Automotive Sampling System database (NASS) from 1995-2005. Using this cohort, we tested for an interaction (effect modification) between outcome was "serious" injury (Abbreviated Injury Scale [AIS] R 3) in any body region. Multivariable logistic regression models, stratified by drivers versus passengers, included: airbag, height, weight, plus 10 important crash confounders. Interaction terms with p < 0.10 were considered significant. For each occupant, we calculated the probability (risk) of injury both with and without an air bag. All analyses included clusters, strata, and weights. 52,552 drivers and 14,732 passengers were involved in MVCs over the 11-year period, of which 2.5% drivers and 2.6% passengers were seriously injured. Height modified the effect of an airbag for passengers (p = 0.08), though this was less prominent for drivers (p = 0.15). Weight did not modify airbag effect for passengers (p = 0.81) or drivers (p = 0.54). Passengers with height 50300-501100 had a small protective airbag effect (risk of injury –2%, 95% CI = –1% to -3%), yet airbags appeared harmful for passengers R60300 (risk +5%, 95% CI +2% to +9%) and %401100 (risk + 4%, 95% CI = –2% to +10%). A similar, though less pronounced, relationship was found among drivers. It is concluded that the effectiveness of airbags on serious injury depends on an occupant’s height, particularly among front-seat passengers. While airbags are modestly protective for mid-stature occupants, they appear to increase the risk of injury for large and small stature adults. Occupant body weight did not modify the effect of airbags. (Author/publisher)
Abstract