A car occupant might be killed if struck by another occupant who was catapulted forward, backward, or sideways in a crash. Cummings and Rivara estimated the association between the death of a car occupant and the restraint use of another occupant in the same car.

The authors matched “target” pairs in reports of cars that crashed and resulted in the death of one of the target occupants. The authors used a method designed to eliminate confounding related to vehicle speed, crash location, and other variables. By assessing information related to restraint use of other passengers, the study sought to determine the effect of restraint use, or lack of it, on the target occupant, whether that target occupant was restrained or not. Target pairs considered in the study were both in the front seat or both in the back seat, or one seated behind the other. Occupants with a front-back diagonal relationship were not considered.

The risk of death for a front target with a restrained occupant seated behind was nearly the same as that of a front target with no one seated behind. However, when the person behind the front target was unrestrained, the risk of death was greater than with no one seated behind and was greatest when the front target was restrained.

The risk of death for a rear target with a restrained occupant seated in front was similar to that of a rear target with no one seated in front. However, when the front occupant was unrestrained, the risk of death to the rear occupant was greater than if no one were seated in front.

The risk of death from an unrestrained occupant seated next to the target was greater than if there was no one seated beside the target. If the target was restrained, the risk of death was less if the adjacent person was restrained, compared with having no one seated next to the target.

An unrestrained target was less likely to die if the person adjacent also was unrestrained than if that occupant was restrained. This latter finding may have been caused by the likelihood that when two unrestrained persons are seated side-by-side, they are thrown in the same direction, rather than against each other. However, the authors speculate that the finding also may have been a confounding error.

The authors established associations between restraint use and nonuse by occupants seated in relation to each other. If the associations are causal, the authors estimate that use of restraints by rear occupants may prevent about one in six deaths of restrained front-seat targets. Similarly, use of restraints by front-seat occupants would reduce rear-seat occupant deaths. The authors conclude that all persons in the same car should wear restraints to reduce crash deaths, especially deaths that occur because of the impact of unrestrained passengers on restrained passengers.