Why you should dismiss the CARRS-Q bicycle helmet research
CARRS-Q’s Analysis of Police & Hospital Data
This is the final post in a series (Read Part One, Part Two, Part Three, Part Four, Part Five, Part Six) looking at the non-peer reviewed CARRS-Q publication entitled “Bicycle Helmet Research“, published in 2011 and widely relied upon to support mandatory helmet laws for cyclists. This publication was commissioned by the State Government to support its policy of mandatory helmet laws in Queensland, Australia, in the face of criticisms of its lack of support for Brisbane’s public bike hire scheme.
In addition to selective reliance pre-existing analyses of the efficacy of helmet wearing discussed in previous posts, CARRS-Q also sought to conduct its own analysis of data. Data was sourced from police records and hospital data provided by the State Government’s Department of Transport and Main Roads.
The data was restricted to the years 1993-2008, and so cannot give any comparisons of injuries before and after mandatory helmet laws were introduced in Queensland in 1991. The limitations to this data were recognised by CARRS-Q as follows:
“Reliable earlier data were not available and so all of the crash data analyses relate to after the helmet law and penalty were introduced.” (CARRS-Q, 2011, page 25)
“Research indicates there is significant underreporting of bicycle crashes … International figures suggest between 11% and 13%of bicycle crashes are reported in police statistics. The reporting rates are even lower in Australia, with analysis of Western Australia data suggesting that only 3.5% are reported. Actual reporting rates may be even lower, as these rates are calculated using hospital data.” (CARRS-Q, 2011, page 25)
“There are also suggestions that police-reported data will not only underestimate the magnitude of the cyclist injury problem, but it is likely to be skewed to serious injury crashes and those that involve motor vehicles.” (CARRS-Q, 2011, page 25)
“Within the hospital data, only the primary reason for hospitalisation is reported.” (CARRS-Q, 2011, page 25)
“[C]rashes that occur on private roads or on public paths that are not within the road reserve (e.g. segregated bicycle paths through public paths) are excluded from official records.” (CARRS-Q, 2011, page 25)
“[A]ny changes in the patterns of crash severity, hospitalisation rates and other non-fatal data from 2006 onward may reflect changes in the reporting system, rather than changes in any actual injury outcomes.” (CARRS-Q, 2011, page 25)
Data analysis by CARRS-Q was restricted to after helmet laws were introduced, which is of some concern given that the Victorian research discussed in a previous post showed significant anomalies between pre-helmet law injury rates and post-helmet law injury rates that could not be attributed to helmet use. Additionally, the data used probably only represents somewhere between 3.5% to 11% (possibly even lower) of all bicycle crashes, and is likely to only include crashes that occur on roads and only include the most serious injuries.
Limiting the analysis only to a period after helmet laws were introduced means that:
“… the data analyses cannot provide information on whether the introduction of helmet legislation led to a reduction in the percentage of injured cyclists who had head injuries.” (CARRS-Q, 2011, page 39)
Similar to some of the research discussed in a previous post, CARRS-Q concludes that the proportion of head injuries and facial injuries was lower for those riders who were wearing a helmet. However, also consistent with that previous research, the CARRS-Q analysis of the data noted that helmet wearing rates were lower for children and youths:
“Recorded helmet use rates are highest (approximately 80%) among injured cyclists aged 25 years or older. Helmet use trends are the lower for younger riders, with similar rates for primary and secondary school aged children (5-11 group and 12-17 group).” (CARRS-Q, 2011, page 39)
“Wearing rates were lowest by cyclists aged under 20, with a steady increase with age.” (CARRS-Q, 2011, page 40)
As demonstrated by the research discussed in a previous post, these age groups are much more likely to have a crash. CARRS-Q also found that:
“[b]oth helmeted and non-helmeted cyclists aged 0-4 demonstrated a high proportion of head injuries, while cyclists without helmets also had a higher proportion of lower limb injuries in this age group.” (CARRS-Q, 2011, page 39)
“Most cyclists killed are males in collisions with motor vehicles with 15-29 year olds being most commonly involved.” (CARRS-Q, 2011, pages 39 and 40)
Limiting the data only to a period when helmets were required by law, and only to roads where helmets must be worn, means that any non-helmeted crash victims included in the survey fall into a very specific category: law-breakers. As well as the likelihood that non-helmeted cyclists belong to an age group more likely to have a serious crash, it can also be assumed that any adults riding without a helmet against the law may also be risk takers (the ‘risk’ is a hefty fine – up to $150 in some states in Australia).
CARRS-Q also concluded that injury severity was lower for cyclists wearing a helmet. However limiting the analysis only to the very small percentage of the most serious injuries will give a false representation of any association between helmet use and crash severity.
CARRS-Q also analysed Australian bicycle fatality data it obtained from:
“ … published reports and on relevant websites. Primary references will include the Australian Transport Safety Bureau reports, “Cycle safety: A national perspective.” (2004), “Deaths of cyclists due to road crashes” (2006) and “Road Deaths Australia 2008: A statistical summary” (2009). Some of the detailed data (such as helmet wearing) is dated and any fatalities occurring in off-road riding are not included.” (CARRS-Q, 2011, page 34)
From this data CARRS-Q concluded that:
“… the number of cyclists killed per year has approximately halved since 1991 (when helmet legislation was introduced in most jurisdictions.” (CARRS-Q, 2011, page 35)
Here is the graph used by CARRS-Q, generated from the data analysed, to justify this claim:
(Source: CARRS-Q, 2011, page 35)
Despite CARRS-Q’s claims, what the above graph clearly shows is that there was no clear drop in fatalities at the time helmet laws were introduced, followed by a constant plateau as helmet laws remained constant, which would be the expected result if helmet laws had any effect on injuries and no effect on cycling participation. In fact, what the graph shows is that while there was an initial drop in 1992, the number of cyclist fatalities in 1993 actually exceeded the number of fatalities immediately before the introduction of the helmet laws. Coincidentally, the penalty for not wearing a helmet was introduced in 1993. What follows is an unsteady but gradual decline in fatalities. This graph more appropriately supports the contention that helmet laws have simply continued to reduce the number of cyclists since their introduction.
CARRS-Q admits that:
“First and foremost in a discussion of the effectiveness of bicycle helmets, it should be noted that the datasets from which these reports are created do not record the helmet use of injured persons. Direct comparisons of the outcomes of injured persons wearing or not wearing helmets are therefore not possible.” (CARRS-Q, 2011, page 36)
So what does the CARRS-Q analyses of existing data, and all the previous research discussed in previous posts actually prove?
CARRS-Q makes the headline-grabbing claim that current helmet wearing rates are halving the number of cyclists’ head injuries in Queensland. This claim makes a number of rather dubious assumptions. In its conclusion CARRS-Q says that, because only crashes can be examined and there is no way to randomly assign helmeted and unhelmeted participants to certain road conditions:
“ … it can always be argued that the riders who wear helmets are somehow intrinsically safer riders than those who do not, or that those who ride on footpaths are somehow different than those who ride on the road or that people who chose to cycle would have been healthier anyway.” (CARRS-Q, 2011, page 41)
The actual conclusions that can be drawn from the above fall significantly short of CARRS-Q’s bold claim.
There are some consistent findings in the research discussed above and in previous posts, including the peer-reviewed research CARRS-Q chose to ignore. Such as:
- Children and youths are most likely to suffer a cycling related crash that is serious enough, or possibly serious enough, to be reported.
- Children who do crash are most likely to suffer a minor head laceration that is superficial (i.e. lots of blood, nothing seriously hurt, the type of inury that an adult would have avoided by instinct or experience, for example holding their hands out break their fall).
- Children, especially teenagers, are also the most likely not to wear a helmet.
- Children, especially teenagers, are also most likely to give up cycling if helmet wearing is mandatory.
- Helmet wearing significantly reduces the number of people choosing to cycle across all age groups.
- Significantly increased numbers of people cycling make cycling safer for those that cycle, however that increased safety is likely to diminish if cycling becomes the most popular form of road transport.
- The biggest killer of cyclists is motor vehicles, specifically articulated or rigid trucks.
- Cycling is inherently a safe means of transport.
- In social economic cost terms, increasing cycling will result in savings across the areas of health, transport infrastructure, pollution and productivity that outweigh any costs that can be associated with cycling-related injury.
- The possible decreased likelihood of a serious head injury that can be attributed to wearing a helmet, is probably negated by the likelihood of a neck injury.
These findings boil down to a couple of salient points:
First, the difference in the effect of helmet laws on children and adults means that any non-age group specific assessment of helmet wearing rates for crash victims is flawed. There are significant differences in cyclist injury rates for children compared to adults. Children cyclists are much more likely to suffer injury:
(Source: CARRS-Q, 2011, page 37)
New research by CARRS-Q itself, confirms that teenagers and other cyclists who fail to wear helmets (in jurisdictions where they are required to wear them by law) are much greater risk takers.
Any change in the age demographic of riders over a period of time is going to significantly change the injury statistics over that same period of time. Unlike the efficacy of helmets which should be clear between a time when helmets were not mandatory and helmets were mandatory (e.g. pre- and post-introduction of mandatory helmet laws), changes in the age demographic of cyclists will evoke a gradual change in statistics, such as can be seen in the graph discussed above, entitled Figure 16: ‘Numbers of cyclist (rider and pillion) fatalities in Australia and Queensland from 1991 to 2009’.
Secondly, on a community scale, the economic advantages from increased cycling far outweigh the possible disadvantages of reduced helmet use.
Thirdly, there is no clear, unequivocal evidence that wearing a bicycle helmet provides any significant protection from serious or life-threatening injuries including neck injuries. It is only clear that helmets protect from minor injuries to the head such as superficial lacerations (cuts and bruises etc.)
Fourthly, the lack of real, unequivocal evidence allows political agendas to pick and choose certain findings, so that bad policy decisions can appear to be supported by ‘science’.
This concludes our comprehensive assessment of the CARRS-Q document used by the former Queensland Government to boost support for mandatory bicycle helmet laws.