Protocol•

Protocol of a cluster-randomised trial to improve adolescent bicycling safety education programme efficacy

Abstract

Introduction Second to motor vehicles, bicycles contribute to more childhood injuries than any other consumer product. Youth bicycling safety education programmes are one avenue of prevention, but despite a plethora of programmes, little is known about their impact on bicycling safety behaviour. This paper summarises a cluster-randomised trial to evaluate the impact of a bicycle safety education programme for young adolescents.

Methods and analysis Adolescents aged 9–12 years old who bike at least two times per week and a parent/guardian are recruited via email listservs, flyers and word of mouth. Participants are screened for eligibility before they are consented/assented into the study. Adolescent-parent dyads are randomised to one of the three study arms (Control, Bike Club, Bike Club Plus) based on the school they attend. The Bike Club intervention arm evaluates the bicycle safety club programme and the Bike Club Plus intervention arm evaluates the addition of an active parent engagement component. Control group participants receive no intervention, but they are told that other participants in the study will receive the education intervention so that they understand why they are asked to record their bicycling trips for two separate weeks. The intervention involves 12 hours of classroom and on-bike lessons covering a variety of bike safety skills for independent riding in mixed traffic. Adolescents record their bike rides using a Global Positioning System (GPS)/video camera system for 2 weeks, 1 week before and 1 week after the education programme for participants in the intervention arms and approximately 1 month between data collection weeks for participants in the control group. All participants complete baseline and follow-up surveys. Primary outcomes include rates safety-relevant events (eg, crashes, near crashes) and safety behaviours (eg, helmet use, traffic law violations, hazard avoidance) per 100 miles/min ridden, and the proportion of positive and negative behaviours relative to the overall number of instances of a given event behaviour. Secondary outcomes include differences in knowledge test scores, mean ratings of perceptions, mean ratings of self-efficacy, and identification and recall of cues to action related to bicycling safety.

Ethics and dissemination This study has ethical approval from the University of Iowa Institutional Review Board (IRB# 202105148). Study results will be disseminated through presentations at national and international scientific conferences, peer-reviewed manuscripts, outreach to stakeholders and digital media outlets.

Trial registration number NCT05265689.

What is already known on this topic

  • There are a plethora of youth bicycle safety education programmes, but evaluations have primarily focused on knowledge change only, as opposed to behaviour change.

  • There is no gold standard programme or set of core bicycling competencies to date.

What this study adds

  • This study uses a naturalistic approach with an innovative video/GPS data acquisition system to examine behaviour change following completion of a comprehensive bicycle safety education programme.

  • This study also evaluates the impact of adding a parent training component to the programme on youth bicycling safety behaviours.

How this study might affect research, practice or policy

  • Results from this study can be used to develop a gold standard bicycling safety programme and accepted set of core competencies for youth bicycle safety education that can be implemented by the practice community.

  • The naturalistic methods and protocols used in this study can be used for future road safety intervention research.

Introduction

Bicycling affords young adolescents one of their first opportunities to independently navigate the challenges of the traffic environment. As such, bicycling has many physical and neurodevelopmental benefits. However, bicycling is also the most common cause of recreation-related head injuries, and bicycles are a leading contributor to childhood injuries.1 2 In the USA, bicycling injuries among youth 19 and younger lead to more than 135 000 emergency department visits and 7700 hospitalisations annually.3 School and community-based youth bicycle safety education programmes teaching bicycle handling and traffic safety skills are abundant but rarely evaluated.4 The few bicycle safety education programmes that have been evaluated only measured knowledge change or had limited evaluation of behaviour change (eg, helmet fit).4 As such, there is a gap in evidence related to the impact of bicycle safety education on bicycling safety behaviour change during real-world riding.

Beyond school or community-based education, parents are typically a main source of safety information and teacher/modeller of safety-related behaviours throughout childhood.5–8 However, parents often do not use their position to its full potential in reducing their child’s injury risk.7 Parents often lack knowledge of bicycling safety best practices and very few existing bicycle safety education programmes include active parent engagement.4 9

Examination of changes in real-world riding and active parent engagement are key areas addressed by this ongoing study. Real-world bicycling behaviours pre- and post-intervention are captured using an innovative video/GPS system called Pedal Portal 2 and annotated with a graphical user interface (GUI) using a protocol developed specifically for the outcomes of this trial. Additionally, parents are engaged in the intervention in one study arm through a 30-minute virtual training session to evaluate the potential for added impact on behaviour change.

Objectives

The two aims of this cluster-randomised trial are to (1) determine if early adolescents in two bicycle safety education intervention groups (Bike Club and Bike Club Plus with an added parent engagement and training component) have increased safety behaviours (eg, scanning for traffic) and reduced safety-relevant events (eg, crash, near crash) compared with the control group that does not participate in the education programme and (2) determine if early adolescents in two bicycle safety education intervention groups have improved bicycling safety knowledge (eg, hand signals), perceptions (eg, susceptibility to injury) and self-efficacy efficacy (eg, confidence in choosing a safe route) compared with the control group. Safety behaviours and events will be measured through an objective on-bicycle video system with tailored outcome coding, called Pedal Portal 2, adapted and enhanced from the original Pedal Portal.10 We hypothesise that intervention group participants (Bike Club or Bike Club Plus) will have increased safety behaviours and reduced safety-relevant events compared with the control group, and that participants whose parents are engaged and trained (Bike Club Plus) will have better outcomes than those whose are not (Bike Club). As measured through pre and post surveys, it is hypothesised that intervention group participants will also have improved bicycling safety knowledge, perceptions and self-efficacy compared with the control group, and Bike Club Plus participants will have greater improvement in those outcomes as compared with Bike Club participants. If evidence shows increased bicycling safety (behaviours, knowledge and perceptions), approaches from this study could be widely implemented to develop a gold standard adolescent bicycling safety programme.

Methods and analysis

Study design and overview

This cluster-randomised controlled trial, conducted from 2022 to 2026, evaluates a community-based bicycle safety education programme with and without an active parent training component. The target population is early adolescent bicyclists (ages 9–12) and a parent/guardian from Johnson County, Iowa. Randomisation into the three study arms occurs at the site level (ie, elementary school), with two intervention groups and one control group (see figure 1). As such, each elementary school is considered a cluster which is why this is considered a cluster-randomised trial. This study uses complete block randomisation so that each elementary school is randomly assigned to each study arm once (rotating to a different study arm each year during the 3 years of data collection) to ensure balance across study arms in environmental features of the different catchment areas (eg, road infrastructure, traffic density, etc).

Figure 1
Request permissionrequest permission icon
Figure 1

Study groups and stages of study participation.**Timing between baseline, intervention and follow-up were designed to align with what was most logistically feasible and also to capture immediate effects of the intervention.

Adolescents’ bicycles in all study groups are equipped with Pedal Portal 2, an innovative bicycle-mounted GPS/video system developed by the team to objectively observe bicycling risk exposure and behaviours (figure 2). The Pedal Portal 2 System consists of a handlebar-mounted front-facing and rear/rider-facing camera; and a microphone, along with handlebar-mounted or frame-mounted primary and backup GPS modules. The system powers on/off automatically through vibration detection. The system also includes a handlebar-mounted push button, synced with the camera and GPS data. Tamper-proof screws and labels indicating the system contains a GPS-tracker are used as theft-deterrent measures.

Figure 2
Request permissionrequest permission icon
Figure 2

A view of the Pedal Portal 2 system, consisting of (A) front-facing camera; (B) rear-/rider-facing camera, (C) event push button; (D) main GPS module; (E) microphone; (F) bag containing battery, memory card and motion sensor; (G) back-up GPS module.

System data will be coded to measure bicycling exposure (eg, hours, miles travelled), routes, and the types and rates of safety-relevant events (eg, near crashes, crashes) and behaviours (eg, following traffic rules, scanning for traffic). Participants in the intervention groups (Bike Club and Bike Club Plus) receive a 12-hour bicycle safety education programme. After the education programme is complete, participants in Bike Club Plus receive a parent training session via videoconferencing (eg, Zoom) on bicycling safety best practices, child development related to bicycling, home practice strategies and individualised feedback on their child’s bicycling performance during the education programme. The control group does not receive any programming or training.

Recruitment

Study fliers are distributed digitally via email listservs and partner newsletters, including school-specific online parent bulletins. Team members distribute fliers at community and school events and to local businesses. Team members also identify potential participants by working directly with the school programme coordinators. Eligible participants are identified through a screening process conducted via phone or email with parents/guardians. Eligibility criteria for the study include the following:

Inclusion criteria:

  • Age 9–12.

  • Lives in Johnson County, Iowa.

  • Willing to participate in bicycling safety education programme.

  • Adolescent has access to a bicycle and rides independently at least two times per week.

  • Adolescent has not previously completed a formal bicycle safety education programme.

  • No siblings already enrolled in the study (only one adolescent per family will be enrolled).

  • Fluency in English for the adolescent.

  • Fluency in English, Spanish or French for at least one parent/guardian.

  • Legal guardianship of adolescent by at least one participating adult (thus excluding wards of the state).

Exclusion criteria:

  • Does not agree to or meet one or more of the inclusion criteria.

  • Having a condition which would be a safety risk on a bicycle or performing physical activity, including but not limited to a seizure disorder or mobility issue.

Participants are randomised to a study arm at the site level. The site is determined by the child’s elementary school or home address if they are homeschooled. Responses are monitored to maintain balance in age and gender between the study arms. During conversations with parent(s), the study team (team) introduces the study and high-level participant requirements based on study arm.

Sample size calculation

To determine the sample size for Aim 1, we used data from one of our previous naturalistic riding studies11 involving a two-sample comparison of boys and girls. Based on a test to contrast group-level means, a sample size of 35 parent-adolescent dyads per group will ensure 90% power to detect a difference of 1.2 near crashes per 100 miles (2.9 vs 1.7) of riding, or 17 dyads per group to ensure 90% power to detect a difference of 2.3 errors per 100 miles (2.8 vs 5.1). The computation assumes the data in each sample arises from a Poisson distribution, where the mean equals the variance.

To determine the sample size for Aim 2, the overall pre- and post-knowledge test scores from a preliminary evaluation of Bike Club was used. In that evaluation, a mean increase of 13.7% (SD=17.5%) in overall knowledge test scores was found following the intervention. That study did not include a control group, but a mean increase of 2% is hypothesised in overall knowledge for the control group over the same time period as the intervention groups. It is also assumed that the SD in the control group will be comparable to the SD in the intervention groups. Using these data and estimating the sample needed for a two-sided t-test, 48 participants will be needed per group to ensure 90% power to detect an 11.7% mean difference in knowledge test scores between an intervention group and the control group. The sample sizes needed per group will be the same for detecting the difference between two means in a three-group analysis of variance (ANOVA) comparison.

The pool of eligible adolescents (ages 9–12) in the study area (Johnson County) greatly exceeds the minimum 144 (48 per study group) needed for adequate power and can accommodate attrition. Due to small sample sizes in our preliminary studies and lack of prior studies reporting within-cluster correlation estimates, it was not possible to account for a clustered design in the sample size calculations. To avoid making incorrect assumptions, the sample size was increased to 180 (60 dyads per study group) to indirectly account for these effects and attrition (which was less than 5% in our preliminary studies).

Informed consent/assent of participants

Eligible adolescent-parent dyads indicating interest in the study are scheduled for an initial in-person meeting with two team members at the dyad’s home or other agreed on location (eg, neighbourhood centre). At the first in-person meeting, the dyads are consented/assented.

Intervention

Approximately 2–4 weeks after baseline data collection, adolescents in the intervention study groups (Bike Club and Bike Club Plus) participate in a 12-hour bicycle safety education programme. Additionally, parents of adolescents in Bike Club Plus participate in 30-minute virtual training session soon (usually 1–2 days) after their child has completed the programme.

Bicycle safety education programme (Bike Club)

The bicycle safety education programme, Bike Club, consists of 3-hour instructional sessions for 4 consecutive days during the summer held at local elementary schools. Each programme hosts a maximum of 12 children. Two co-instructors lead the sessions with assistance from one to two additional aides.

The bicycle safety education programme is based on the Let’s GoBiking!12 programme, identified in our bicycle education inventory as one of the most comprehensive.4 The programme uses a theory-driven approach with appropriate cognitive, motor and perceptual development considerations (eg, ability to combine developing motor and cognitive skills for bicycle riding). Programme components are tied closely to targeted outcomes (table 1), including 15 classroom and on-bike modules, with pre- and post-tests, cues to action (eg, stickers, bookmarks), and passive parent engagement (tip sheets sent home to parents summarising each day’s lessons).

Table 1
|
Safety behaviours and safety-relevant events to be coded from video and GPS data and their ties to intervention components

Each module is introduced in a classroom setting through educational videos, demonstration and discussion. Adolescents then transition to practising these skills through on-bike drills in a controlled setting consisting of simulated roadway scenarios within a closed blacktop or parking lot at the school. Team members keep track of each participant’s daily progress. At the conclusion of the programme, participants receive a certificate of completion and written assessment of skill achievement, with emphasis on areas of proficiency and areas for improvement.

The Let’s GoBiking! Programme was augmented for this study to include on-road bicycle rides focused on demonstrating and practising the safety behaviours. These additional components include neighbourhood and capstone bike rides. During days two and three of the programme, the participants are taken on short (2–4 miles) neighbourhood group rides to practise these skills and a 10 mile capstone ride on the final day to demonstrate and practise safety and navigation skills. Completing rides in real-world scenarios (as opposed to protected areas only) provides encoding specificity which we hypothesise will translate to better maintenance of skills and knowledge. The premise of encoding specificity in learning theory is that information is better retained when it is taught in the context that matches where it will be executed.13 Route planning is a lesson included on day 3 using paper maps and online route-planning tools, with a focus on selection of low-traffic roads and bike facilities (eg, trails and bike lanes).

Team members collaboratively deliver the programme with community partners. The long-term goal is to train-the-trainer so local programme personnel can implement the programme independently in the future. Programmatic and bike maintenance supplies are shared among the project partners.

Parent training session (Bike Club Plus)

Following the completion of the educational programme but before the follow-up week of data collection, parents of adolescents in Bike Club Plus participate in a 30-minute virtual training session with one to two team member(s) who assisted with the programme. The team discusses the lessons taught throughout the week, safety considerations for independent and on-street riding related to child development, individualised feedback on the child’s performance, and tips for practising skills at home. The parent is also reminded to read through the parent tip sheets sent home each day of the programme.

Data collection

Surveys

At the initial in-person meeting, the parent and adolescent each complete a succinct baseline survey to collect demographics, bicycling experience and bicycle safety knowledge. Among adolescents, additional questions guided by the Health Belief Model (HBM)14 are included that relate to perceptions of severity, susceptibility, benefits and barriers; cues to action; and self-efficacy. The HBM is a widely used framework for health behaviour interventions, including traffic safety.14–17 Among parents, additional questions from the Early Adolescent Temperament Questionnaire—Revised (EATQ-R)18 related to inhibitory control, aggression and sensation seeking subscales are included as proxies for risk-taking and impulsivity tendencies of their child. The EATQ-R is commonly used and validated temperament measurement designed for children aged 9–15 years old.18

Naturalistic riding data

At the initial meeting, the team also installs the Pedal Portal 2 data collection system on the adolescent’s bicycle. Participants are asked to record all bike rides for 1 week using this system. The rider is instructed to stop and press the button during rides to mark points in the video if crashes, near crashes or other perceived threats occur. At the initial meeting, participants are given examples of crashes, near crashes and threatening situations (eg, obstruction in the road) to help support consistency across participant reports. Once the rider stops in a safe location, they will press the button and verbally describe the event that occurred into the system’s microphone. These data help the team quickly identify safety-relevant events and allow the adolescent to immediately describe the event in detail to reduce recall bias. An additional backup GPS module is installed on the bike to improve data quality and completeness.

Trip diaries

Participants are asked to maintain a simple written trip diary during each 1-week data collection period (before and after the intervention programme). The trip diary collects trip purpose, date, time, an overall stress rating of each ride and descriptions of any safety-relevant events.

Compensation

After 1 week, the team removes the Pedal Portal 2 System, collects the trip diary, discusses issues or notable events from the week and compensates the adolescent with a $25 gift card. This process is repeated with a shortened survey and 1 week of data collection after the intervention, with compensation of an additional $50 gift card. Data collection is completed between April and September, 2022 to 2024.

Data management

All identifiable participant information is stored separate of study data. A participant ID crosswalk document is saved in a password-protected project database.

Survey data

The team saves the completed baseline and follow-up survey responses to a password-protected Qualtrics database. Throughout data collection, a team member performs a weekly check of survey responses for completeness. On completion of the data collection season, all responses are downloaded, cleaned and coded for analysis.

Video, GPS data and trip diary data

Following each participant’s baseline or follow-up data collection week, team members transfer all files from the memory card into a password-protected database. Overall quality and completeness of these data are assessed to ensure the Pedal Portal 2 system is working properly before being deployed for further data collection.

Trip separation

Raw video files are compared with trip diary entries and videos corresponding to each trip are copied into separate folders. The team makes notes directly on the trip diary regarding data quality, diary entry accuracy and noteworthy observations. The hand-written diaries are scanned and saved, and the information about each trip is recorded in a password-protected database.

GPS repair

Due to environmental conditions (eg, heavy tree coverage) and/or participant behaviour (eg, riding before GPS signal lock), the Pedal Portal 2 system sometimes does not capture complete and/or accurate GPS information. These instances are identified during the trip separation process and recreated using an online temporospatial tool to trace the route in a series of segments containing start and end points associated with timestamps observed in the video. These segments can be exported in GPX file format and synced with the corresponding video when loaded into a GUI being built and customised for this project. To minimise the need for GPS repair, the Pedal Portal 2 system integrates a supplemental GPS device that uses Wi-Fi and cellular networks to obtain a GPS signal more quickly and consistently.

Safety Monitoring Committee

A Safety Monitoring Committee (SMC) was established to identify potential human subject protection issues. The SMC reviews any events that initiate a suspicious behaviour, crash report or violation of subject privacy. Reports are generated when a team member witnesses behaviour that is dangerous to the bicyclist or to the general public or is illegal, or crashes resulting in property damage, injuries or fatalities. The SMC non-investigator subject matter experts advise the team on an appropriate course of action, including discontinuation of the study or removal of a participant.

Planned data analyses

Study measures

Aim 1 measures

The primary outcome variables for Aim 1 include rates of safety behaviours and safety-relevant events and proportions of both positive and negative safety behaviours relative to the overall number of instances of a given event behaviour (table 1). These outcomes will be coded via review of video and GPS. Information on these behaviours and events will be augmented via contextual information about what happened, drawn from the written trip diaries.

To analyse Aim 1 measures, the team is developing a GUI that allows the data annotator/coder to simultaneously watch three data streams (front-facing video, rear/rider-facing video and GPS) and annotate events and observations through a series of menus (figure 3). The annotated events are associated with a timestamp and location that will be exported from the GUI as one trip-level spreadsheet.

Figure 3
Request permissionrequest permission icon
Figure 3

A screenshot of the graphical user interface (GUI), consisting of two video streams (front and rear-/rider-facing), a map window, and a series of dropdown menus and dialogue boxes.

Prior to annotation, we will establish a baseline inter-rater reliability between all annotators. Before annotation can begin, a kappa statistic of 0.85 or greater must be achieved between all possible pairs of annotators through a minimum of 100 independently annotated videos. To avoid information bias, annotators will be blinded to the participant study arm and data collection period (baseline vs follow-up) using randomly generated trip IDs and will not discuss their observations with one another.

Aim 2 measures

The primary outcome measures for Aim 2 assess difference in knowledge, perceptions (mean ratings of susceptibility of injury, severity of injury, benefits of safety behaviours and barriers to change), self-efficacy and cues to action (eg, whether one has a reminder system for proper hand signals) from baseline survey to follow-up survey.

Covariates

Potential confounders or effect modifiers to be examined for Aim 1 include rider, trip and built environment/neighbourhood characteristics, based on previous research. Potential covariates or effect modifiers to be examined for Aim 2 include rider characteristics.

Rider characteristics

  • Survey: Age, sex, bicycling experience, past bicycling instructions given by parents and risk-taking tendencies and impulsivity.

Trip characteristics

  • Video: Time of day, traffic volume (motor vehicles, pedestrians, bicycles), ride type (alone, with others—adult present, with others—no adult present), road condition (potholes, cracks, etc), parked cars and other road user hazards (trash bin, debris, etc).

  • GPS: Length of trip (time and distance), speed limit, roadway functional class (arterial, collector, local).

  • Trip diary: Trip purpose (commute, errand, recreation/social), as assessed from trip diaries.

Built environment/neighborhood characteristics (based on catchment area of each study site)

  • GPS: Population density, employment density, intersection density, median traffic volume, roadway functional class distribution, miles of bicycle facilities, street density (centerline miles per square mile of land area).

Analytical plan

Descriptive statistics will be examined for all variables at both baseline and follow-up (pre- and post-intervention). Analyses will focus on examining the differences between the study groups, which will be coded as a three-level categorical variable: (1) Bike Club, (2) Bike Club Plus and (3) Control. Rates will be calculated for each behaviour/event by determining numerators (number of each type of behaviour or event, miles ridden, minutes ridden; see table 1) and denominators (total miles or minutes). Mean differences in the rates and proportions of safety behaviours and safety-relevant events per trip will be estimated for pre- and post-intervention periods. Models will be adjusted for relevant dyad characteristics and other potential covariates (eg, age, bicycling experience).

Missing data

First, missing data patterns will be examined. If missing data appear to be isolated and unrelated to study outcomes or potential covariates, it will be assumed to be missing completely at random and complete-case analyses will be conducted. Otherwise, a multiple imputation method will be used to accommodate missing data.19

Aim 1 analytical plan

The primary outcomes for Aim 1 are the number of safety behaviours and safety-relevant events (see table 1) per 100 miles/min ridden and proportion of positive and negative behaviours relative to the overall number of instances of a given event behaviour. These behaviours and events will be identified as part of the video/GPS and trip diary data coding procedure. We will analyse and model each of these behavioural/event categories separately.

For rate-based behaviours and events listed in table 1, generalised linear mixed models (GLMMs) will be constructed with trip as the unit of analysis, response variables defined as counts of behaviours and events, and group as the main explanatory variable. Random participant-level effects will be incorporated to adjust for within-subject correlations arising from multiple trip outcomes recorded for the same adolescent. The models will be based on a Poisson or negative binomial distribution and a log link function, given the outcomes are count-based (behaviours and events). Trip length (log transformed) will be included as an offset variable, which results in the modelling of rates based on ratios of event counts to trip length. Each model will be fit with offsets based on time (minutes) as well as distance (miles), for comparison of risk by time exposure versus distance covered.

For proportion-based behaviours, the number of behaviours will be recorded and treated as binomial random variables. These outcomes will be constructed as proportions of the number of behaviours of a particular type relative to the overall number of identified instances evaluated per trip (eg, number of failures to scan for traffic/number of necessary scanning instances encountered). To compare the mean differences between the Bike Club intervention group and the Control group and between the Bike Club and Bike Club Plus intervention groups, GLMMs will be fit with study arm as the main explanatory variable. Random subject-level effects will be incorporated to adjust for within-subject correlations arising from multiple trip outcomes recorded for the same adolescent. The logit link function will be used and the log of the trip length (miles or minutes) will be included as a covariate.

For all models related to Aims 1 and 2, the impact of clustering at the site-level will be investigated. Site indicators will be included to account for potential site-to-site differences (eg, socioeconomic status (SES), race/ethnicity distributions). Inclusion of more granular indicators nested within sites to reflect potential differences in neighbourhood and environmental characteristics with relevance to bicycling trip generation and risk exposure (population density, employment density, intersection density, median traffic volume, roadway functional class distribution, miles of bicycle facilities and street density) will also be considered. Relevant comparisons will be conducted using contrasts.

Modelling diagnostics, including residual, will be used to assess adequacy of fit for all final models. Potential covariates will be considered and assessed for inclusion (see the Covariates section). Variable inclusion will be guided by the use of information criteria (eg, Akaike information criterion).

Aim 2 analytical plan

The primary outcome for this aim will be differences in knowledge test scores and differences in mean ratings of self-efficacy and perceptions between baseline and follow-up measures. We will construct traditional linear regression models, more specifically analysis of covariance models, to compare mean differences between the control and intervention groups. Normality will be assessed for the scores and if the distributions are highly skewed, generalised linear models (GLMs) will be employed based on the gamma distribution with a log link function. Another outcome for this aim will be cues to action, which will be constructed as the proportion of ‘yes’ responses relative to the overall number of cues to action. The GLM for this outcome will be based on a binomial distribution with logit link. To compare the mean differences between the Bike Club intervention group and Control group and between Bike Club and Bike Club Plus intervention groups, GLMs with group as the main explanatory variable will be fit.

Ethics and dissemination

This study has ethical approval from the Institutional Review Board (IRB) at the participating institution (IRB-01-202105148). No significant risks are associated with participation in the study. The normal risks associated with riding a bicycle exist. If a protocol change is needed, a modification must be approved by IRB before implementation.

Common dissemination strategies will be used to disseminate the study findings: (1) traditional academic outreach (eg, peer-reviewed publications, presentations); (2) media outreach (eg, radio, TV, social media) and related materials (eg, reports, infographics); (3) personal contacts (eg, professional networks); (4) key stakeholders and organisations (eg, local, state, and national safety advocates and educators); (5) distribution of actionable recommendations to appropriate institutions and organisations; (6) creation of actionable recommendations for additional research.

Intellectual property and data generated from this project will be administered in accordance with both University and NIH policies. Access to databases, protocols and associated software tools generated under the project will be available for educational, research and non-profit purposes.

Patient and public involvement

Patients and members of the public were not involved in the design of the study and reporting of the study. Members of our partner community organisations are involved in the delivery of the intervention. Results will be shared with participants and to stakeholders as outlined in the Ethics and dissemination section.

Article metrics
Altmetric data not available for this article.
Dimensionsopen-url