Risk Compensation Theory Should Be Subject to Systematic Reviews of the Scientiﬁc Evidence

86 Injury Prevention 2001;7:86–88

OPINION

Risk compensation theory should be subject to systematic reviews of the scientiﬁc evidence

D C Thompson, R S Thompson, F P Rivara

Background suggests that individuals provided with a Many readers of Injury Prevention are quite protective device such as a bicycle helmet or an familiar with the debate over bicycle helmet automobile seat belt will act in a riskier manner use. The core of this debate is the opinion on because of the sense of increased protection one side that helmets are eVective and thus from the helmet or seat belt and thereby nullify should be worn, countered on the other side by the protection aVorded by the helmet or seat the belief that risk compensation negates this belt. The theory of risk compensation is not protective eVect of helmets. A systematic applicable to case-control studies of helmet review on helmet eVectiveness has been eVectiveness per se. Its potential applicability is published in the Cochrane Library.1 The to the impact of widespread use of helmets, objective of the Cochrane review was to deter- particularly legislation to require helmet use, mine whether bicycle helmets reduce head, and its net protective eVect. brain, and facial injury for bicyclists of all ages involved in a crash. The principles required of WHAT ARE THE ARGUMENTS FOR RISK high quality evidence based reviews were COMPENSATION? followed: a comprehensive literature search, The theories of risk homeostasis and risk com- pre-established study selection criteria, and pensation are well summarized by Gerald most importantly a critical review of study Wilde and John Adams.2–4 Those who argue methods. A well conducted systematic review that risk compensation must be taken into identifies and considers all the literature (peer account before bicycle helmets are adopted as a reviewed, government reports, and unpub- safety measure have said: lished papers), and rates the study quality. (1) Encouraging helmet use would have seri- Appropriately, such reviews only include better ous adverse consequences on the public health, designed and conducted studies. The evidence without making any significant diVerence to is then summarized across all the studies. the dangers of riding.5 The literature search for the Cochrane (2) Wearing of a helmet influences cyclists’ review yielded five studies meeting the pre- behaviour, thereby aVecting the likelihood of established criteria for inclusion. The strengths them being involved in such an incident in the and weaknesses of five case-control studies of first place. Cyclists are less likely to ride bicycle helmet eVectiveness were carefully cautiously when wearing a helmet owing to evaluated. The scientific evidence which indi- their feeling of increased security. In this way, cates that bicycle helmets protect against head, they consume some, if not all, of the benefit that would otherwise accrue from wearing a brain, severe brain and facial (upper and mid- 67 face) injuries has been well established. Addi- helmet. There have been no systematic reviews of the Harborview Injury tionally, the evidence indicates that helmets Prevention and provide injury protection in all type of crashes evidence for the relevance of risk compensation Research Center, including those involving motor vehicles. to bike helmets. Mayer Hillman states the evi- Department of Based on this review the authors recom- dence for risk compensation is “overwhelm- Pediatrics, University ing”.6 of Washington, Seattle mended that as a policy bicycle riders of all D C Thompson ages should be encouraged to wear helmets. F P Rivara The purpose of publishing health research, and WHAT EMPIRICAL EVIDENCE EXISTS FOR RISK discourse about it, is to improve the health of COMPENSATION BEHAVIOUR? Department of the public. We believe that the evidence If risk compensation plays a role, one would Preventive Care, indicates such a strong protective eVective of expect the gains from helmet eVectiveness to be Group Health erased or strongly mitigated by increasingly Cooperative of Puget helmets, that the net eVect on the health of the Sound, Seattle public will be positive. risky riding habits of helmet wearers which R S Thompson would neutralize any protective eVect provided by helmet wearing. What do the empiric data Correspondence to: What about the theory of risk show? Diane C Thompson, Harborview Injury compensation? Prevention and Research WHAT IS RISK COMPENSATION? (A) Bicycle riders Center, Box 359960, 325 Risk compensation is frequently raised during The evidence from time series studies in Ninth Avenue, Seattle, WA 98104, USA debates about helmet promotion and legisla- Australia, New Zealand, Europe, and the [email protected] tion. Briefly put, risk compensation theory United States indicates that increased rates of

www.injuryprevention.com Risk compensation theory and bicycle helmets 87

helmet use resulting from multifaceted edu- available to the public from the National Tech- cational campaigns and/or legislation are nical Information Service, Springﬁeld, Vir- linked to signiﬁcant decreases in bicycle related ginia. It is also indexed in the Transportation head injuries.8–13 Given that helmets are very Research Information Service (TRIS) data- eVective, cyclists would have to increase their base.20 21 Authors of a systematic review would risk taking fourfold to overcome the protective obtain and evaluate all available research eVect of helmets. This seems unlikely. before arriving at a conclusion.

(B) Motorcycle riders Summary and conclusion on risk The closest analogy to bicycle helmet use is compensation theory arguments mandatory motorcycle helmet use laws in the In summary the empirical evidence to support United States. Although there is general agree- the risk compensation theory is limited if not ment that motorcycle helmets reduce head and absent. There are a number of studies in the brain injury when a crash occurs, many motor- traYc literature that point out problems or cyclists dislike helmets. Risk compensation show data at odds with the risk compensation/ theory would propose that a motorcyclist homeostasis theory. No systematic review of might drive more recklessly if legislation the evidence for risk homeostasis has been requires helmet use. Motorcyclists may also be conducted. considered vulnerable road users, since motor- We recommend that interested readers con- cycle crashes usually result in serious injury to sult a comprehensive discussion of the risk the motorcyclists themselves and not to cars compensation debate presented by James Hed- and other motor vehicle passengers. In the lund at the Fifth World Conference on Injury United States nearly all 50 states passed laws Prevention and Control.22 James Hedlund pro- requiring motorcycle helmets in the mid- vided his personal view: “I believe the evidence 1960s. In 1976 almost half of the states is overwhelming that every safety law or repealed their laws. This provided an oppor- regulation is not counterbalanced by compen- tunity for a natural experiment. Fewer motor- sating behavior”. We suggest risk compensa- cyclists wore helmets following the repeal of tion is an appropriate area for systematic motorcycle helmet laws. This resulted in a 25% reviews. to 40% increase in motorcycle deaths.14–18 These results strongly imply that wearing a Conclusions on recommending the use of helmet does NOT lead to large increases in risk taking.18 bicycle helmets Based on the solid empirical evidence for bicycle safety helmet eVectiveness, we are confident (C) Motor vehicle drivers and passengers in recommending their use, and policies to An extensive systematic review of automobile encourage their use. However, additional criti- safety interventions by the non-federal national cisms of our Cochrane review have been raised Community Preventive Services Task Force by Bill Curnow and Dorothy Robinson. These has been sponsored by the Centers for Disease criticisms and our replies are published on the Control and Prevention since 1996. The task Cochrane injuries group web site (http:// force looks at both the benefits and the risk for www.cochrane-injuries.ich.ucl.ac.uk/ any given intervention. Based on results of sys- HelmetComment.htm). Criticisms from Rich- tematic reviews, the task force makes recom- ard Keatinge and Mayer Hillman will be mendations on population based interventions published along with author’s replies in the to promote health and prevent disease, injury, next edition of the Cochrane reviews. Please disability and premature death, and to reduce read this series of interesting debates. In our environmental hazards. The task force found opinion, we have provided well founded that child safety seats, seat belts, and alcohol answers to the criticisms. laws all contributed to substantial reductions in We feel there is strong scientific evidence for motor vehicle injuries and deaths. The task recommending or mandating bicycle helmet force recommended a number of community- use. This is a first step in reducing bicycle wide information and enforcement campaigns related head injuries. Encouraging cycling, to promote these measures.19 building a bicycle friendly infrastructure, and Based on risk composition theory, John promoting safe cycling instruction are also Adams has long opposed seat belt legislation.34 important activities. These activities are not The Adams essay published on the Cato Insti- mutually exclusive. Promoting bicycle helmet tute web site discusses risk compensation and use does not exclude other road safety seat belt legislation. The essay explains the approaches. There are many aspects to bicycle theory of risk management and uncertainty but injury prevention, helmet use is just one it is not a critical systematic review. Infor- technique, one which has proven eVective. mation on the United States experience with seat belt legislation is omitted from the discus- 1 Thompson DC, Rivara FP, Thompson R. Helmets for pre- sion. The reason provided in reference number venting head and facial injuries in bicyclists (Cochrane 3, is that “calls by Cato staV to the National review). The Cochrane Library, Issue 3, 2000. Oxford: Update Software. Highway TraYc Safety Administration to 2 Wilde G. Target risk. Toronto: PDE Publications, 1994. obtain research results about how many lives 3 Adams J. Risk. London: UCL Press, 1995. 4 Adams J. Cars, cholera and cows: the management of risk and have been saved through seat belt use were uncertainty. Washington, DC: Cato Institute, No 335, unsuccessful”.4 However, this information is 1999. http://www.cato.org/pubs/pas/pa-335es.html

www.injuryprevention.com 88 Thompson, Thompson, Rivara

5 Keatinge R. Comment 3, Cochrane injuries group web site concluded that the law had reduced injuries 2001. http://www.cochrane-injuries.ich.ucl.ac.uk/ HelmetComment.htm. and fatalities. The fact that in Sweden as a 6 Hillman M. Comment 4, Cochrane injuries group web site whole the number of deaths and injuries suf- 2001. http://www.cochrane-injuries.ich.ucl.ac.uk/ HelmetComment.htm. fered by car occupants increased after the 7 Hillman M. Cycle helmets: the case for and against. London: law suggests that it must have been possible Policy Studies Institute, 1993. 8 Vulcan AP, Cameron MH, Watson WL. Mandatory bicycle to find other sets of hospitals which found helmet use: experience in Victoria, Australia. World J Surg the opposite result.2 1992;16:389–97. 9 Carr D, Skalova M, Cameron M. Evaluation of the bicycle x Claims based on statistical significance. One helmet wearing law in Victoria during its first four years.No time in 20 researchers who use the conven- 76. Melbourne: Monash University Accident Research Center, 1995. tional 5% test of significance will find 10 Pitt WR, Thomas S, Nixon J, et al. Trends in head injuries “significance” by chance. Given the well among child bicyclists. BMJ 1994;309:877. 11 Ekman R, Schelp L, Welander G, et al. Can a combination known desire of researchers to “prove” their of local, regional and national information substantially hypotheses, one in 20 is probably a generous increase bicycle-helmet wearing and reduce injuries? Experience from Sweden. Accid Anal Prev 1997;29:321–8. estimate of the ratio of tests published to 12 Rivara FP, Thompson DC, Patterson MQ, et al. Prevention tests done. of bicycle-related injuries: helmets, education, and legislation. Annu Rev Public Health 1998:19:293–318. x Studies that ignore system eVects. Prohibit- 13 ScuVham P, Alsop J, Cryer C, et al. Head injuries to ing motorcycling, for example, would pre- bicyclists and the New Zealand bicycle helmet law. Accid Anal Prev 2000;32:565–73. vent the loss of lives in motorcycle accidents. 14 Evans L. TraYc safety and the driver. New York, NY: Van Nostrand Reinhold, 199l. But to demonstrate that it had saved lives 15 General Accounting OYce. Highway safety: motorcycle helmet one would have to track the alternative forms laws save lives and reduce costs to society. Washington, DC: US General Accounting OYce, 1991. of transport and thrill seeking adopted by the 16 Fleming HS, Becker ER. The impact of the Texas 1989 banned motorcyclists—everything from sky motorcycle helmet law on total and head-related fatalities, severe injuries, and overall injuries. Med Care 1992;30:832– diving to driving old bangers in a way that 45. pumps as much adrenaline. 17 Kraus JR, Peek C, McArthur DL, et al. The eVect of the 1992 Calfornia motorcycle helmet usage law on motorcy- x Empirical evidence of risk compensation. In cle crash fatalities and injuries. JAMA 1994;272:1506–11. Britain there is one cycling fatality for every 18 Evans L. Cycle helmets and the law: even when the science is clear policy decisions may still be diYcult BMJ 25 million kilometres cycled—not all of 1994;308:1521–2. them of course due to head injuries. The risk 19 US Community Preventive Services Task Force recommen- dations. MMWR Morb Mortal Wkly Rep 2001 (in press). compensating behaviour required to oVset 20 National Highway TraYc Safety Administration. Fourth the claimed benefits of helmets would report to congress: eVectiveness of occupant protection systems and their use. Washington, DC: NHTSA, May 1999. require an extra fatal error once in many 21 National Highway TraYc Safety Administration. EVective- millions of kilometres—a behavioural ness of lap/shoulder belts in the back outboard seating positions. (DOT HS 808 945 NHTSA technical report.) Washing- change unlikely to be directly observable. ton, DC: NHTSA, June 1999. Hedlund, whom they cite respectfully, hav- 22 Hedlund J. Risky business: safety regulations, risk compensation, and individual behavior. Inj Prev 2000;6:82–90. ing surveyed the debate about risk compensation, says “if experiments cannot provide useful Response from Adams and Hillman evidence, and if evaluations are contaminated We did NOT accept that bicycle helmets are by poor data and uncontrolled factors, we are eVective in reducing head injuries. We had left with theory”. The theory supporting risk hoped that by putting it in italics they would compensation is well supported by empirical have noticed our rider in the event of an accident. evidence in cases where the risks are large: tra- For evidence of their predisposition also to peze artists will attempt manoeuvres with include inadequate science supportive of their safety nets that they would not contemplate argument see our letter in the BMJ.1 without them. The contention of those who We wish them luck in their systematic review would introduce measures that would crimi- of all the tens of thousands of articles that have nalize self risk (for example, riding without a a bearing on risk compensation. In undertaking helmet) is that this eVect vanishes when the risk this review we suggest that in devising inclusion is smaller. Given the dismal record of the pro- and exclusion criteria for their more refined fil- hibitionists, we suggest that the burden of proof ters they be wary of the following1: lies with them. x Studies that deal with small subsets of populations—such as the oft cited study of 1 Adams J, Hillman M. Bicycle helmets. BMJ admissions to 16 hospitals in Sweden after 2001;322:1063–4. the introduction of a seat belt law, which 2 Adams J. Risk. London: UCL Press Ltd, 1995: 113–34.