Inclusion of adult vehicle occupants in matched-cohort studies of child restraint effectiveness

Background

Many researchers believe that only child occupants should be used in matched-cohort studies of child restraint effectiveness. The objective of this study was to determine if the inclusion of adults in these studies biases estimates of restraint effectiveness.

Methods

We conducted data simulations to represent 10,000 collision-involved vehicles carrying a mix of children and adults. We set the effects of age, adult seat belt use, child seat belt use, and child safety seat use to known values. We made age and crash severity modifiers of the adult and child seat belt risk ratios (RR) and of the safety seat RRs, and we made crash severity a modifier of the age RRs. We also made collision severity a confounder of the restraint RRs by creating an association between restraint use and collision severity. RRs were estimated using conditional Poisson regression and compared with the true values.

Results

RRs for death estimated from the simulated data were nearly identical to the true values. The average child safety seat RR was 0.42 (true value 0.40) and the average child seat belt RR was 0.54 (true value 0.53). Age effects were also correctly estimated; child age RR was 1.69 (true value 1.66) compared with adults.

Conclusions

These findings demonstrate that matched sets of adult and child vehicle occupants, together with conditional regression, can be used to estimate associations between restraint use and the risk of death. No bias resulted from the effect modification. Conditioning on the vehicles eliminated confounding by crash severity.