"Spinning" helmet law statistics
The examples below show how information seems to have been distorted to create a favourable "spin" on reality and avoid embarrassment about the true effects of bicycle helmet laws.
1) "Spin" injury reductions from reduced cycling into a "benefit" of helmet laws!
Official reports showed 43% and 46% reductions in teenage cycling in the first and second years of the helmet law in Melbourne, Victoria, plus reductions of 3% and 11% for children up to 11 years 6. Despite a bicycle rally at one site in the second year, numbers of adult cyclists were also 29% and 5% below pre-law levels.
With fewer cyclists (especially teenagers who generally have a higher risk of injury than adults) there should have been fewer cycling injuries, a fact demonstrated by the large reductions in non-head injuries (see Fig 1 below). Yet a newspaper article quoting Max Cameron from the Monash University Accident Research Centre (MUARC) implied that the entire 40% reduction in head injuries was due to helmet wearing 7.
With such a large and obvious reduction in non-head injuries, most likely because of reduced cycling, it seems highly inappropriate to claim that the entire 40% reduction in head injuries was due to helmets.
2) "Spin" injury reductions from improved road safety into a "benefit" of helmet laws!
Campaigns against speeding and drink-driving commenced about the same time as Victoria's bicycle helmet law. Pedestrian deaths fell from 159 (1989) to 93 (in 1990, the year the helmet law was introduced) 8. The British Medical Journal reported that road accident costs were reduced by an estimated 100 million dollars for an outlay of 2.5 million 9. Table 1 (below in section 2, Details) shows that deaths and serious head injuries (DSHI) to pedestrians fell by 26% and other serious injuries (OSI) fell by 17%.
Accounting for improved pedestrian safety, the reduction in cyclist DSHI was just 23%, which is less than the 36% reduction in numbers of cyclists counted in the 1991 survey 1. This suggests the risk of injury per cyclist increased. Helmets do not prevent other serious injuries, yet the reduction in OSI (26% after accounting for improved pedestrian safety) was better than the reduction in deaths and serious head injuries.
The "spin" that the 40% reduction in head injuries was due to helmets bears little or no relationship to the reality that the entire effect can be explained by safer roads and reduced cycling.
3) Hide key information on numbers of adults counted in the pre-law survey
In Melbourne, Monash University Accident Research Centre (MUARC) conducted three surveys of cycling in May 1990 (before the law) and post-law in May 1991 and May 1992. In 1990, 1567 adult cyclists were counted but only 1106 (29% less) in 1991. In 1992, when counts were inflated by a bicycle rally passing through one site, there were 1484 adult cyclists (still 5% less than 1990).
Numbers counted are a good indicator of cycle use, yet MUARC ignored the 1990 adult count and used a much earlier survey at a different time of year (Dec/Jan 1987/88, that counted only 1079 adult cyclists) as the pre-law "baseline". So instead of reductions from 29% to 5% in adult cycling, based on actual numbers counted, MUARC claimed that there was an "estimated increase in adult use of 44%" 6.
A later report by CARRS-Q made an even more dramatic claim: "In Melbourne adult cyclist numbers doubled after the helmet legislation was introduced". MUARC researchers muddied the waters even further by claiming: "Because the 1990 survey did not cover adult bicyclists, it was not possible to fully examine the change in their bicycle use" 6.
Pretending there was no information on adult cyclists or that adult cycle use could not have been predicted from the numbers of adults counted in 1990, and so claim an "increase" in adult cycling from a non-comparable "baseline", is another example of misleading and deceptive "spin".
4) Ignore obvious warning signs that helmets couldn't possibly be as effective as claimed
MUARC researchers noted that their models predicted head injury rates would fall to zero before helmet wearing reached 100%! 2 This impossible result should have warned them to look for other factors affecting head injury rates ... e.g. that head injury percentages for cyclists and pedestrians show remarkably similar trends (see graph below in section 4, Details).
If they had looked at the data for pedestrians, MUARC researchers would have seen that numbers of pedestrians with concussion fell by 29% and 75% respectively in the first and second years of the helmet law 11. Investigating these factors might have led to more balanced reporting of the effect of helmet laws, instead of "spin" that failed to distinguish between any real benefits of helmet laws and injury reductions due to improved road safety and reduced cycling. And then even more "spin" to hide information on numbers of adults counted in the main pre-law survey and avoid admitting that injury rates per cyclist had actually increased.
5) Ignore risk compensation and safety in numbers
When children ran an obstacle course wearing a helmet and wrist guards, tripping, falling and bumping into things increased by 51% 12. There would be little point of making helmets compulsory if the increased helmet wearing encouraged cyclists or drivers 13 to take more risks, resulting in increased injuries per cyclist, counteracting any benefits of helmets.
Similarly, increased injury rates because of reduced safety in numbers would also be counterproductive 4, 14. Several jurisdictions introduced helmet laws but few measured cycle use reliably enough to compare the change in injury rates with the change in cycling. The information that is available suggests that helmet laws increased injury rates.
When Alberta, Canada made helmets mandatory for children, numbers of child cyclists halved but injuries increased, suggesting the risk of injury more than doubled 15. In NSW, head injuries to child cyclists fell by 29% but numbers counted in observational surveys fell by 36% and 44%% in the first and second years of the helmet law, suggesting an 11-27% increase in the risk of head injury per child cyclist 2.
In Victoria, child cycling was estimated to have fallen by 33% and 36% in the first and second years of the helmet law but child cycling injuries fell by only 22% and 25% respectively, suggesting that the risk of injury increased by 16-17% 1.
The apparent increase in the risk of injury per cyclist suggests that, even ignoring the lost health and environmental benefits of reduced cycling, helmet laws were detrimental to public health because of risk compensation and reduced safety in numbers.
6) Ignore the health and environmental costs of reduced cycling and reduced safety in numbers
After allowing for injury costs, an Australian government report estimated the net health benefits of cycling at 0.75 cents per kilometre in 2013. For transport trips, there were additional savings per km in vehicle operating costs (35 cents), reduced congestion (20.7 cents), infrastructure (5.2 cents) and environmental benefits (5.9 cents) 16.
In 1985/86, bicycle travel accounted for 3.9% of all trips in Australia, including 1.6% in Sydney and 5.0% in the rest of NSW 17. The same amount of cycling today (2.24 km per person per week, about 85% for transport purposes) would generate estimated health, environmental and other benefits of $2.4 billion per year.
As noted above, in Victoria and after accounting for the improvement in pedestrian injuries, the fall in head injuries to cyclists was less than the fall in cycling, implying that injuries per cyclist increased compared to what would have been expected without the law. The lost health and environmental benefits from even a 25% fall in cycling – over half a billion dollars per year, according to the above estimates – is a totally unacceptable price to pay for a law that did not achieve its stated objective of making cycling safer.
7) Why would non-enforced laws with no long-term effect on helmet wearing be beneficial?
In Ontario, Canada, the helmet law for children was not enforced. After a temporary increase, helmet wearing fell back to pre-law levels by 1999. In contrast, head injury rates trended downward (see graph below in section 7, Details) and were much lower in 2001/02 than in the peak helmet-wearing years of 1996-97.
Yet a paper discussing the effects of helmet laws, published in 2003, omitted the key data for 1999 that showed helmet wearing had returned to pre-law levels, instead reporting helmet wearing rates only until 1997 18.
Omitting this information misled people into believing the law was effective, yet had not discouraged cycling. It's hard to imagine how a non-enforced law with no long-term effect on helmet wearing could possibly have been effective. Lying by omission, to hide the ineffectiveness of a law, seems like another unacceptable form of "spin".
1) 'Spin' injury reductions from reduced cycling as a "benefit" of helmet laws!
A 1996 press article quotes Mr Max Cameron, a senior researcher at Monash University Accident Research Centre (MUARC), that his "studies of bicycle-related hospital admissions showed conclusively that helmets worked. For four consecutive years after helmets became compulsory, we had a 40% drop in head injuries over what we had before".
Fig 1 (above, from Carr et al. 19) shows a substantial fall in non-head injuries as well as the 40% reduction in head injuries. This implies that much of the so-called "benefit" must have been due to reduced cycling, not helmets.
By not mentioning the fall in non-head injuries, people were misled into thinking that the entire effect was due to helmets when this was obviously not the case. Pretending that the harm to public health from discouraging a healthy, environmentally-friendly activity was a good thing because fewer cyclists means fewer cycling injuries is an unhelpful case of "spin".
2) Misleadingly count other road safety improvements for cyclists as "benefits" of helmet laws
|In Victoria, campaigns against speeding and drink-driving were introduced about the same time as the bicycle helmet law. The British Medical Journal reported that total accident costs were reduced by an estimated £100M for an outlay of £2.5M. 9.|
Fig 2 above shows pedestrian deaths and the timing of the bike helmet law. The "benefits" for pedestrians seem more impressive than for cyclists!
To see how much of the claimed benefits for cyclists were due to safer roads, Table 1 below compares Victorian Transport Accident Commission data on deaths and serious head injuries (DSHI) and other serious injuries (OSI) before and after the bicycle helmet law. In the two post-law years, pedestrian DSHI fell to 74% of pre-law numbers and pedestrian OSI to 83%.
Assuming the improved road safety would have generated similar benefits for cyclists, we'd expect 53.6 cyclist DSHI and 167 OSI without the helmet law (Table 1 above). The "true" effect of the law was therefore to reduce DSHI to 77% of the expected number and OSI to 74%.
Helmets don't prevent other serious injuries so the fall in OSI must be due to the reduction in cycling. The marginally lower fall in DSHI than OSI implies there was no additional benefit of increased helmet wearing on serious head injuries ... i.e. the main effect of the law was to discourage cycling.
3) Hide important information on numbers of adult cyclists counted just before the law; instead compare surveys at different times of year
In Melbourne, Victoria cyclists were counted in three annual surveys. The first was just before the helmet law in May 1990. It was followed by two post-law surveys in May 1991 and May 1992 at the same 64 sites.
Each site was counted for 10 hours per survey, covering the same time periods including weekend and weekday use. All cyclists were counted and the times taken to cycle through marked areas also recorded, except for adults in May 1990. The three surveys in May used similar protocols to an earlier survey conducted at a different time of year – December and January 1987/88.
Cycling is seasonal – comparing surveys at different time of year is misleading
Cycling injuries vary substantially with time of year (Fig 1), indicating there are large seasonal differences in the amount of cycling.
The Dec/Jan survey does not seem particularly comparable with the other three surveys, presumably because it was carried out at a different time of year. For example, the higher number of children under 12 might reflect the holiday period. Children and adults also cycled faster in 1987/88 (children: 1.05 in 1987/88 vs 1.80 in 1990; adults 0.48 in 1987/88 vs 0.88 in 1990, Table 2 above).
Strong correlation between cyclists counted and total time to cycle through marked areas
The Melbourne surveys were unique in that, as well as counts, times taken to cycle through marked areas were recorded. Under the same conditions (e.g. the same time of year) total numbers were highly correlated with total times. Unless average speeds change dramatically, or are unduly affected by traffic conditions, both should reflect the change in cycling due to helmet laws.
This is demonstrated in Fig 3 (left) showing teenage cycling. Both numbers counted and total times increased substantially from Dec-Jan 87/88 to May 1990.
This could be due to the different time of year, or perhaps cycling increased in popularity from Dec 1987 to May 1990.
The helmet law resulted in a big decline in both numbers of teenagers counted and total time taken from 1990 to 1991 (Fig 3).
If times had not been recorded for teenagers in 1990, the total time could have been estimated from numbers counted in 1990 and average times to cycle through the marked areas in 1991 (1.10) and 1992 (1.00). Multiplying the count of 1293 teenage cyclists in 1990 (Table 2) by 1.05 (the average of 1.10 and 1.00) leads to a fairly similar estimate (13.6) to the actual value of 13.1 (Table 2). In contrast, as shown in Fig 3, drawing a straight line between the 1987/88 and 1990 surveys (blue dotted line in Fig 3) results in a ridiculous "estimate" of teenage cycling (8.2) that is nothing like the true value of 13.1!
Figure 4 to the left shows that a realistic estimate for adult cycling derived from numbers counted would have led to the conclusion that the helmet law discouraged adult cycling.
Instead, a paper published by Max Cameron and colleagues claimed: "Surveys in Melbourne also indicated a 36% reduction in bicycle use by children during the first year of the law and an estimated increase in adult use of 44%" 6.
Cameron's "estimated increase" is even larger than the invalid inflated estimate (Fig 4) obtained by ignoring numbers of adults counted in 1990.
A report by CARRS-Q10 went even further and claimed: "In Melbourne adult cyclist numbers doubled after the helmet legislation was introduced." Despite inflated estimates due to a bicycle rally passing through one site in 1992, fewer adult cyclists were counted in both post-law surveys than an identical survey just before the helmet law. How could any serious researcher claim that numbers of adult cyclists "doubled"?
When counts provide almost as much information as times, why would any serious researcher claim: "Because the 1990 survey did not cover adult bicyclists, it was not possible to fully examine the change in their bicycle use"? 6
Cameron's estimates (20 hours of cycling per week for every man, woman and child in Melbourne) simply don't add up!
The problem of simply drawing a straight line between the results from the Dec 87/Jan 88 survey and the 1991 and 1992 surveys is illustrated in Figure 7 of Cameron's AAP paper 6. The "estimate" of total cycle use in 1990 is much lower than the sum of the three age categories, indicating major errors in their methods. Another obvious error is the estimate of 60 million hours of cycling per week in Melbourne (reported in the paper to have a population of 3 million), an average of 20 hours cycle use per week for every man, woman and child in the city!
Such basic checks of plausibility (that the total is consistent with the sum of child, teenage and adult cycling, or the plausibility of an average of 20 hours cycle use per person per week) appear to have been omitted from Cameron's work!
Exaggerating the benefits of helmet laws by claiming reductions in cycling and effects of safer roads as "benefits" of helmet laws seems inappropriate. Counting fewer adult cyclists in the post-law surveys but claiming the "number of adult cyclists doubled" might be considered even more inappropriate.
The results in Section 2 above imply that, after accounting for the improvement in road safety the reduction in deaths and serious head injuries was less than the reduction in other serious injuries to cyclists in collisions with motor vehicles and all plausible estimates of the reduction in cycling (discounting those noted in Section 3 to be invalid). Instead of a public benefit, the law almost certainly caused public harm by discouraging a healthy, environmentally-friendly form of transport.
Instead of a public benefit, the law almost certainly caused public harm by discouraging a healthy, environmentally friendly form of transport.
4) Ignore obvious signs helmet wearing wasn't the only reason for drops in percent head injury
|MUARC's errors in evaluating the effect of helmet laws became evident in 1992 when they reported that their predictions showed that head injury rates would fall to zero before helmet wearing reached 100%! (see Figure 12 of Cameron et al. 2 to the left).|
|Such impossible results should have alerted the researchers to look for other factors affecting head injury rates, such as the remarkably similar trends in head injury rates of cyclists and pedestrians shown in Figure 2 of Robinson 1 (reproduced to the left).|
If MUARC had included both pedestrian injuries and the substantial (21% to 24%) reductions in severely injured cyclists who did not have head injuries 2 in a comprehensive model, perhaps they might have concluded that after accounting for the 46% reduction in teenage cycling 6 and the improvements in road safety from reduced speeding and drink-driving, the increased helmet wearing achieved very little benefit and perhaps substantial harm from the large reductions in teenage and therefore future adult cycling.
"Solve" the problem by fitting models implying that helmets become less effective as more people wear them
It seems implausible that helmets would become less effective as more people wear them. In 1992, Cameron et al. 2 admitted this was the case, stating that a linear relationship was to be expected "if the effectiveness of helmets in reducing head injuries is constant, and the cyclists saved from head injury sustain other severe injuries requiring hospital admission".
Most evaluations of helmet wearing seem to have encountered this problem, which is illustrated in Table 3 to the left.
Before the law in New Zealand, helmet wearing increased from 30% to 43% of adults and this was accompanied by a 9.7 percentage point reduction in percent head injury.
But the increase from 43% to 93% of cyclists reduced head injuries by only 3.2 percentage points.
So an increase of one percentage point in helmet wearing before the law reduced the head injury percentage by 0.74 percentage points. In contrast, an increase of one percentage point in helmet wearing because of the law reduced the head injury percentage by 0.06 percentage points.
The most likely explanation is that, as is evident for helmet wearing and head injury rates in Ontario, Canada (see section 8), helmet laws were introduced when head injury rates were trending down for reasons unrelated to helmet wearing.
5) Ignore risk compensation and safety in numbers
When children ran an obstacle course wearing a helmet and wrist guards, tripping, falling and bumping into things increased by 51% compared to without 12. There would be little point of making helmets compulsory if the increase in helmet wearing encouraged cyclists or drivers 13 to take more risks, resulting in increased injuries per cyclist and counteracting any benefits of helmets.
For similar reasons, increased injury rates because of reduced safety in numbers would also be counterproductive 4, 14. Many jurisdictions introduced helmet laws but few measured cycle use reliably enough to compare the change in injury rates with the change in cycling. When this was possible, evidence suggests that helmet laws increased injury rates.
After Alberta, Canada made helmets mandatory for children, numbers of child cyclists halved but injuries increased, suggesting the risk of injury more than doubled 15. In NSW, head injuries to child cyclists fell by 29% but numbers counted in observational surveys fell by 36% and 44%% in the first and second years of the helmet law, suggesting an 11-27% increase in the risk of head injury per child cyclist 1.
In Victoria, child cycling was estimated to have fallen by 33% and 36% in the first and second years of the helmet law but injuries fell by only 22% and 25% respectively, suggesting that the risk of injury increased by 16-17% 1.
The apparent increase in the risk of injury per cyclist suggests that, ignoring the lost health and environmental benefits of reduced cycling, helmet laws were detrimental to public health because of risk compensation and reduced safety in numbers.
6) Ignore the health and environmental costs of reduced cycling and reduced safety in number
After accounting for injury costs, an Australian government report concluded that the net health benefits of cycling amounted to 0.75 cents per kilometre in 2013. For transport trips, there were additional savings per kilometre of 35 cents in vehicle operating costs, 20.7 cents in reduced congestion, infrastructure savings of 5.2 cents and environmental benefits of 5.9 cents per kilometre 16.
In 1985/86, bicycle travel accounted for 3.9% of all trips in Australia, including 2.8% in NSW (1.6% in Sydney and 5.0% in the rest of NSW) 17. The same amount of cycling today (2.24 km per person per week, about 85% for transport purposes) has estimated health, environmental and other benefits of $2.4 billion per year.
In Victoria, after accounting for the improvements in pedestrian injuries, Table 1 (section 2 above) shows that the fall in head injuries for cyclists in motor vehicle crashes was less than the reduction in cycling. This means that injuries per cyclist increased, compared to what would have been expected without the law. As noted in section 6 above, the risk of head injuries to child cyclists in NSW increased by 11-27% and the risk of all injuries to child cyclists in Victoria increased by 16-17%.
The lost health and environmental benefits from even a 25% fall in cycling – more than half a billion dollars per year based on the value per km cycled from the recent government report 16 – is a totally unacceptable price to pay for a law that did not achieve its stated objective of making cycling safer.
7) Claim non-enforced laws with no long-term effect on helmet wearing are beneficial
|The helmet law for children in Ontario, Canada was not enforced. After a temporary increase, helmet wearing rates fell back to pre-law levels by 1999. As shown to the left, head injury rates trended downward and were much lower in 2001/02 than 1996-98, when helmet wearing peaked.|
Unfortunately, other researchers investigating helmet laws were not told about the return to pre-law wearing rates for many years. Indeed, a paper published in 2003 reported helmet wearing rates to 1997 with no mention of the all-important data for 1999 18.
These problems were noted in a response published on the BMJ website on 14 January 2007 3 summing up the debate on the paper by Robinson 20. Fig 2 of the published response shows a declining rate of head injuries, which bears little relationship to helmet wearing rates.
Given the lack of relationship between helmet wearing and head injury rates, it seems unlikely that the helmet law had any real benefit. Many years later a full analysis of the effects of helmet laws throughout Canada confirmed that "injury rates were already decreasing before the implementation of legislation and the rate of decline was not appreciably altered on introduction of legislation" 21.
|Disentangling the effect of different factors such as increased helmet wearing and safer roads is not easy. Fig 3 of the debate summary 3 (left) highlights the importance of considering the safety of other road users such as pedestrians.|
If this is not done, the Canadian researchers who naïvely claimed helmet laws were beneficial because there was a greater decline in head injuries in provinces that passed legislation 22 would also have to conclude the same was true for pedestrians!
The same researchers also naively concluded that helmet laws did not discourage cycling, based on the results from wildly fluctuating surveys that would have been incapable of detecting a 25% decline in the amount of cycling. The laws were later found to have no long-term effect on helmet wearing rates 23. Perhaps the main effect of widely-ignored laws (such as Ontario's child helmet law) is to teach children to disregard road safety laws, another possible reason why injury rates might increase.
8) Measuring changes in cycle use
Although telephone surveys can provide information about cycle use, they are subject to large sampling variation, difficulty in recalling the exact number of bicycle trips in the past few months, or wishful thinking about how often people feel they ought to participate in a healthy activity.
In Western Australia, responses to a specific question about whether adults would cycle more if not legally required to wear a helmet produced a much higher estimate of the detrimental effect of helmet laws (the number who said they would was equal to 64% of current adult cyclists) than estimates of the effect of the law on the cycling of other people in their family 1.
Similarly, in South Australia a telephone survey found no significant decline in the amount people said they cycled (Tables 1a and 1b) but there was a large, significant drop in how much they had actually cycled in the past week 24. In 1990 (pre-law), 17.5% of males aged at least 15 years reported cycling in the past week (210 out of 1201, Table 5a), compared to 13.2% (165 out of 1236) post-law in 1993. For females, 8.1% (102 out of 1357, Table 5b) had cycled in the past week in 1990 compared to 5.9% (98 out 1768) in 1993 24.
These reductions (24% for males, 26% for females aged at least 15 years) are statistically significant (P < 0.005 for males, P = 0.025 for females). Responses were also sought from one child in the household but no information was provided on how the child was chosen and information on children's cycling was available for only 25-26% of households. The survey did not reveal any significant change in children's cycling but this might be explained by a sampling bias towards younger children, who were noted in Victoria to be much less affected by helmet laws than teenagers 1. Destinations for children (but not adults) included "own property", suggesting the intention was to cover riding in the backyard. The high proportion of children who listed this (47% of males, 59% of females) suggests a possible bias towards young children. Marshall and White also noted a 38.1% decline in children's cycling to school but no obvious reduction in commuter cycling 24.
Commuter cycling, like cycling by young children, was probably less affected than teenage cycling, making the effect of laws harder to detect, especially in the presence of other trends. One useful way to examine the impact is to compare census data on cycling and walking to work in states with and without enforced helmet laws in 1991 (above) 25. In states with enforced helmet laws (red lines) the 1991 census showed a reduction in cycling. In contrast, cycling to work increased on average in states without enforced laws (blue lines), with sharp declines evident in the next (1996) census when helmet laws were enforced in these states. Walking to work showed almost identical declining trends irrespective of whether helmet laws were enforced at the time of the 1991 or 1996 census.
The results strongly suggest that helmet laws did indeed reduce cycling to work, more so in regional areas. The discrepancy between regional areas and capital cities might relate to perceived safety. In areas where cyclists are already wearing helmets because of perceived danger, the impact of helmet laws in discouraging cycling is likely to be less than in areas where helmet wearing is low and cycling is generally considered safe. For example, UK surveys show much higher helmet wearing rates in London (69.5%) than elsewhere (29.9%) 26.
New Zealand also introduced helmet laws in January 1994. Cycling to work declined in New Zealand (from 11.6% in 1989/90 to 7.3% in 1997/98 to 4.3% in 2004-08 and cycling to secondary school from 18.6% (1989/90) to 10.6% (1997/98) to 4.9% (2004-08) 27. The total amount of cycling fell from 15 minutes per person per week in 1989/90 to 9 minutes in 1997/98, and remained at this level (see Appendix).
9) Portray decreased risks as an "increase"
Exposure to Montreal’s public bicycle share scheme (PBS) more than doubled the likelihood of cycling (odds ratio = 2.86) after the second season of implementation.28
Safety also improved substantially with a 50% reduction (from pre-implementation to season two of implementation) in the number of collisions per 100 person-days of cycling, although total numbers of collisions and near misses did not decrease.29
Another study considered Montreal and four other cities (Boston, Miami Beach, Minneapolis, Washington D.C.) that implemented PBS. Head injuries fell by 14% (from 319 to 273 per year) while moderate or severe head injuries fell by 27%. There was an ever larger reduction in other injuries from 437.5 to 272 per year.30 There was no similar improvement in five non-PBS cities where head injuries decreased by 4% and those classed as moderate or severe increased by 6% (from 180.5 per year to 192). Non-head injuries also increased by 6% (see graph).30
Astonishingly, an article on the University of Washington's website discussed an analysis of the above data and a publication with lead author from the Harborview Injury Prevention and Research Center,30 claiming the "risk of head injury among cyclists increased 14 percent". In response to letters pointing out that head injuries were actually 14% lower, the authors responded that the "conclusion that bike safety has improved after the institution of the PBS is not warranted without denominator data" (i.e. information on the amount of cycling).31
However, as well as the published data showing substantially increased cycling in Montreal (and a halving of the number of collisions per 100-person days), a letter by Prof Kay Teschke32 points out that cycling to work in PBS cities increased by an average of 33%, compared to 18% in non-PBS cities, again suggesting that cycling increased more in PBS cities while head injuries decreased, leading to real improvements in safety, environmental gains and better health.
Other evaluations have shown that PBS generate substantial benefits. In 2007, Velib was said to have driven Paris "cycling mad".33 Since then, many cities have successfully introduced public bikshare schemes and improved health. Compared to car users, the estimated annual change for Barcelona's 181,982 Bikeshare users was 10.5 to 12.5 avoided deaths from increased physical activity offset by 0.03 deaths from road traffic incidents and 0.13 deaths from air pollution34.
In its first 12 months (to March 2012), 7.4 million trips were made on London's Bikeshare scheme. An evaluation in the BMJ estimated net health benefits (after subtracting losses from injuries sustained during those trips) of 72 additional years of healthy life for men (who accounted for 71% of cycling time) and 15 for women.35
After New York launched its Citi-Bike scheme in May 2013, cycling in the Citi-Bike area spiked by 25% with more than five million Citi-Bike rides to November 2013.
Yet in November 2013, the number of cyclist fatalities in NYC, year to date, was lower than any year since record-keeping began in 1983.36
Source: American Journal of Public Health in August 1991. The letter comments on unbelievable statistics in the paper by Diane Thompson and colleagues from the Harborview Injury Prevention and Research Center (e.g. that a 5-9 year old child who had one injury should expect additional injuries every 3.3 miles).
The extraordinary claim that head injuries increased when they decreased is not the first implausible claim from the Harborview Injury Prevention and Research Center. A letter to the editors of the AJPH in 1991 (see box above) questioned their estimates of extraordinarily high injury rates - e.g. that a 5-9 year old child who has had one injury should expect additional injuries every 3.3 miles. Their claim that helmets prevent 85% of head injuries has now been withdrawn by two US Federal Government agencies 45.
10) Probable increases in risk portrayed as decreases
The previous section provides evidence of increased cycling and improved safety after implementation of bikeshare schemes. This again raises the question of whether decreases in cycling because of helmet laws are likely to have the opposite effect of increasing injury rates. It certainly appears to be true in the long term.
For example, in New Zealand from 1989/90 to 2011, average time spent cycling (on roads and footpaths) fell by 79% for children aged 5-12 (from 28 to six minutes per person per week) and 81% for 13-17 year olds (52 to 10 mins/person/week). The decreasing trends in cycling were accompanied by large increases in hospital admissions per million hours of cycling from crashes not involving motor vehicles, quadrupling in 15-19 year olds (from 11.6 in 1989 to 45.9 injuries per million hours in 2011) and more than doubling for children (from 39.5 to 85.4 per million hours) and adults (from 15.9 to 32.3 per million hours.37
The largest falls in cycling (reductions of 21 minutes per week for 13-17 year olds and 13 minutes per week for children) and increases in injury rates were from the pre-law survey in 1989/90 to the first survey (1997/98) after the helmet law.37
A comparison of hospital admissions in NSW38 with cyclists counted in surveys39,40 also suggests that injury rates increased. The graph below shows a clear decrease in child cycling at road intersections (r), recreational areas (l) and schools (e) after helmets were made compulsory for children. In April 1993, 44% fewer child cyclists were counted than in April 1991. However, head injuries fell by only 29%, suggesting that the risk of children's head injuries increased.1
The data for adults are harder to interpret because the second survey (in April 1991) was in sunny conditions but the pre-law survey was at a different time of year (October) and conditions were overcast in Sydney and interrupted by rain in some areas.41 Increased cycling because of better weather or other seasonal variation may well have masked the effect of the introduction of a helmet law for adults.
The graph above shows a longer time series of counts for cyclists of all ages at the same sites (25 road intersections) in Sydney,40 compared to numbers of hospital admissions of cyclists with head and arm injuries.38 Despite the downward trend from 1991 onwards in numbers of cyclists counted at the Sydney sites, head and arm injuries trended upwards, again suggesting that cycling became more dangerous after helmet laws were introduced.
Graph from Olivier et al. in which the pre-law values for adults were obtained by ignoring the fact that adults over 20 were not counted at some sites in 1991, and for both adults and children appear to have been calculated from post-law trends assuming that the law has no effect!
Published information does not always reflect this reality. For example, the graph (left) shows how the survey data were represented by Jake Olivier and colleagues.42 The only comparable data for children in 1990 and 1991 are the surveys at road intersections where, as shown in the detailed graph above, counts were marginally higher in 1991 than 1990. In contrast, Olivier's graph shows a decrease in children's cycling which appears to have been calculated by "estimating" the amount of children's cycling in 1990. Yet a survey of NSW schoolchildren who owned bikes found that 51% of those who hadn't cycled in the past week said this was because of the helmet law.20 This implies that post-law trends in children's cycling were strongly influenced by the helmet law, so there is no valid way of determining pre-law trends, let alone seasonal effects. It is both misleading and invalid to estimate the amount of cycling in 1990 assuming the helmet law did not affect trends, then use the results to demonstrate the effect of the law!
The graph of Olivier et al. for adults is even more misleading because cyclists over 20 were not counted at some recreational areas in 1991.43 Comparable counts were available only for 1992 and 1993. It is therefore not possible to determine whether the cause of the big increase from 1992 to 1993 was random variation or the continuation of a trend that started before the law. Other surveys (e.g. a figure equivalent to 64% of current adult cyclists in Western Australia saying they would cycle more if not legally required to wear a helmet20) suggests that helmets laws do indeed deter cycling. Using non-comparable data for 1991 and (as was pointed out for children) assuming the helmet law did not affect trends in order to estimate the amount of cycling in 1990 in a graph to demonstrate the effect of the law, are both misleading and invalid.
Although many factors affect the decision to cycle, the bulk of evidence suggests that helmet laws and portrayal of cycling as a dangerous activity in order to persuade cyclists to obey helmet laws is an important factor that affects the amount of cycling.
The evidence cited here also suggests that in order to avoid embarrassment, some researchers exaggerated the benefits of helmet laws. This appears to be particularly true of the assessment by MUARC of Victoria's helmet law, where the law was claimed to have reduced head injuries by 40% despite, as shown above, most of this reduction appearing to be due to reduced cycling and safer roads.
In 1990, 1567 adult cyclists were counted but only 1106 (29% less) in an identical survey in 1991. In 1992, when counts were inflated by a bicycle rally passing through one site, there were 1484 adult cyclists (still 5% less than 1990). Although it would have been possible to estimate the total adult cycling time from the number of adults counted, MUARC chose instead to use a survey taken some years earlier at a different time of year (Dec/Jan 1987/88 that counted only 1079 adult cyclists) as the pre-law "baseline".
So instead of the observed reductions of 29% and 5% in adult cycling, various inconsistent claims were made about 'increases' in adult cycling, even that "in Melbourne adult cyclist numbers doubled after the helmet legislation was introduced" 10."
This is a truly remarkable claim to make when both post-law surveys counted fewer adult cyclists than the identical pre-law survey in May 1990.
The evaluation of helmet laws in Canada was similarly problematical in that the researchers delayed reporting the fact that child helmet wearing had fallen to pre-law levels by 1999 until 2006. Perhaps this information would have generated embarrassing questions about how a non-enforced law that had no long-term effect on helmet wearing could possibly have generated the claimed reductions in head injuries.
The ethics of such "spin" are questionable. Strong justification is required for laws that deprive people of the freedom to cycle without a helmet, or have the potential to increase health costs by discouraging a healthy, environmentally-friendly form of transport, or increase the risk of injury because of risk compensation or reduced safety in numbers.
With evidence suggesting that the risk per cyclist is higher than would have been expected without the law, action should be taken to correct the problem.
Further details and sources of information
1. Robinson, D.L., Head injuries and bicycle helmet laws. Accid Anal Prevent, 1996. 28: p. 463-475.
2. Cameron, M.H., L. Heiman, and D. Neiger, Evaluation of the bicycle helmet wearing law in Victoria during its first 12 months, 1992, Monash University Accident Research Centre Report No 32.
3. Robinson, D.L. Postscript. Rapid reponse to 'No clear evidence from countries that have enforced the wearing of helmets'. Available at http://www.bmj.com/content/332/7543/722.2?tab=responses. BMJ, 2007.
4. Robinson, D.L., Safety in Numbers in Australia: more walkers and bicyclists, safer walking and bicycling. Health Promotion Journal of Australia, 2005. 16(1): p. 47-51.
5. Robinson, D.L., Changes in head injury with the New Zealand bicycle helmet law. Accid Anal Prev, 2001. 33(5): p. 687-91.
6. Cameron, M.H., et al., Mandatory bicycle helmet use following a decade of helmet promotion in Victoria, Australia - an evaluation. Accid Anal Prevent, 1994. 26(3): p. 325-37.
7. Strong, G., Hardhat defiance cops night in cell, in Sunday Age 12 May 1996.
8. ATSB, Road Fatalities Australia 2001 Statistical Summary, 2002, Australian Transport Safety Bureau.
9. Powles, J.W. and S. Gifford, Health of nations: lessons from Victoria, Australia. BMJ, 1993. 306: p. 125-7.
10. Haworth, N.L., et al., Bicycle helmet research: CARRS-Q monograph 5. 2010.
11. Robinson, D.L., Head Injuries and Helmet Laws in Australia and New Zealand, 2012, Bicycle Helmet Research Foundation. Available at http://cyclehelmets.org/1241.html.
12. Morrongiello, B.A., B. Walpole, and J. Lasenby, Understanding children's injury-risk behavior: Wearing safety gear can lead to increased risk taking. Accid Anal Prev, 2007. 39(3): p. 618-23.
13. Walker, I., Drivers overtaking bicyclists: Objective data on the effects of riding position, helmet use, vehicle type and apparent gender. Accident Analysis and Prevention, 2007. 39(2): p. 417-425.
14. Jacobsen, P.L., Safety in numbers: more walkers and bicyclists, safer walking and bicycling. Inj Prevent, 2003. 9(3): p. 205-9.
15. BHRF Alberta's helmet law – children's cycling halved, injuries increased per cyclist. Available at http://cyclehelmets.org/1250.html. 2012.
16. Australian Government, Walking, Riding and Access to Public Transport. Supporting active travel in Australian communities. Ministerial statement, 2013, Department of Infrastructure and Transport.
17. INSTAT, Day-to-day travel in Australia 1985-96, 1988, Federal Office of Road Safety: Report CR 69.
18. Parkin, P.C., et al., Influence of Socioeconomic Status on the Effectiveness of Bicycle Helmet Legislation for Children: A Prospective Observational Study. Pediatrics, 2003. 112(3): p. e192-196.
19. Carr, D., M. Skalova, and M. Cameron, Evaluation of the bicycle helmet law in Victoria during its first four years, 1995, Rpt 76 Monash Univ Acc Res Centre Melbourne.
20. Robinson, D.L., No clear evidence from countries that have enforced the wearing of helmets. BMJ, 2006. 332: p. 722-725.
21. Dennis, J., et al., Helmet legislation and admissions to hospital for cycling related head injuries in Canadian provinces and territories: interrupted time series analysis. BMJ, 2013. 346.
22. Macpherson, A.K., et al., Impact of mandatory helmet legislation on bicycle-related head injuries in children: a population-based study. Pediatrics, 2002. 110(5): p. e60.
23. Kary, M. Errors and omissions in Canadian research group's bicycle helmet papers. Bicycle Helmet Research Foundation website. Available at http://www.cyclehelmets.org/1244.html. 2012.
24. Marshall, J. and M. White, Evaluation of the compulsory helmet wearing legislation for bicyclists in South Australia Report 8/94, 1994, South Australian Department of Transport, Walkerville, South Australia.
25. BHRF Changes in cycle use in Australia - http://www.cyclehelmets.org/1194.html (accessed December 2013).
26. Sharratt, C., L. Walter, and O. Anjum, Cycle helmet wearing in 2008, 2009, TRRL (available at http://www.trl.co.uk/online_store/).
27. NZ Ministry of Transport, How New Zealanders Travel. Spreadsheet available at http://www.transport.govt.nz/research/travelsurvey/latestresults/. 2009.
28. Fuller, D., et al., Impact evaluation of a public bicycle share program on cycling: a case example of BIXI in Montreal, Quebec. American journal of public health, 2013. 103(3): p. e85-e92.
29. Fuller, D., et al., The impact of implementing a public bicycle share program on the likelihood of collisions and near misses in Montreal, Canada. Preventive Medicine, 2013. 57(6): p. 920-924.
30. Graves, J.M., et al., Public Bicycle Share Programs and Head Injuries. American Journal of Public Health, 2014. 104(8): p. e106-e111.
31. Graves, J.M., I.B. Pless, and F.P. Rivara, Graves et al. Respond. American journal of public health, 2014: p. e1-e1.
32. Teschke, K. and M. Winters. Letter to the Editor of the American Journal of Public Health, available at http://cyclingincities-spph.sites.olt.ubc.ca/files/2014/06/Graves-AJPH-as-submitted.pdf. 2014.
33. Chrisafis, A., The City's Gone Cycling Mad, in The Guardian 2007.
34. Rojas-Rueda, D., et al., The health risks and benefits of cycling in urban environments compared with car use: health impact assessment study. BMJ, 2011. 343.
35. Woodcock, J., et al., Health effects of the London bicycle sharing system: health impact modelling study. BMJ, 2014. 348.
36. NY City, Making Safer Streets, 2013, New York City Department of Transportation. Available at http://www.nyc.gov/html/dot/downloads/pdf/dot-making-safer-streets.pdf.
37. http://cycle-helmets.com/. Trends in cycling, walking and injury rates in New Zealand. Available at: cycle-helmets.com/newzealand-road-users.html. 2014.
38. Olivier, J., S.R. Walter, and R.H. Grzebieta, Long term bicycle related head injury trends for New South Wales, Australia following mandatory helmet legislation. Accident Analysis & Prevention, 2013. 50: p. 1128–1134.
39. Walker, M., Law compliance among cyclists in New South Wales, April 1992. A third survey, 1992, Road and Traffic Authority Network Efficiency Strategy Branch.
40. Walker, M. Bicycling in Sydney: law compliance and attitudes to road safety in Velo Australis. 1996. Fremantle, Western Australia.
41. Walker, M., Law compliance and helmet use among cyclists in New South Wales, April 1991, Road and Traffic Authority Network Efficiency Strategy Branch.
42. Olivier, J., et al., Statistical Errors in Anti-Helmet Arguments, in Australasian College of Road Safety Conference – "A Safe System: The Road Safety Discussion" 2013: Adelaide.
43. Smith, N. and F. Milthorpe, An observational survey of law compliance and helmet wearing by bicyclists in New South Wales - 1993., 1993, Roads and Traffic Authority.
44. NZ Ministry of Transport, New Zealand Household Travel Survey 2009–2013 Cycling. Available at:http://www.transport.govt.nz/research/travelsurvey/latestresults/, 2013.
45. Washington Area Bicyclists Association, June 2013. Available here.
New Zealand Household Travel Survey data 44 showing the decline in cycling. New Zealand's helmet law commenced in January 1994.