Article Text

A systematic review of correct bicycle helmet use: how varying definitions and study quality influence the results
1. R S Lee1,
2. B E Hagel2,
3. M Karkhaneh3,
4. B H Rowe3,4
1. 1
O’Brien Centre for the Bachelor of Health Sciences Program, University of Calgary, Calgary, Alberta, Canada
2. 2
Departments of Paediatrics and Community Health Sciences, Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada
3. 3
School of Public Health, University of Alberta, Edmonton, Alberta, Canada
4. 4
Department of Emergency Medicine, University of Alberta, Edmonton, Alberta, Canada
1. Dr B Hagel, C4-434, Alberta Children’s Hospital, 2888 Shaganappi Trail NW, Calgary, Alberta, Canada T3B 6A8; brent.hagel{at}calgaryhealthregion.ca

## Abstract

Background: Bicycle helmets effectively reduce the risk of bicycle-related head injuries and trauma; however, they must fit properly to be effective. Little is known about the prevalence of correctly worn helmets and factors associated with proper helmet use.

Objective: To examine proper bicycle helmet use through a systematic review.

Methods: Comprehensive searches of electronic medical databases were performed, and completed by grey literature and reference list checks to identify eligible studies. Studies eligible for inclusion had to involve cyclists and report on the prevalence of correct or incorrect helmet use. Two reviewers independently selected studies and data were extracted regarding the prevalence and factors influencing proper helmet wearing of cyclists.

Results: An inclusive search strategy led to 2285 prescreened citations; 11 of the studies were finally included in the review. Overall, correct helmet use varied from 46% to 100%, depending on the criteria used by researchers to define proper helmet use; stricter criteria reduced the proportion of properly worn helmets. Adulthood, female sex and educational interventions were associated with correct helmet use in some studies. Self-reported poor helmet fit (OR = 1.96; 95% CI 1.10 to 3.75), posterior positioning of helmet (OR = 1.52; 95% CI 1.02 to 2.26) and helmet loss in crash (OR = 3.25; 95% CI 1.82 to 5.75) increased the risk of head injury. In addition, educational programmes on helmet use in schools increased correct helmet use among schoolchildren.

Conclusions: This systematic review outlines the current state of the literature including the variability in research methodology and definitions used to study proper helmet-wearing behaviour among cyclists.

View Full Text

## Statistics from Altmetric.com

Bicycling is a universally popular means of transportation, outdoor recreational activity and exercise enjoyed by all age groups, especially young children and adolescents.12 Given this popularity, it is not surprising that cycling injuries can and do occur. Head injuries account for ∼20% of cycling injuries reported to emergency departments, and they make up most of the injuries that result in fatalities.34 Two systematic reviews have shown that bicycle helmets significantly reduce the risk of head, brain and even facial injuries.56 Consequently, helmet promotion as an intervention is an integral part of most bicycle injury prevention strategies.2

### Inclusion/exclusion

Study eligibility was checked by two reviewers (RL, BH) using standardised inclusion and exclusion criteria. The inclusion criteria were as follows: original data; study population included bicyclists; outcome measures (reports on the prevalence of helmet use and the prevalence of correct/incorrect helmet use); and description of cyclist characteristics (reports on age, sex, location, riding companion, their helmet use, behavioural and geographic factors). Studies were excluded from the review if they failed to provide original data (eg, review articles, case study reports, observational studies with no data analysis or comparison) or failed to report the prevalence of CHU.

### Selection process

Firstly, the titles and abstracts of the initially identified citations were screened by one reviewer (RL). Then, a list of potentially relevant studies was developed and the full papers were retrieved. Two reviewers (RL, BH) independently reviewed the full papers of the potentially relevant studies for inclusion. Disagreements were resolved by repeat review and consensus of the reviewers or third-party adjudication, if needed.

### Quality assessment

The Downs and Black checklist was used to evaluate the methodological quality of the included studies in the systematic review. The modified checklist included 20 applicable questions covering the subscales reporting: external validity, bias, confounding and power (appendix A, an online supplement).12

### Statistical analyses

Data from individual studies were extracted and tabulated. Sources of heterogeneity between studies were explored including age, sex, location of study, year of study, non-intervention versus intervention studies, and studies examining helmet fit and injury risk. No formal statistical pooling was performed because of the heterogeneity of the included studies. Assessment of publication bias was not possible because of lack of information.

## RESULTS

### Selection of studies

Using an inclusive search strategy described, 2285 citations were identified. From these citations, 61 were found to be potentially relevant (appendix B, an online supplement). After independent review of the full reports, 11 citations were included in the final review.137111316 Studies were excluded primarily because they failed to report on the prevalence of correct or incorrect helmet use. Disagreements were uncommon; one study7 required repeat evaluation and a consensus decision; none required third-party adjudication.

### Description of studies

Of the 11 studies, all were peer-reviewed publications, except one, which was published in an academic report.14 Six studies were conducted in the USA,7101516 three in Canada,1311 and two in Australia.1314 Six studies were observational surveys on the prevalence of bicycle helmet use,13891114 two studies evaluated an educational intervention on bicycle helmet use by schoolchildren,1316 one study evaluated helmet fit during a well-child visit in a doctor’s office,10 and two studies examined the influence of helmet fit on head injury risk.715Tables 1, 2 and 3 provide detailed information on each study.

Table 1 General descriptions of the included studies
Table 2 Results from cross-sectional surveys and intervention studies by year of study
Table 3 Results from helmet evaluation studies

### Quality assessment

It was difficult to conduct the quality assessment, as most of the studies had a primary outcome of helmet prevalence, with CHU prevalence as an ancillary outcome. Added to this, there were three different types of cross-sectional, case–control and interventional studies with different criteria. All studies met the criteria for clearly describing the study’s objectives, main outcomes and main findings; however, in six studies,7810131516 there was limited reporting of subject characteristics, making it challenging to assess issues of external validity. Only a few studies actually provided an assessment of random error (in the form of confidence limits or p values) for the estimates of CHU.137101115 There was little attempt to capture and examine the characteristics of those with, versus those without, proper helmet fit, apart from the case–control study on helmet fit and risk of head injury by Rivara et al7 and the stratification by age undertaken in the cross-sectional study by Harlos et al.11

### Overall assessment

A review of included studies involving more than 15 000 cyclists found a substantial variation in the prevalence of CHU. In one repeat cross-sectional survey in which CHU was defined as fastened helmet chin strap, the prevalence of properly worn helmets was 94–100%, with little difference between 1991 and 1992.14 Three other studies, which applied more than one criterion, reported the prevalence of CHU to be 72.6–90.7%.1911 Another study showed that older age groups had a higher prevalence of CHU (96.8% for ages >19 vs 88.9% for ages 16–19).11 Two studies with similar and more rigid criteria of CHU reported the prevalence of properly worn helmets to be 46–49%.38 One study conducted during a well-child visit to a paediatrician showed that only 4% of families had no errors in a timed attempt to fit a helmet.10

Of the two reports based on data from a case–control study of helmet fit, one noted a discrepancy between expert (21%) and self-reported (66%) ratings of proper helmet fit.15 Self-reported helmet fit was reported to be excellent by 66% of the non-head-injured control group, with the helmet centred on the head in 85%.7

### Heterogeneity

There was considerable heterogeneity among the studies, and this was explored in the following subgroup analyses.

#### Age

Of the studies that reported CHU by age group, one found little difference between children (5–11 years), adolescents (12–17 years) and adults (18+ years) (table 2).14 Other reports indicated a greater prevalence of CHU among adults compared with children, although the actual prevalence estimates showed substantial inter-study variability (fig 1).811

Figure 1

Prevalence (Pr) of correct helmet use by age estimated from logistic regression analysis. The data of Finch et al14 on age as a proportion of correct helmet use ranged from 92% to 100% and were not included in the analysis. (A) Data from Harlos et al.11 (B) Data estimated from proportions and totals provided in Twomey et al.8

#### Sex

From the two studies that provided a sex distribution for CHU, one reported little difference between males and females, with all estimates ranging from 94% to 100% (table 2).14 On the basis of a randomised intervention trial intervention and control groups at baseline, data from Hall et al13 indicate that the prevalence of CHU was 42% (95% CI 37% to 48%) for girls and 27% (95% CI 24% to 31%) for boys before a peer-led classroom curriculum intervention on CHU (fig 2).

Figure 2

Prevalence (pr) of correct helmet use by sex for children 10–11 years old. Data from Hall et al.13

#### Location

Only one study14 provided data for evaluation of CHU by location. It reported that there was no relation between CHU and footpaths (94–99% correct depending on age group) and roads (96–99% depending on age group). Two other studies1316 restricted observations to school sites: in the former, CHU was 32% pre-intervention and in the latter CHU ranged from 9% to 64% at the baseline assessment of an educational intervention.

#### Year of study

All included studies were conducted from 1993 to 2003 except one16 in 2007. Figure 3 shows predicted versus actual prevalence of CHU by calendar year. The overall trend based on a logistic regression model fitting calendar year to predict the logit (ie, logarithm of the odds) of CHU prevalence suggests a general decline since 1995 (p<0.001).

Figure 3

Actual versus predicted prevalence (Pr) of correct helmet use by calendar year and country. Data from Finch et al,14 Thompson et al15 and Van Houten et al16 were not included as absolute numbers not available for entry into the logistic regression model. Data from Rivara et al were based on self-reported fit  =  excellent. Logistic regression model: logarithm of the odds (correct) ∼ study year.

#### Intervention

Two studies1316 examined the influence of educational interventions on the prevalence of helmet use and CHU. In one that consisted of peer data collection of helmet use, education on wearing a helmet properly, goal setting, public posting of correctly worn helmet percentages, and free helmet giveaways, researchers observed a substantial increase in the prevalence of CHU after implementation. The improvements in CHU prevalence were from 9% to 40%, from 64% to 80% and from 30% to 78%, respectively, for the three participating schools (table 2).16 In a similar study that used a whole-school intervention consisting of the development and review of the school’s road safety policy, home activities for students’ families, educational resources for students and a peer-led classroom curriculum for grade 5 and 6 students, the CHU was recorded before the intervention, but not after.13 This study found a relatively low prevalence of CHU at the pre-intervention stage for both control (32%) and intervention (27%) groups. It also reported a higher prevalence for girls than boys at the 27 participating schools (control group: boys = 27%, girls = 40%; intervention group: boys = 22%, girls = 38%).

#### Helmet fit and risk of injury

One study examined the relationship between helmet fit and the risk of head injury. The authors reported an almost twofold greater risk of head injury with helmets that fit poorly compared with those with excellent fit (odds ratio (OR) = 1.96; 95% CI 1.10 to 3.75). This study also suggested a significantly higher risk of head injury with helmets tipped posteriorly (OR = 1.52; 95% CI 1.02 to 2.26) and with a helmet that came off in a crash (OR = 3.25; 95% CI 1.82 to 5.75) (table 3).7

## DISCUSSION

This comprehensive systematic review presents the evidence from the published literature on proper helmet-wearing behaviour. Using an exhaustive search of electronic databases and grey literature sources, this review identified 11 studies reporting on correct cycle helmet use. These studies involved more than 15 000 cyclists, and thus results from this review provide valuable insight into the prevalence of, and factors associated with, correct bicycle helmet use.

There was a wide variation in the prevalence of CHU among included studies; from relatively low (46–49%)38 to high (72.6–100%).191114 This discrepancy may be explained by the variety of definitions used by researchers to assess CHU. The study reporting the highest rate of correct use assessed proper use with the criterion of a fastened helmet chin strap,14 but most studies used more than one criterion, including assessment of the positioning of the helmet on the head and helmet chin straps. The helmet must sit level on the forehead and be positioned squarely on the head without exposing the forehead or the back of the head. In addition, two studies310 also included an extra criterion for the stability of the helmet in the definition of CHU. Overall, the studies using more than one criterion for defining helmet fit reported a lower proportion of CHU.138111316

Females and adults were found to have a higher prevalence of correctly worn helmets.111314 It was reassuring to observe that changes including those interventions involving in-school programmes and educational campaigns positively influenced the prevalence of CHU by children.16

#### What is already known on this topic

• Bicycle helmets effectively prevent head and brain injuries from bicycle collisions.

• Improper helmet fit and wearing position increases the risk of head and brain injuries.

• Little is known about the factors that affect correct helmet use (CHU).

• The prevalence of proper helmet wearing among cyclists varies from 46% to 100% depending on the nature of the study and CHU definition.

• There is some evidence that the age and sex of the cyclist influences CHU.

• School-based helmet educational interventions appear to influence CHU in children.

• A number of methodological issues in the evaluation of CHU were identified that can guide future research in this area.

Discrepancy between self-report and expert assessment, reported in two studies,1015 revealed that proper fit perceived by study participants was higher than the actual helmet fit assessed by experts. Findings on the prevalence of CHU, factors influencing CHU, and the difference between perceived and actual helmet fit are important when considering the increased risk of head injury as a result of poorly fitting helmets.7 It seems that these measures must be taken into account in order to influence cyclists’ behaviour in proper helmet wearing and safe cycling.

### Limitations

There are limitations to this systematic review. We limited our review to English-language studies and did not contact authors for any unpublished work. The published articles used a variety of definitions for CHU, leading to differences in the results of CHU prevalence. All included studies were conducted in Canada, the USA or Australia, potentially limiting the generalisability of our findings accordingly. A small sample size in some studies may limit the power to detect associations of cycling characteristics with the prevalence of CHU.11 Many of the studies involved unobtrusive and direct observations of cyclists at certain sites associated with fluctuations in the number of cyclists throughout the day, times of high volume of riders, and large groups of cyclists at one time. This potentially leads to an inability of observers to collect data on every cyclist encountered and to accurately record all the required information.39 Most of the studies were conducted in urban or suburban areas, limiting generalisability to rural locations.

## IMPLICATIONS FOR INJURY PREVENTION

This systematic review revealed few studies specifically examining the prevalence of, and factors associated with, CHU. Most included studies focused primarily on the prevalence of helmet use, with the inclusion of some data on the prevalence of CHU. With the variety of CHU definitions used by each study, more rigorous, consistent definitions of CHU would improve our understanding. With attributed risk of head injury due to improper helmet fit,7 more timely research on CHU is urgently needed to enable researchers, policy makers and other stakeholders to target the risk factors associated with improper helmet use. The results of interventional studies also suggest that school educational programmes on proper helmet wearing are effective in increasing the prevalence of CHU among schoolchildren.1316 Thus, we suggest incorporating CHU in any bicycle helmet programme including school-based education programmes, programmes involving retailers selling bicycle equipment, and those in paediatric care centres.

## CONCLUSIONS

Although proper fitting of bicycle helmets has been shown to reduce the risk of bicycle-related head injuries, the evidence indicates that the prevalence of proper helmet wearing among cyclists varies from 46% to 100% depending on the definition of CHU. This review also found an existing relationship between cyclist characteristics, including age and sex, and CHU. In addition, school-based helmet educational interventions appear to be effective in influencing children’s behaviour, although more research is required on the prevalence and factors associated with proper helmet use.

View Abstract

## Footnotes

• Appendices A and B are published online only at http://injuryprevention.bmj.com/content/vol15/issue2

• Funding: BH holds the position of Professorship in Child Health and Wellness funded by the Alberta Children’s Hospital Foundation, through the support of an anonymous donor and Canadian National Railway Company. In addition, he is supported by a Population Health Investigator Award from the Alberta Heritage Foundation for Medical Research and a New Investigator Award from the Canadian Institutes of Health Research. BR is supported by a 21st Century Research Chair in Emergency Medicine from the Government of Canada (Ottawa, ON).

• Competing interests: BR was the lead investigator and a co-investigator on two of the included studies; however, he did not complete the quality assessment. One included study is Alberta-based. BH is the guarantor of this study. All authors fulfil the criteria for authorship. No one else fulfils the criteria for authorship.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.