall to the ground from a bicycle. It is likely that most of the 183 UK cyclist deaths recorded in 1997 would have resulted form velocities and energies in excess of the average cycle helmet’s ability to prevent the tragic outcome.’ However, there is no research evidence to support the 2nd statement; we do not know how the fraction of the cyclists who wore helmets, or the crash circumstances. It is possible to check the validity of most of these statements by carrying out experiments. It is possible to measure experimentally helmets' ability to mitigate rotational impacts, but to assess the extent to which these impacts may be translated into real world injuries and deaths is rather more problematic; that would also need to be informed by careful diagnostic scanning and post-mortem examination. It is also hard to see how any experiment could assess the extent to which helmets might mitigate real cyclists injuries of the most serious kinds, because all cycle crashes are unique. While it is true that the proportion of seriously and fatally injured cyclists in the UK who were wearing helmets is unknown (and in some cases irrelevant, since the largest single cause of cyclist death is reportedly crushing by left-turning goods vehicles), it is a matter of record that in Australia today the proportion of helmeted cyclists suffering serious or fatal head injuries is the same as the proportion of cyclists who wear helmets. A large volume of real-world time-series data shows no correlation between even steep rises in helmet use, and reductions in recorded (i.e.serious or fatal) head injury. In their tests the authors identify significant differences in rotational forces in different helmets; they suggest the addition of a test of rotational impacts into the EN 1078 helmet standard. This may be unrealistically optimistic: EN 1078 is substantially weaker than earlier standards;

Linear Impacts
The authors suggest that Burdett makes the following claims:

They set out to address these. It is not clear that they address (a); further development of the models is clearly required if there is to be a realistic attempt to model the human head more closely. Claim b is not stated by Burdett, although he acknowledges that his text could be construed in that way. The authors identify steps which have been taken by manufacturers to mitigate this issue (which is, therefore, clearly a real one). Claim c is a rather selective representation of the FAQ, reproduced as if quoted in full. The full text is:

The text in bold has been omitted; it is not clear why the authors have not used ellipses in the usual way to denote edited text.

Authors' Conclusions
The authors' conclusions are as follows: Most criticisms of current bicycle helmet designs are not valid: the lack of a scalp on test headforms does not lead to inappropriate designs; the foams used in helmets with large vent areas do not cause excessive pressure on the head; there is a gradual rise in peak impact force with impact kinetic energy, not a just sub-lethal level for minor impacts. Current designs provide adequate protection for oblique impacts on to a road surface, in terms of the peak linear and rotational head accelerations. Helmet designs, with long extensions at the front and rear, do not appear to cause excessive rotational head acceleration. However the coverage at the side of the head is felt to be inadequate. It is recommended that an oblique impact test, using a headform with realistic scalp and wig, is included in EN 1078 with measurements of rotational acceleration. First and foremost, the criticisms identified are not "most criticisms" of helmet designs. The fundamental criticism of helmet design is not addressed: that it meets a standard in terms of impact which is unlikely to cause serious or fatal injury in the first place, and that these standards are actually declining as the evidence for lack of efficacy against serious and fatal injury mounts[5]. Neither are the claims as addressed actually the claims as made. The first conclusion is problematic: the forces are clearly different when a scalp and/or wig is used, and to state that this does not result in inappropriate designs is therefore to rely on a coincidence. The second is interesting, but does not fully quantify the way in which large impact forces might be spread in helmets with very deep ridges as are the current vogue. The authors identify a number of problems with modelling kerbstone and hemispherical impacts. It is clear that, as they state, it is not possible to have an optimum design for all impacts. The third claim, "Current designs provide adequate protection for oblique impacts on to a road surface, in terms of the peak linear and rotational head accelerations", seems to be founded on the assumption that the standards represent adequate protection, but the recommendation that an oblique impact test with wig and/or scalp be included is a recognition that this may not be the case. Finally, the analysis assumes that the helmet is correctly fitted and secured. Evidence suggests that this is at best a minority case.

Summary
This paper makes important comments regarding the modelling of impacts with current helmets. It opens an interesting debate. It does not amount to a rebuttal of the idea that helmets do not, in practice, prevent measurable numbers of serious and fatal injuries. In as much as the claims of helmet sceptics are addressed, they are somewhat misrepresented. Some of the authors' base assumptions are open to question. The tone and language of the paper displays a degree of bias: this is unfortunate in that it detracts from the technical merit of the content.

References
 * [1] The Cochrane Collaboration and bicycle helmets, Curnow, W.J. Accident Analysis and Prevention, 2005; 37 (3) 569-574.
 * [2] Hormone replacement therapy and coronary heart disease, Pettiti D, International Journal of Epidemiology, 2004;33:461-463
 * [3] The hormone replacement - coronary heart disease conundrum: is this the death of observational epidemiology? Lawlor DA, Smith GD & Ebrahim S, International Journal of Epidemiology, 2004;33:464-467
 * [4] Costs and benefits of the NZ helmet law, 2001, Robinson DL.
 * [5] Helmet Standards and Capabilities, 2004, Walker B