Objective To determine whether either the inclusion of adults in matched cohort studies of passenger vehicle occupants or modification of age effects by collision severity biases child restraint risk ratios biases estimate of child restraint effectiveness.

Methods Monte Carlo data simulations were conducted to represent 10 000 collision-involved vehicles carrying a mix of children and adults. The effects of age category, adult seat belt use, child seat belt use, and child safety seat use were set to known values. Age was a modifier of the adult and child seat belt risk ratios and of the safety seat RR, and crash severity was a modifier of the age RRs. We also created an association between restraint use and collision severity to confound the restraint-death risk associations. RRs were estimated using conditional Poisson regression and compared with the true values.

Results Estimated RRs for death were identical to the simulated values. The average child safety seat RR was 0.42 (true value 0.42), and the average seat belt RRs for children were 0.54 (true value 0.54) for those aged 0–3 years and 0.61 (true value 0.61) for those aged 4–17. Age effects were also correctly estimated, 1.69 (true value 1.69) for those aged 0–3 and 1.35 (true value 1.35) for those aged 4–17.

Conclusions Neither the use of matched sets of adults and children nor the modification of age effects by collision severity biases estimates of child restraint effectiveness in matched cohort studies.