Risk compensation in children’s activities: A pilot study

Abstract

BACKGROUND:

The intent of protective equipment (PE) in sports and leisure activities is to reduce injuries. However, some postulate that any safety measure prompts riskier behaviour, a phenomenon known as ‘risk homeostasis’ or ‘risk compensation.’ This study explores one approach to examining this in children. The rationale for this pilot study was to establish if children between six and 16 years old could answer questions about risk-taking sensibly and which questions, if any, could be eliminated; to establish the reliability of response; and to determine the numbers needed for a definitive study.

METHODS:

Sixty-three children with nonsevere injuries, ages six to 16 years, were interviewed while waiting to be seen at the Montreal Children’s Hospital emergency department. An interviewer administered a questionnaire comprising three sections. The first part only applied to those who were injured in an activity for which some form of PE was available (n=19). The second part examined customary risk-taking behaviour using the thrill and adventure seeking scales of a standardized questionnaire (Zuckerman) (n=63). The third section posed hypothetical questions about likely risk-taking when using PE to those who had engaged in such activities (n=58).

RESULTS:

The approach and questionnaire proved feasible with this age group. The responses suggest that children wearing PE were more likely to report increased risk-taking than those who did not wear PE. For most of the hypothetical questions, the majority also reported changes toward riskier behaviour when using PE. However, those wearing PE scored lower on the thrill and adventure seeking scale, suggesting that they are, by nature, less venturesome.

CONCLUSION:

The results indicate that risk compensation may modify the effectiveness of PE for children engaged in sports and leisure activities. Conversely, the findings also suggest that those wearing PE may be a cautious subgroup.

Children are frequently injured in sports or leisure activities and these injuries may be serious. However, for many activities protective equipment (PE) such as bicycle helmets are available to reduce the number or severity of these injuries (1–6). Nevertheless, skeptics theorize that the use of PE may make wearers feel safer and prompt them to take more risks. This theory is known as ‘risk compensation’ or ‘risk homeostasis’ (7–9).

Risk compensation (RC) postulates that humans have a built-in level of acceptable risk-taking and that our behaviour adjusts to this level in a homeostatic manner (8). The positive result is that when we perceive danger, we are inclined to be more cautious such as by driving more slowly in a snowstorm. The corollary argues that, when a safety measure is introduced, more risks are taken (10,11).

Much of the literature purportedly in support of RC uses aggregated epidemiological data from road safety studies (eg, the effect of seat belt or motorcycle helmet laws on injury rates). There is, however, much disagreement (12–14). Proponents of RC argue from these examples that safety measures increase the confidence of drivers and thus prompt them to take other risks. Some of the studies cited by Wilde (9) note an increase in driving speeds after seat-belt legislation was introduced. Although a decrease in the per capita death rate per crash with the use of seatbelts followed, they found no overall decrease in motor vehicle death rates, presumably because other types of fatal injuries (eg, involving pedestrians) increased. Others arrive at a different conclusion (15).

In addition to the many questions about the scientific merit of these studies (13,14), there is reason to question whether risk homeostasis applies in other situations, and specifically, to children. Although one recent report concludes that parents are more likely to allow children to engage in riskier activities when the child uses PE (16), few studies have attempted to examine the issue directly with children. We thought that by asking children specific questions in a blinded manner about their perceived behaviour with and without the use of PE, it may be possible to do so.

The main goal of this study was to determine if children between the ages of six and 16 could respond well to direct questions asking about risk-taking when protective equipment is used and when it is not. We also wanted to establish if such questions could be asked in a hypothetical manner and to find questions that could be eliminated. Finally, we wanted to estimate the reproducibility of responses and to determine the number of subjects ultimately needed for a definitive study.

We believe it is essential to study this issue because it is not likely that many injury prevention programs take RC into account. If the phenomenon genuinely exists, radical new approaches to injury prevention involving fundamental changes in views towards risk-taking would be needed.

SAMPLE AND METHODS

The Montreal Children’s Hospital emergency department is divided into a medical and surgical section. The participants for this study were recruited from the surgical emergency department. Inclusion criteria included injured children ages six to 16, whether or not PE was available for the activity in which they were injured. Exclusion criteria included those younger than six years of age and those who went directly to the crash room or who were judged to be in sufficient distress to make the administration of the questionnaire difficult.

Informed consent forms following written and verbal explanations were sought from all parents as well as from children over the age of seven. We also sought permission for the interviewer to call one week later for a repeat interview to determine the questionnaire’s reliability.

The questionnaire established baseline characteristics including age, sex, activity and use of PE at the time of injury. For those injured in activities for which protective equipment is available and recommended (bicycling, scooter riding, skateboarding, football, hockey), we added five items with yes or no responses to determine whether they thought they had engaged in more dangerous behaviour on the day of the injury. The interviewer then asked if the child was wearing PE at the time of the injury and was thus blinded with respect to whether PE was used, and the child was blinded to the specific purpose of the study.

All children were then asked 10 forced-choice questions from Zuckerman’s Thrill and Adventure-Seeking subscale (17). This scale has been validated; higher scores are indicative of risk-taking propensity. Finally, hypothetical questions were asked of all children having worn PE on any previous occasion (other than the day of the injury), to explore whether they thought they would change their behaviour when wearing PE.

The Kappa statistic was used to assess reproducibility; that is, to see if the same answers would be given at a later date and when administered on the telephone.

RESULTS

Eleven children or their parents refused the initial interview and 4 refused the second interview. Fifty-nine per cent of the respondents were boys between seven and 16 years and the median age for both sexes was 10.5 years (Table 1). Only 19 children were treated for an injury resulting from activities for which PE was available (bicycling, football or hockey). The remainder had injuries in situations for which there was no PE. Fifty-eight children answered the hypothetical questions and all 63 children responded to the Zuckerman subscale.

TABLE 1.

Baseline demographic characteristics (n=63)

	n	Percentage
Sex
Boys	37	59
Girls	26	41
Age (years)
6–9	13	21
10–12	28	44
13–16	15	24
missing	7	11
Injured: PE-related activity	19	30
Bike/skateboard/scooter	16	84
Football	2	11
Hockey	1	5
Injured: Non-PE activity	44	70
Used PE previously	58	92
Bike/skateboard/scooter	57	98
Hockey	18	31
Football	9	23
Zuckerman TAS subscale	63	100

PE Protective equipment; TAS Thrill and adventure seeking

Of the 19 children injured in a PE related activity, five were PE users and their responses were compared with those of the 14 who did not use PE. In general, more users gave responses suggestive of RC than did nonusers. For example, four of the five users (80%) reported riding faster versus five of the 14 nonusers (36%) (Table 2). Riding near busy roads showed little difference and too few children were injured in hockey or football to include in a separate analysis. As for the general question asked to the 19 children referred to above, three of the five users (60%) reported behaving in a ‘more dangerous than usual’ manner compared with only three of the 14 nonusers (21%) (Table 2). For the 10 specific hypothetical questions, there was considerable variability in the responses and no pattern. However, in response to the general hypothetical question, more than 50% reported an inclination to change behaviour with PE use (Table 3).

TABLE 2.

Protective equipment users versus nonusers: Responses to risk-taking questions

Reported behaviours*	PE users (n=5) n (%)	Non-PE users (n=14) n (%)
Riding faster (n=16)	4 (80)	5 (46)†
Damage to bike, etc. (n=16)	3 (60)	4 (36)†
Riding near busy roads (n=16)	1 (20)	2 (18)†
More dangerous than usual (n=19)	3 (60)	3 (21)
Zuckerman Score >5 (n=19)	1 (20)	7 (50)

^*Responses to questions about behaviour on day of injury;

^†based on n=11 road users. PE protective equipment

TABLE 3.

Response to hypothetical questions about behaviour assuming protective equipment used

Sports related behaviours	Yes (n [%])	No (n [%])	Unknown (n [%])
Bicycle: (n=57)
Ride faster	22 (39)	35 (61)	0 (0)
Ride near busy roads	16 (28)	40 (70)	1(2)
Hockey: (n=18)
Cross checking	8 (44)	5 (28)	5 (28)
Block shot with body	14 (78)	4 (22)	0 (0)
Football: (n=9)
Spearing	4 (44)	5 (56)	0 (0)
Tackle harder	7 (78)	2 (22)	0 (0)
General: (n=58)
More confident	51 (88)	7 (12)	0 (0)
Feel safer	55 (95)	3 (5)	0 (0)
More aggressive	33 (57)	25 (43)	1 (2)
Try more dangerous	34 (59)	24 (41)	0 (0)

Hypothetical questions asked to children who have engaged in such an activity at a previous date. Questions 1–2 relate to road activities (biking, in-line skating, skateboarding, scooter riding); 3–4 relate to hockey; 5–6 relate to football; 7–10 are general questions

Reproducibility

Twenty-three families were telephoned within a week of their initial visit to establish the questionnaire’s reproducibility. This was assessed using the Kappa statistic, an aggregate measure for all questions. There was perfect agreement for all nonhypothetical questions, and somewhat lower agreement for the hypothetical questions.

Question elimination

All questions seemed worth retaining, although in future studies, the added benefits of the Zuckerman scale may depend on the specific questions being asked.

Sample size conclusions

Assuming the same pattern of responses in a definitive study, and equal numbers in users and nonusers, about 200 subjects (or more) would be required to detect such differences with alpha=.05 and power (1-beta)=.80.

DISCUSSION

RC is a theory that is highly contentious but of utmost importance to injury prevention strategists. As described by Thompson et al (13), “Briefly put, risk compensation theory suggests that individuals provided with a protective device such as a bicycle helmet or an automobile seat belt will act in a riskier manner because of the sense of increased protection from the (device) and thereby nullify the protection afforded.” Accordingly, one proponent has even argued that seat belts are more harmful than helpful (7), although others have found no such evidence (15).

In fact, the implications of the theory extend beyond protective devices as is evident in the contention that drivers of sports utility vehicles are more aggressive because the size of the vehicle makes them feel safer. Although continued declines in injury mortality rates in most countries make it difficult to accept the widespread offsetting effects that the theory predicts, there are nonetheless many ardent believers and some who declare that the evidence for risk compensation is ‘overwhelming’ (12). However, Thompson et al (13) in their examination of the empirical evidence reach precisely the opposite conclusion: that at least with respect to bicycle and motorcycle riders and car drivers and passengers, “supportive evidence is limited if not absent.” Others hold a more balanced view: Hedlund (10) states that “the evidence is overwhelming that every safety law or regulation is not counterbalanced by compensating behaviour,” but as supporters of RC note, he also asserts that “some laws and regulations as well as safety measures voluntarily adopted, are counterbalanced by compensating behaviour.” In short, the debate is far from resolved.

We are only aware of two other studies that examine RC with respect to children. Both used questionnaires administered to the mother, and the second also had children complete the measure (16,18). The results indicate that PE may result in riskier behaviour, albeit in lab-like situations. Ours is the first study to attempt to test the RC theory using children who have been injured. The responses show a tendency toward agreement with the RC hypothesis. For example, in general, children using PE such as helmets report going faster and more often reported damage to their bike, perhaps because wearers took risks resulting in more severe crashes. However, other behaviours showed few differences. Such results suggest that the RC theory may operate in a selective way with respect to the individual or with respect to particular activities.

The main limitation of our study is the small sample. Consequently, formal statistical analysis was not justified and no firm conclusions could be drawn. There are, however, other limitations that we acknowledge.

Reported behaviour may differ considerably from observed behaviour. Some studies suggest that adults tend to over-report safety behaviours (19), or that parent-reported safety behaviours fail to predict unintentional injuries in children (20). Other investigators, however, suggest good concordance between reported and observed safety behaviour (21,22). We cannot say if reported versus actual behaviour is different for PE users compared with nonusers. For most questions, there was good reliability but their validity remains uncertain.

Children wearing PE at the time of the injury were more likely to score lower on the Zuckerman scale, suggesting that although the more protected subgroup thought that they were engaging in more dangerous activities, they are generally more cautious. This may further reinforce their more cautious approach to risk-taking, resulting in positive answers to the questions about riding faster and behaving more dangerously following an injury. Therefore, it may be the child’s perception of his or her activity following an injury that prompted the responses rather than the use of PE at the time.

The hypothetical questions showed considerable variability in responses. It seems that only a small proportion of children believed they performed specific activities differently when wearing safety equipment. However, a considerable proportion of the same children reported a change with respect to their general feelings and behaviours when wearing safety equipment. Assuming that the responses to such questions are valid and that the sample is reasonably representative, this suggests that RC may occur in some but not all children.

Another limitation is the exclusion of individuals who went to the crash room or were too severely injured to be interviewed. If these individuals were the true ‘risk compensators,’ then we would have underestimated the effect.

Evaluating RC in sports where PE is mandatory (eg, helmets in ice hockey) is particularly difficult if children never have occasion to participate without the equipment. Further, these results may be confounded if all those wearing PE were participating in organized league play, whereas all non-PE participation was, for example, on outdoor rinks with little supervision. Then recollection of behaviour with and without a helmet would be confounded by the type of participation.

Clearly, a larger study with appropriate statistical power, focused on particular activities where PE use is not mandatory, including those at the upper end of the injury severity spectrum and with attention to the issue of validation of reported versus actual behaviour, is needed. If it yielded similar findings, we would need to think differently about prevention strategies. In particular, those planning preventive programs would need to take into account the likelihood of RC and to find ways to persuade children to lower their level of acceptable risk-taking.