Road Crash Casualties: Characteristics of Police Injury Severity Misclassification

Epidemiologic road crash injury data is based on police data in most countries. While validity is of some concern, it may be the only such data available at a national level. The authors discuss French police severity classifications of both "slight" and "serious" casualties. Using a five year study of Rhone County, police data and a road trauma registry were linked so that classification of 14,342 casualties were available using both New Injury Severity Score (NISS) and police data. Comparison of "slight" and "serious" casualties is made between both NISS 1-15 and 16-75 data and police data. Multivariate analyses of police severity misclassification and over and under classification probability are conducted as a crash and casualty characteristic function. Positive predictive value is 35% and police classification sensitivity is 72% with a Kappa estimation of 0.41. Overclassification is most likely in pedestrian and motorcyclist casualties, with relative risk at 1.4 and 1.2 respectively, as well as by police in rural, rather than urban, areas, with rural relative risk at 3.1. During the study period, overclassification decreased and underclassification increased. When interpreting police data severity results, misclassification characteristics must be taken into account. Unbiased nationwide severity figure estimates are being developed.

  • Availability:
  • Authors:
    • Amoros, Emmanuelle
    • Martin, Jean-Louis
    • Chiron, Mireille
    • Laumon, Bernard
  • Publication Date: 2007-2


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01045357
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Mar 27 2007 1:20PM