Martinez 1996.

Methods	Non‐RCT
Participants	Participants were from 2 high schools (grades 10, 11 and 12) in the US, matched for socioeconomic factors but geographically separated, with enrolment of participants occurring in 4 sections of a physics class. Number of participants: 129 students in the intervention group and 74 students in the control group.
Interventions	Intervention: a 5‐component course consisting of audio‐visual aids, physical demonstration and a didactic lecture. A researcher delivered the course over 1 week, with each contact lasting 1 hour. The 5 components were basic energy lesson; safety features of vehicles including seat belts; occupant kinematics and forces/crash prevention, e.g. airbags; review of days 1 to 3 and a demonstration of a rollover, students then designed crash vehicles; the students tested their crash design. Control: usual physics lesson.
Outcomes	All outcomes measured by questionnaire 2 weeks, and 6 months after intervention. Self‐reported behaviour and practices (seat‐belt use, speeding and drink driving). Safety knowledge (physics of crashes, demographics of people involved in crashes and characteristics of automobiles).
Injury mechanisms	Pedestrian, cycle, motorcycle, vehicle (non‐specific).
Notes	No sample size calculation; non‐significant results may have occurred due to lack of power. There was a difference in the school grade of control and intervention groups at baseline.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation to intervention/control (selection bias) (for non‐RCT and CBA studies)	Unclear risk	Allocation was not described, except that they were 'chosen'.
Blinding of participants and personnel (performance bias) All outcomes	High risk	It was clear which group the participants were in as the intervention group received the lessons and the control group had lessons as normal ‐ blinding not possible.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Paper did not report who analysed the data.
Incomplete outcome data (attrition bias) All outcomes	High risk	Only results for students with scores for before instruction (time T1), and at 2 weeks (T2) and then T1 and T3 scores (6 months after instruction was completed) were included. No imputation for the missing data was carried out. There was also a large dropout rate in the control group at T3 (differential).
Selective reporting (reporting bias)	Unclear risk	Unclear how many children were in each group for the analyses and the authors did not mention removing outliers.
Other bias	Unclear risk	Methods of adjustment used in the regression modelling not described.
Risk of bias due to confounding (for non‐RCTs and CBA studies)	High risk	There were slightly more males in the intervention group, and a difference of 1 school grade between most of the intervention and control group. Schools were matched on socioeconomic status.