This study examined two 2002 New York State public databases: police crash reports and hospital discharge data. Probabilistic linkage applies Bayesian statistical methods to link databases; observed agreements and disagreements of common fields are used to determine the probability of a match. For this study, selected match fields were date and hour of crash, geographic location, gender, age, birthday, name, injury status, crash type, and occupant type.
We built a logistic regression model to impute missing values on safety belt use and adjust potentially misclassified values. The backseat occupant safety belt use was adjusted from the recorded 69% to 35% in New York State in 2002. Driver's alcohol involvement, male gender, and nighttime crashes were associated with non-use of safety belts by adult backseat occupants. Non-use of safety belts by backseat passengers was related to higher rates in hospitalization, traumatic brain injury, and vertebral cord injury.
LA - en SN - 0022-4375 UR - http://dx.doi.org/10.1016/j.jsr.2005.10.014 ID - ref1 ER -