Background
In the Netherlands, cycling is a common mode of transportation and bicycle trips account for up to 28% of all transfers made [
1]. Dutch people cycle a mean daily distance of 2.9 km and over 40% of inhabitants cycle at least once a day, the highest level of any country in the European Union [
2,
3]. Although the level of road safety in the Netherlands is high, roughly 78,400 injuries per year are treated at an emergency department (ED), of which 13,300 (17%) result in hospital admissions [
4]. Even though the total number of bicycle related fatalities declined between 1996 (239) and 2016 (189), the number of bicycle related deaths remained at a high level with an average number of 189 casualties in the last 5 years [
5]. In 1996, 20% of all traffic deaths were bicycle related, increasing to 30% in 2016 [
5].
Traumatic brain injury (TBI) is the main cause of severe morbidity and mortality after an accident involving bicyclists. More specifically, 32% out of all hospitalized cyclists with severe injury has a head or brain injury [
1,
4,
6]. The incidence of bicycle related TBI treated at the ED has increased with 54% between 1998 and 2012, while the overall incidence of bicycle related injuries treated at the ED remained relatively stable in that period. In 2012, bicycle related TBI was treated 43 times per 100,000 persons in the ED and resulted in subsequent hospitalization in 64% of the TBI cases [
6].
Most survivors of this form of injury remain impaired for life. TBI has been associated with a decline in cognitive capacity, long lasting physical disability and handicap, and the development of mental illness [
7‐
9]. The consequences of TBI often prohibit survivors from returning to full employment, reduce their quality of life, and have been linked to an increased risk of suicide [
10‐
13]. Scholten et al. valued the mean healthcare costs and costs due to productivity loss resulting from bicycle related TBI at € 19,620 per case for the Netherlands, of which € 4940 healthcare costs and € 14,680 productivity losses per case. Total costs were estimated at 74.5 million euros in 2012 [
6].
About 75% of all bicycle-related head injuries are caused by single-bicycle accidents, i.e. accidents without any motorized vehicles involved. In most cases this involves falls or collisions with an obstacle [
14]. Polinder et al. (2016) identified these injuries as a priority area for prevention [
15]. The use of bicycle helmets has been found to be an effective measure of preventing head and brain injuries, especially in the case of these single-bicycle accidents [
16‐
19]. It is associated with a 51% reduction in odds for head injury, according to the most recent and comprehensive systematic review on the subject [
18]. Several countries have introduced legislation, which enforces the use of helmets. Examples are Spain, Sweden, Australia, New Zealand, certain states in Canada, and the United States [
17]. However, many countries only require children to wear helmets. A review of population-based studies that compare injury data before and after the introduction of legislation that enforces use of bicycle helmets found decreases in head injuries and mortality in certain parts of Canada and the United States [
17]. However, some other similar studies found only a small or no obvious effect of legislation on number of injuries in New Zealand and certain parts of Australia and Canada [
20,
21].
Bicycle helmets are not mandatory in the Netherlands and are generally only used by young children, mountain bikers and racing cyclist [
22]. According to a survey commissioned by the Royal Dutch Touring Club, 73% of adults and 84% of children under 17 never wear a helmet [
23]. Although a considerable amount of research into the efficacy of both the bicycle helmet and legislation that enforces its use has been performed, the cost-effectiveness of such a law is unknown in the Dutch context. The current research aims to explore the cost-effectiveness of mandatory helmet use by comparing costs and benefits of legislation with the costs and benefits of the existing situation, i.e. voluntary helmet use in a small part of all cyclists. This study can support decision-making concerning a legalization to prevent bicycle related TBI and death [
24,
25].
Discussion
The current study aimed to assess the cost-effectiveness of enforcing bicycle helmet use through legislation as a means of preventing bicycle related TBI and mortality in the Netherlands [
49,
50]. Without considering different age groups, the current intervention is not cost-effective in light of the Dutch reference value for cost-effectiveness of € 20,000 per QALY [
50]. However, the results show a large fluctuation in cost-effectiveness between age groups. It is notable that all ICERs are most favorable in people aged 65 years or older, despite the fact that this group incurred no costs for lost productivity due to retirement. The absence of savings related to productivity losses was offset by a substantially higher incidence rate and higher mean medical costs when compared to the other groups (Table
2 and
Appendix 3). This age group is the only group with an ICER per DALY averted that is below the threshold of € 20,000. In addition, the ICER per death prevented is approximately 26 and 11 times lower as for the youngest and middle age group, respectively.
As no Dutch utility figures for injuries associated with traffic accidents were available, we choose to use disability figures for injuries and DALY outcomes in our analysis. Although use of DALY as an outcome measure in economic evaluations is less common than use of QALYs, it is accepted as well, as appeared from a review of Oostvogels et al. [
51]. Therefore, the intended health effect is expressed as DALYs averted following mandatory helmet use. Both QALY and DALY are a form of health-adjusted life expectancy (HALE). The primary difference between the two outcome measures is that QALYs primarily measure health gains following interventions, while DALYs measure health losses from disease and death [
52]. Another difference is that DALYs, by adjusting life years for disability caused by one single health problem such as TBI, do not consider comorbidity and thus tend to be relatively larger than QALYs when comorbidity is present. Therefore, use of DALYs could lead to an overestimation of the disease burden [
52] and a somewhat flattered ICER. In absence of a clear threshold value in the Netherlands for cost-effectiveness when DALYs are used as outcome measure, we chose to use the same threshold value for DALYs as is commonly used for QALYs in the Netherlands, namely € 20,000 [
50]. This threshold value is much more stringent than the WHO guidelines on cost-effectiveness that state that any intervention with a cost-effectiveness ratio below the GDP per capita (€ 43,100 for the Netherlands in 2017) should be regarded as a cost-effective intervention. Should we have followed this standard, our conclusions on cost-effectiveness of mandatory helmet use would have been more positive than they currently are.
When interpreting the results, it should be remembered that we chose not to include the often cited substitution effect and risk compensation effect in our analysis [
53,
54] . The former is related to the fact that mandatory helmet use could discourage current cyclists to use bicycles but other forms of transportation instead for (part of) their journeys [
53,
54]. The latter is said to increase risk for injury by increased risk taking behavior by either the helmet wearing cyclists themselves or by other traffic participants [
55,
56]. However, both mechanisms are not undisputed either [
57]. We decided not to include these factors in our model, due to their uncertain nature. Nonetheless, we should bear in mind that the possible presence of substitution and risk compensation effects could have led to less favorable cost-effectiveness estimates.
Our findings are in line with the existing literature. However, most previous economic evaluations of helmet laws have used cost-benefit methods rather than cost-effectiveness analysis [
36,
43]. Economic evaluations that use a societal perspective of costs generally agree that a helmet law is not beneficial, i.e. that the costs outweigh the associated benefits [
43]. Although many studies evaluate the effectiveness of bicycle helmet laws, not many evaluate their cost-effectiveness. To our knowledge, no contribution has been made to this field in the last two decades. In one of the few papers on cost-effectiveness, Hendrie et al. focused on the cost of public education campaigns and police enforcement and not on medical costs and productivity losses [
38]. They did however measure the cost of purchasing helmets, which was also by far their largest cost item. They estimated the cost per TBI prevented to be € 99,123 for their aggregated data model and € 212,769 for their individual pooled data model [
38]. This is substantially higher than the €31,028 per TBI prevented that we found, probably mainly because they did not consider the healthcare savings associated with less TBI. In another cost-effectiveness paper, cost per (discounted) life saved was estimated to between € 85,727 to € 110,332 for children from 5 to 12 years, between € 673,193 to € 793,338 for children from 13 to 18 years old and between € 863,340 to € 984,404 for adults [
58]. This is substantially lower than the estimates that we found. Interestingly, the cost per death averted in this study is higher for adults than it is for children, while we found an inverse relationship. The most recent cost-effectiveness research was done in 2000 by Kopjar & Wickizer, and found the same age gradient [
59]. They found the risk of head injury and the largest reduction in absolute risk of head injury due to wearing a helmet to be the highest in children. We found the opposite. Our data clearly show that, in the Netherlands, 0.115% of all people older than 65 years had TBI in 2017, while only 0.050% of children under 15 years was admitted to an Emergency Department with this type of trauma. This might be because the cycling infrastructure in the Netherlands gives this age group a feeling of safety, while they are in fact more vulnerable [
60]. To our knowledge, no other research has looked at the cost per DALY averted or QALY gained to date.
Results of the sensitivity analyses show that the ICERs were most influenced by the efficacy of bicycle helmet use that was assumed in the model. The efficacy of bicycle helmets is often debated in the scientific literature [
18,
61‐
63]. Generally, two types of research have been performed. Firstly, case-control studies, in which brain damage between people that did and did not wear a helmet is compared, and secondly, ecological studies, in which the period before and after the introduction of an intervention to stimulate helmet use is compared. Case-control studies tend to find a higher efficacy of bicycle helmets than time series analysis [
64]. Both types of research are vulnerable to confounding factors in their own regard. The meta-analysis by Olivier et al. that was used in this study includes recent research from several countries over several years and was therefore the best available study [
18]. Additionally, they checked for evidence of time trends and publication bias, which they did not find. Therefore, this meta-analysis was deemed more comprehensive and likely more fitting the Dutch context, as opposed to results from ecological studies, which generally relate to developments over time in one specific country.
In the Netherlands, the share of electric bicycles out of all newly sold bikes rose from 15% in 2011 to 31% in 2017 [
65]. According to the Dutch cyclists’ federation, about 6% of all bicycles in the Netherlands are electric bicycles [
66]. However, more than a quarter of all bicycle related deaths were related to use of electric bicycles. Out of these deaths, three quarters are people aged 65 years and older [
67]. The relatively high mortality under users of electric bicycles hints to a high incidence of TBI in this group. Therefore, users of electric bicycles seem to be a very relevant group for the intervention under study. Unfortunately, the EDs in the Netherlands do not systematically register the use of an electric bicycle by patients admitted with TBI. Hence, we could not stratify for the use of electric bicycles in our research, while we know that between 2010 and 2017 there was an increase of the selling of new e-bikes of 77%.
Strengths of this study
The main strength of the study lies in the use of recent data regarding incidence, medical costs, and disease burden from the Dutch Injury Surveillance System and Dutch Burden of Injury Model. Therefore, to our knowledge, this is the first paper that reports the disease burden of bicycle related TBI in DALYs and the societal costs associated with preventing them.
Limitations of this study
The availability of data used in this research restricted us from performing probabilistic sensitivity analysis (PSA). PSA would have resulted in more reliable and specific ICERs by adding information about the distribution of the model parameters. However, information about the distribution was unavailable for a large number of parameters. The probability of TBI is unrelated to the intensity of bicycle use, while those who cycle more are at greater exposure to TBI risk than those who rarely use a bike. No data are available about bike use intensity. Unfortunately, data on electric bicycles use were not available so we could not distinguish this group of cyclists in this economic evaluation. The fact that we only had healthcare costs available for the first year leads to an underestimation of the cost-effectiveness of the use. Finally, we did not include police enforcement and regulation costs in our study, due to data limitations. The ICERs as estimated in this study may be less beneficial when these costs would have been taken into account. On the other hand, the fact that we only included TBI and not all other injury costs, such as related to fractures, nor other costs of accidents, such as material damage and indirect costs of traffic jams, may have prevented us from reporting more favorable ICERs.
Our results are specific for the Dutch cycling context and, as a consequence, not directly transferable to other countries’ settings. First, in the Netherlands cycling is much more common than in other European countries, Denmark excluded. The infrastructure with cycle lanes and other traffic aspects, such as right of way for cyclists and legal responsibility of motorized traffic in any traffic accident, regardless of actual responsibility, differs enormously from other countries. Consequently, injury risks differ substantially, independent of helmet use. Second, in the Netherlands most cyclist currently do not wear a helmet, whereas in most other countries helmet use is common.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.