Slashing Two-wheeled Accidents by Leveraging Eyecare

The primary aim of STABLE is to determine whether the provision of spectacles for the correction of myopia can reduce the average number of crash-near-crash (CNC) events among young motorcycle drivers in Vietnam.

Road traffic crashes kill more people aged 5 to 29 years worldwide than any other cause and are predicted to become the 7th leading global causes of death for all ages by 2030. The Southeast Asia has among the world's highest burdens of traffic mortality, with 20.7 fatalities per 100,000 population. The United Nations reports that motorcycles account for 93.3% of all motorized vehicles on the road in Vietnam, the highest number as a proportion of all motorized vehicles of any Association of Southeast Asian Nations country.

The investigators recently published a meta-analysis in Lancet Global Health which revealed that drivers with poor vision in Low and Middle-Income Countries (LMICs) had a 46% greater risk of experiencing a road traffic crash than those with good vision Risk Ratio (RR) 1·46, 95% Confidence Interval 1·20 to 1·78, p<0·001, 13 studies). Nevertheless, the impact of poor vision on the safety of young road users in LMICs, where many drivers do not undergo vision testing, remains poorly understood. There is a particular dearth of information for adolescent drivers, who are at greatest risk for crash-related death and the lack of reliable data has proven a significant barrier to inference of cause and effect.

Trials with robust measurement of CNC events with appropriate equipment under conditions of naturalistic driving are needed, and these have become possible only recently, due in part to falling costs and aggressive miniaturisation of GPS and on-board video systems.

STABLE will be the first cluster randomised controlled trial to investigate whether there is a link between vision correction with glasses and road safety in a LMIC.

Rationale for Intervention: The intervention in STABLE is the provision of glasses for the correction of presenting distance visual acuity <= 6/12 in the better-seeing eye due to un- or under-corrected myopia following an eye examination. The intervention will be applied to male and female motorcycle users between ages 18 and 23 enrolled at universities in Ho Chi Minh City, Vietnam

Rationale for Comparator The comparator in STABLE is standard eyecare. For university students in Vietnam, this requires the individual to pay for eyecare services provided by trained ophthalmologists or optometrists and access to good-quality eyecare services amongst university students is low. This is illustrated by the high proportion either not wearing glasses, or having glasses with incorrect power in the unpublished study noted above.

STABLE is an investigator-masked mixed-methods naturalistic driving study using a stepped-wedge cluster randomised trial (SW-CRT) design. It uses a SW-CRT design because, in a conventional parallel group cluster randomised trial, half the clusters would be randomly assigned to the intervention condition and half to the control condition. This approach would necessitate identifying participants with un- or under-corrected myopia and withholding glasses from them if they were randomly assigned to the control group until after the completion of the trial. The STABLE investigators maintain that it would be unethical to withhold glasses from motorcycle drivers who have been identified as needing them through the trial's processes for this length of time. Therefore, working with local stakeholders in Vietnam they agreed on the use of the SW-CRT design, which balances the need to conduct a randomised evaluation of the intervention with the ethical demand to provide glasses to trial participants as soon as their need for vision correction is identified.

Primary objective: is to provide reliable estimates of the impact that vision correction with glasses has on the number of CNC events in a Low- and Middle-Income Country (LMIC) setting.

Secondary objectives: are to investigate the effects of study interventions on self-reported visual function; self-reported crashes for comparison with DAS recorded crashes, and the total delivery cost per crash event avoided.

Recruitment Methods: The research team will approach the Board of Executives of the five universities and explain the purpose of the study and invite them to join the study. If they agree to participate and depending on their assignment, trial partners Asia Injury Prevention Foundation will work with each university's Department of Student Affairs and/or the Student Unions to announce, advertise and follow up on trial recruitment activities. The trial will be advertised on posters, through email and through fan page of either the Department of Student Affairs or the Student Unions.

Informed consent: to participate in STABLE will be sought from 5,200 students in 5 faculties from each university (a total of 25 clusters), to perform a self-assessment of their vision using the WHO's free smartphone-based screening app, WHOeyes. Data Acquisition System(DAS) units will be fit to the motorcycles of 625 drivers (25 per cluster), with an estimated 250 (40% of those that fail the WHOeye self-assessment) judged to be eligible for vision correction and therefore assigned to the intervention condition over the course of the trial. A 30% dropout is anticipated, yielding complete datasets from 175 trial participants.Those who meet the inclusion criteria for DAS fitting and informed consent will undergo the informed consent procedure. 35 students will be recruited from the university to be involved in the pilot study, and 175 students (35 per cluster) will be recruited from each of the other five universities for the main trial.

Sequence generation: At each sequence (step), a centralised, password-protected web-based system will be used to randomly select five clusters from those that have not yet been assigned to STABLE's intervention group. The PI on site, or an individual with delegated authority, will access the system to generate the list of five randomly selected clusters.

Diversified Technical Systems (DTS) has developed the DAS for STABLE to support the research questions and objectives of the trial. The DAS will be connected to two colour, wide-angle view video cameras: forward-facing and driver-facing. Additional data streams include: 3-axis accelerometer, Geographic Positioning System (GPS), braking and ground speed.

DAS will be fitted on site to the motorcycles of eligible students who provide informed consent to participate in the trial. For the pilot study this will take place in the week prior to the start of the pilot study. For the main trial this will take place in the month preceding the start of the trial. It will take approximately 7 days to complete DAS fitting at each university, and approximately 35 days to fit the DAS to the motorcycles of all 625 participants.

Data collection and study Instruments: Trial data collection instruments will be used to capture the outcome measures. Data collection from DAS will begin after each DAS unit has been installed on the motorcycles of consenting participants. In the main trial, the first eye examination will take place 12 weeks after the last DAS unit is fitted. After the first eye examination DAS data will only be collected from those who:

• are in a cluster that has not yet been randomly selected for transition from the control period, or
• have undergone an eye examination and been diagnosed with uncorrected myopia or under-corrected myopia (and thereby entered the intervention group) The DAS systems will be removed from the motorcycles of those participants who have had an eye examination and do not meet the inclusion criteria for assignment to the intervention group. This process will be repeated at each sequence, until DAS data are being collected from only those who meet the inclusion criteria for assignment to the intervention group (myopes only). DAS data will be collected from these DAS units up until the time they are removed from the motorcycles. Data collection from all 35 clusters will cease after the final 5 clusters in the sequence have completed 12 weeks of follow up.

During the first three transitions, DAS units will be removed from students not entering intervention conditions, while their data are retained for secular trend modelling. In the final two transitions, DAS units will remain to continue collecting non-intervention driving data, strengthening secular trend assessment within the stepped-wedge design.

Prospective participants will be contacted by email and Short Messaging Service and asked to complete a screening questionnaire to know their age, ownership of bike.

Participants who meet the inclusion criteria for DAS fitting and informed consent will be contacted again and asked to provide the following information by questionnaire:

Data Collection

• KoboToolbox
• CRF 1 - Eligibility for Informed Consent and DAS Fitting
• CRF 2 - Baseline Assessment (VFQ 25, Dula Dangerous Driving Index)
• CRF 3 Part 1 - Before DAS Fitting
• CRF 3 Part 2 - After DAS Fitting
• CRF 5 - Vision Function Follow-up Assessment
• CRF 6 - USB Swap
• CRF 7 - Crash Report 19
• MACRO
• CRF 4 - Vision Examination
• Protocol Deviation Form
• Adverse Event Form
• Serious Adverse Event Form
• Off Study Form
• PI Sign Off Form Data will be collected using a mix of electronic and paper Case Report Forms (CRF).

Data capture by AIPF will be facilitated via KoboToolbox. Data capture by NICTU will be facilitated via MACRO.

Assessments in the event of a crash: Crash and Injury Self-Report Questionnaire (CI SRQ) will be collected from each participant involved in a crash event that results in hospitalisation. During the main trial, research assistants will maintain contact with participants by Short Messaging Service (SMS) on a twice monthly basis to retrieve DAS data and ensure that they are motivated to continue with the study.

Statistical Methods: Baseline demographic and clinical characteristics will be reported for all participants who are eligible for the intervention group for whom the data is available. Baseline characteristics will be summarised by means and standard deviations, medians and inter-quartile ranges, or numbers and percentages as appropriate.

Analysis of the primary outcome will use the mixed effects Poisson regression model. If the outcome includes a large number of zero values, a zero-inflated mixed effects Poisson model will be applied. In the presence of overdispersion, we will use a zero-inflated mixed effects negative binomial model. If any of these models fail to converge, a linear mixed effects model will be used as an alternative.

Null hypotheses and analyses for secondary outcomes take a similar form to that for the primary outcome. Mean differences for continuous outcomes and risk differences for binary outcomes will be reported (all adjusting for cluster, secular and time effects).

An agreement analysis will be done to compare VFQ-25 total scores at baseline and four weeks after receiving spectacles. Assuming that the response will be from the same participant and the time will be fixed, we will use the ICC and the concordance correlation coefficient (CCC) to estimate the agreement between the baseline and four weeks results.