Selection criteria, data sources, reliability of data
Web Extras: published as supplied by the author
Selection criteria, data sources, reliability of data
The statistical power of detecting effects of a helmet law is proportional to the change in percentage helmet wearing (%HW). If %HW increases by 50 percentage points (e.g. from 35%-85% of cyclists) and (as suggested by case-control studies) helmets prevent 68% of head injuries (HI), HI rates should fall by 46%. Changes this large would be evident in times series. The selection criterion for inclusion was therefore an increase in %HW of at least 40 percentage points in less than 1 year.
Medline/web searches were used to locate jurisdictions that passed helmet laws and determine changes in %HW. Table B provides details of jurisdictions where the required increase was not achieved. Published data were located for all other jurisdictions from the sources listed in Table A. Public hospitals in Australia and New Zealand have computerised recording systems based on the ICD coding system, enabling cycling injuries to be readily identified and separated into head and non-head injuries. Several hundred cyclists were admitted to hospital every year in every jurisdiction (Table A), so the statistics for percent head injury are considered reliable. As shown in Figs 1-3, B, and C, they exhibit smooth trends, rather than random fluctuations. The total of 10,479 head injuries (Table A) is an order of magnitude greater than the 1,196 head injuries (4 studies), 347 brain injuries (3 studies) and 90 serious head injuries (>AIS 2; 2 studies) from which combined odds ratios were derived in the frequently-cited review of bicycle helmet efficacy.
Data for emergency room treatments in Halifax, Nova Scotia are considered more problematical, because annual totals were not available (making it difficult to separate trends from the effect of legislation) and, even before legislation, there were only 15 head injuries (and 401 non-head injuries).
Bike/motor-vehicle collisions: reduced safety in numbers?
Other successful road safety campaigns commenced about the same time as the Australian helmet laws. Pedestrian fatalities fell by 43% (Victoria), 34% (NSW) and 22% (South Australia) in the first 3 calendar years of the helmet law, compared to the 3 years before legislation. Early analyses ignored the effect of such confounding factors.[w2] However, to fully assess the effects of helmet laws, cyclist injury rates should be compared with those of other road users such as pedestrians.
In Victoria, the Transport Accident Commission (TAC) funds all medical treatment for road injury. It was therefore possible to investigate ‘Safety in Numbers’, by comparing pre and post-law rates of death or serious head injuries (DSHI) and other serious injuries (OSI) to pedestrians and cyclists in collision with motor vehicles.[w19] TAC’s definition of DSHI was: death or hospital admission for skull fracture (800, 801, 803, 804) or intracranial injury (851-854) excluding concussion. OSI was defined as ICD9 codes: 599.7, 765.1, 802, 805-839, 860-869, 870.3, 870.4, 871, 878, 885-887, 895-897, 900-902, 940-957, 994.1, 994.7.
Table C compares DSHI as a percentage of all serious injuries (%DSHI) of pedestrians and cyclists. For cyclists, %DSHI fell by 1.7 percentage points to 24.8% in the 2 years after the law. For pedestrians, the fall over the same period was actually greater – 2.5 percentage points. Helmets are popularly believed to prevent death and serious head injury, yet the fall in %DSHI was actually greater for pedestrians than cyclists.[w19]
According to the ‘Safety in Numbers’ rule, injury rates per cyclist are lower when more people cycle. So, did the reduction in cycle-use because of helmet laws increase the risk per cyclist, relative to other road users? The road safety campaign, introduced at the same time as the helmet law in Victoria, reduced pedestrian DSHI to 74% of pre-law numbers (Table C). Cyclist DSHI fell to 57% of pre-law numbers, but there were fewer cyclists – only 69% as many as before the law (Table 1). DSHI should therefore have fallen to (69%x74%)=52% of pre-law numbers for cyclists to enjoy the same injury reductions as pedestrians. The actual fall suggests cyclists did not fare as well with the helmet law as they should have done without it. Increased injury rates following helmet laws was also noted for child cyclists in NSW.
A similar phenomenon was also noted for Australia as a whole. Comparing 1988 (before any helmet law) with 1994 (when all states had enforced laws) cyclist, pedestrian and all road user deaths fell by 35%, 36% and 38% respectively; head-injury deaths fell by 30%, 38% and 42%.[w20] Thus the decline for cyclists was less than other road users.[w20] This suggests that, given the fall in cycle-use, the risk of fatal injury per cyclist increased relative to other road users.
Problems with case-control methodology
Real-life head injury data, before and after helmet laws, contradict the results of case-control studies, where cyclists who choose to wear helmets tend to have fewer head injuries than non-wearers. However, helmet wearers often differ considerably from non-wearers in riding styles and attitudes to risk.
Case-control methodology can fail. We now accept that hormone replacement therapy (HRT) does not decrease (and may increase) the risk of heart disease. Yet a review of what were considered the best quality observational studies (11 case-control studies, 16 prospective studies, 3 cross-sectional studies) concluded that HRT decreased the risk by 50%. These misleading results (and others e.g. from vitamin supplement studies[w15]) were attributed to problems distinguishing between the effect of the treatment (e.g. HRT, or helmet wearing) and the other differences (e.g. socioeconomic status, attitudes to risk) between those who chose the treatment and those who did not. When there are substantial differences (known as confounders) between treatment and non-treatment groups, case-control studies can, and have, produced misleading results, so should be treated with caution
Helmet efficacy vs injury severity
Most of the case-control studies (and data from Halifax, Nova Scotia) used emergency department records. In contrast, Table C and the line drawings show injuries severe enough to warrant hospital admission. Changes in recording or admission policies are more likely to affect minor injuries, e.g. whether concussions are admitted for observation. A published analysis of hospital data for cyclists in bike/motor-vehicle collisions, reported that numbers with head injury (including wounds or concussion) fell by 48% and 70% in the first and second years of the law, compared to 23% and 28% for admissions without head injury.[w2] It was later noted that numbers of pedestrians with concussion fell by an equally-impressive 29% and 75% in the same years. Although the effect was cited as a major success of helmet laws, it probably had nothing to do with helmets.[w19]
The 36,500 injuries in the line drawings represent what actually happened when entire populations of several million cyclists were required by law to wear helmets. They represent all cyclist admissions in all public hospitals, including the most severe injuries. The case-control studies were of cyclists receiving treatment in emergency departments; the vast majority did not require hospital admission. In the 1996 Seattle study, 73% of head injuries did not involve concussion or other brain injury.[w21] Helmets undoubtedly prevent wounds but increase the size and mass of the head. They are designed to deform on impact, but a broken, dented shell may not slide easily on surfaces. This, together with greater torque from increased head size, may increase the risk of rotational injuries, such as diffuse axonal injury. In monkeys, linear accelerations of 665-1230g caused no detectable brain damage, but, in every case, rotational accelerations (348 to 1025g) caused concussion or other brain injury, including subdural haematoma, subarachnoid haemorrhage and intracerebral petechial haemorrhage.[w22] Angular accelerations in dummies wearing bicycle helmets were measured at 12 times the 4500 rads/s2 for onset of vein rupture.[w20]
For serious injuries (>AIS 2), the review of helmet efficacy derived odds ratios from only 2 studies and a total of 90 head injuries >AIS2. With such small numbers and major differences in behaviour and attitudes to risk between helmeted and non-helmeted cyclists, case-control studies are unable to provide reliable information on injuries of this severity, and cannot address other consequences of legislation, such as risk compensation or reduced safety in numbers. The line drawings show all consequences, so represent the most reliable information on helmet laws. The lack of response in %HI to substantial increases in helmet wearing, but substantial reduction in cycle use (Table 1, Fig C) indicate that the risk per cyclist increased relative to that for other road users. Thus the disadvantages of enforced helmet laws outweigh any benefits; it would be better to concentrate on other road safety interventions with few drawbacks and large, demonstrable benefits.
Non-helmeted cyclists have lower mortality rates than non-cyclists; countries where cycling is popular but few cyclists wear helmets have much lower fatality/injury rates per km cycled than those with higher helmet wearing rates.[w17] Governments should therefore provide appropriate facilities to encourage more people to cycle for the health and environmental benefits and make cycling even safer because of ‘Safety in Numbers’.
- Provenance of article
The author is a commuter cyclist who, in association with the Bicycle Federation of Australia, investigated the effect of helmet legislation. She works at the University of New England and has a first class honours degree in mathematics and a PhD in statistics/genetics. Experience in statistical modelling, especially how biased and misleading results can be obtained from fitting incorrect or inappropriate models, motivated her to investigate the contradiction between predicted benefits from case-control studies but minimal changes in head injury rates when helmet wearing increased substantially because of legislation.
- Additional References
- Finch C, Heiman L, Neiger D. Bicycle use and helmet wearing rates in Melbourne, 1987 to 1992: the influence of the helmet wearing law. Rpt 45, Monash Univ Acc Res Centre. 1993.
- Cameron MH, Vulcan AP, Finch CF, Newstead SV. Mandatory bicycle helmet use following a decade of helmet promotion in Victoria, Australia--an evaluation. Accid Anal Prev 1994; 26: 325-37.
- Cook A, Sheikh A. Trends in serious head injuries among cyclists in England: analysis of routinely collected data. BMJ 2000; 321: 1055-.
- Povey LJ, Frith WJ, Graham PG. Cycle helmet effectiveness in New Zealand. Accid Anal Prev 1999; 31: 763-70.
- Smith N, Milthorpe F. An observational survey of law compliance and helmet wearing by bicyclists in New South Wales - 1993. Roads and Traffic Authority.
- Cameron M, Newstead S, Vulcan P, Finch C. Effects of the compulsory bicycle helmet wearing law in Victoria during its first three years. Proc Austral. Pedest & Bicyclist Safety & Travel Workshop.
- Leblanc JC, Beattie TL, Culligan C. Effect of legislation on the use of bicycle helmets. Cmaj 2002; 166: 592-5.
- Macpherson AK, To TM, Macarthur C, Chipman ML, Wright JG, Parkin PC. Impact of mandatory helmet legislation on bicycle-related head injuries in children: a population-based study. Pediatrics 2002; 110: e60.
- Heathcote B, Maisey G. Bicycle use and attitudes to the helmet wearing law. Traffic Board of Western Australia. Traffic Board of Western Australia. May 1994.
- Cycling Scotland. Cycling Scotland's helmet policy statement, the associated briefing paper and links. http://www.cyclingscotland.org/downloads.asp?fileID=29 (accessed 21 March 2005).
- Chipman ML. Butting heads over bicycle helmets. CMAJ 2002; 167: 339.
- Blacktown City Council. Blacktown Bikeplan Study, Final Report. Sydney.
- Walker M. Bicycling in Sydney: law compliance and attitudes to road safety. Proc Velo Australis, Fremantle, Western Australia.
- Webber R. Cycling in Europe. In: Shepherd R, ed. Ausbike 92. Proceedings of a national bicycle conferences, Melbourne, Australia. Melbourne: Bicycle Federation of Australia, March, 1992.
- Lawlor DA, Davey Smith G, Kundu D, Bruckdorfer KR, Ebrahim S. Those confounded vitamins: what can we learn from the differences between observational versus randomised trial evidence? Lancet 2004; 363: 1724-7.
- Parkinson GW, Hike KE. Bicycle Helmet Assessment During Well Visits Reveals Severe Shortcomings in Condition and Fit. Pediatrics 2003; 112: 320-323.
- Cyclehelmets website. Safety in numbers (graph of helmet wearing, % of trips by bicycle and fatalities/billion cycle km). http://www.cyclehelmets.org (accessed March 2005).
- Walker M. Law compliance among cyclists in New South Wales, April 1992. A third survey. Road and Traffic Authority Network Efficiency Strategy Branch.
- Robinson DL. Safety in Numbers in Australia: more walkers and bicyclists, safer walking and bicycling. Health Promotion Journal of Australia 2005; 16: 47-51.
- Curnow WJ. The Cochrane collaboration and bicycle helmets. Acc Anal Prev 2005; 37: 569-73.
- Thompson DC, Rivara FP, Thompson RS. Effectiveness of bicycle safety helmets in preventing head injuries. A case-control study. JAMA 1996;276(24):1968-73.
- Gennarelli TA, Thibault LE, Ommaya AK. Pathophysiological Responses to Rotational and Translational Accelerations of the Head, SAE Paper No. 720970. 16th Stapp Car Crash Conf.; 1972. Society of Automotive Engineers, 1972
- King M, Fraine G. Bicycle helmet legislation and enforcement in Queensland 1991-3: effects on helmet wearing and crashes. Road User Behaviour Section, Queensland Transport. June 1993.
- Robinson DL. Confusing trends with the effect of helmet laws. Pediatrics 2003; Post-publication Peer Review of Pediatrics 2002; 110: e60.
- CIHI. Injury Hospitalizations (includes 2000-01 and 2001-02 data). Canadian Institute for Health Information. March 2003.
- Abularrage JJ, DeLuca AJ, Abularrage CJ. Effect of education and legislation on bicycle helmet use in a multiracial population. Arch Pediatr Adolesc Med 1997; 151: 41-4.
- Borglund ST, Hayes JS, Eckes JM. Florida's bicycle helmet law and a bicycle safety educational program: did they help? J Emerg Nurs 1999; 25: 496-500.
- Kanny D, Schieber RA, Pryor V, Kresnow MJ. Effectiveness of a state law mandating use of bicycle helmets among children: an observational evaluation. Am J Epidemiol 2001; 154: 1072-6.
- Liller KD, Nearns J, Cabrera M, Joly B, Noland V, McDermott R. Children's bicycle helmet use and injuries in Hillsborough County, Florida before and after helmet legislation. Inj Prev 2003; 9: 177-9.
- Rodgers GB. Effects of state helmet laws on bicycle helmet use by children and adolescents. Inj Prev 2002; 8: 42-6.
Posted as supplied by the author
Table A Dates of helmet laws, sources and descriptions of injury data, helmet wearing surveys and other road safety initiatives introduced at the same time as helmet laws
Jurisdiction |
Description of head and other injuries (HI and OI), surveys and other safety initiatives |
South Australia (SA); Marshall and White[6] Helmet law 1/7/91 |
HI: (Fig 1) = hospital
admission with principal diagnosis of skull fracture (800-802.1, 802.4-804.06)
intracranial injury (850-854.16) or wound to the head (870-871, 873-873.39,
873.42, 873.52); concussions (850) were tabulated separately. OI = other
admissions. |
Western Australia (WA); Hendrie et al.[7] Helmet law 1/1/92 |
HI: (Fig 2) = all hospital
admissions with ICD9-CM injuries coded to the AIS body region of head (1998
onwards). For 1979-87, HI = admission with skull fracture (800-804), intracranial
injury (850-854), open wound to head (873.0-873.1), injury to blood vessels
of head or neck (900) or injury to optic or other cranial nerves (950-951).
To 1978, HI = IDC8 codes 800-804, 850-854, 873, 904, 950-951. OI = all other
admissions. |
New Zealand; Robinson[10] Povey et al.,[w4] Helmet law 1/1/94 |
HI: (Fig 3) = admissions
with skull fracture (800–804), concussion or intracranial injury (850–855);
OI = admission with limb fracture (810–829); excludes accidents involving
motor vehicles. |
New South Wales (NSW); Robinson[10] Adult law 1/1/91; <16 years 1/7/91 |
HI: (Fig B) = hospital
admissions with head injury (NSW Health classification); OI = admission with
other injury; cyclists with both head and other injuries were included in
both categories. |
Victoria; |
HI: (Fig C) = any hospital
admission with skull fracture (800-803), concussion/intracranial injury (850-854),
open wound of ear or head (872, 873.0, 873.1, 837.8, 873.9). |
Halifax, Nova Scotia, Canada; Leblanc et al.[w7] Law 1/7/97 |
HI: = concussions, lacerations,
dental injuries or other head injuries causing death or requiring follow-up,
observation in the emergency department, admission to hospital or transfer
to another health facility. OI = all treatments at the health centre except
those listed above. |
See text and Table B for formal selection criteria and jurisdictions where helmet wearing failed to increase by at least 40 percentage points.
Posted as supplied by the author
Table B Jurisdictions not included because percent helmet wearing (%HW) increased by less than 40 percentage points
Queensland, Australia |
All-age helmet law was introduced 1/7/91, but not enforced for 18 months; %HW increased temporarily, then declined until penalties imposed.[w23] |
Ontario, Canada |
Non-enforced law for child cyclists introduced October 1995. HW increased from 45% pre-law in 1995 to 68% in 1996 and 1997, then returned to pre-law levels.[w10] %HI trended down smoothly from 94/95 to 97/98; the largest fall (from 33.9% in 96/97 to 28.5% 97/98) was when %HW was static or declining.[w24] Despite the decline in %HW from 1997 onwards, %HI continued to decline, falling by 26% from 1997/98 to 2001/02, non-HI by 12%.[w25] |
British Columbia, Canada |
All-age helmet law introduced in September 1996. %HW increased from 51%-74% in metropolitan areas, 32%-58% elsewhere. %HW of 6-15 years olds increased from 35%-61%. Declining trend in children’s %HI; the largest fall (7.4 percentage points, from 39.9%-32.5%) was from 1994/95 to 1995/96, a year before legislation.[w24] |
New York State |
Non-enforced law introduced June 1994. In Brooklyn %HW decreased from 5.6% to 4.2% with legislation. In Queens, education plus legislation increased %HW from 4.7% to 13.9%.[w26] |
Florida |
Children’s law from 1997. Legislation increased %HW of 7-12 year old boys from 5.6% to 20.8% in Brownard.[w27] %HW in a survey of elementary school children was 79% in counties with helmet laws, compared to 33% in non-law counties.[w28] In Hillsborough county, %HW increased from 14% in 1996 to 57%, 67%, 56% and 50% in 1997, 1998, 1999 and 2000. No head injury data, but number of child cyclists (0-14 years) injured in motor vehicle accidents declined (135, 125, 140, 125, 100, 77, 101, 87 in 1993-2000)[w29]. No mention of any attempts to estimate changes in cycle-use, so the fall in injuries may have been due to reduced cycling. |
US generally |
State helmet laws (for children under 16) increased average probability of helmet use by 18.4%.[w30] |
Posted as supplied by the author
Table C TAC (Transport Accident Commission) data for average numbers of deaths and serious head injuries (DSHI) and all serious injuries (ASI) per year in Victoria (from Robinson[w19])
Injuries due to collisions with motor vehicles (average number per year) |
Cyclists |
Pedestrians | ||||
DSHI* |
ASI |
%DSHI |
DSHI |
ASI |
%DSHI | |
Pre-law (1988/90) |
72.5 |
274.0 |
26.5 |
285.5 |
828.0 |
34.5 |
Post-law (1990/92) |
41.0 |
165.0 |
24.8 |
211.0 |
660.0 |
32.0 |
Post-law as % of pre-law |
56.6 |
60.2 |
93.6 |
73.9 |
79.7 |
92.7 |
Adjusted for 30% fall in cycling |
80.8 |
86.0 |
*DSHI as defined by TAC (death or hospital admission with skull fracture
(800, 801, 803, 804) or intracranial injury excluding concussion (851-854).
OSI = other serious injury (599.7, 765.1, 802, 805-839, 860-869, 870.3, 870.4,
871, 878, 885-887, 895-897, 900-902, 940-957, 994.1, 994.7). ASI = DSHI +
OSI.
![]() |
![]() |
![]() |