Objective: To determine the accuracy of police, emergency department, and ambulance records in describing motor vehicle crash (MVC) characteristics when compared with a crash investigation report (CIR).
Methods: This study was a retrospective record review. Sixty-three motor vehicle crash (MVC) patients transported to a university hospital emergency department via ambulance and also reported in a crash investigation record (CIR) during the period January 1993 to December 1995 comprised the study population. The crash characteristics analyzed were occupant position (OP), restraint use (RU), air bag deployment (AD), type of impact (TI), ejection (EJ), and external cause-of-injury code (EC). The accuracies of the police report (PR), the emergency department record (EDR), and the ambulance report (AR) for each patient were compared with the CIR by computing percent agreement, with 95% confidence intervals (95% CIs) for each variable and for each data source.
Results: Overall average agreement was 92.9% for PR, 89.7% for EDR, and 80.7% for AR. The overall average agreement for each variable was 98.9% for EJ, 92.1% for AD, 91.5% for OP, 90.5% for EC, 77.2% for RU, and 76.2% for TI. For all but one variable (RU), 95% CIs overlapped between data sources.
Conclusions: The accuracy of data sources used to determine crash characteristics varies. Using a CIR as the standard, the PR was the most accurate. Inaccuracies occurred most frequently for RU and TI. Researchers and clinicians need to be aware of these inaccuracies.