Injuries sustained in road crashes and their consequences are still a major public health issue. In most countries, epidemiologic knowledge of road crash casualties is mainly based on data collected by the police. Counts of road crash casualties are usually distributed according to the police definition of injury severity into fatal casualties, non-fatal serious and slight casualties. Fatal casualties are rather clearly defined and well reported, but this is still not the case for non-fatal serious and slight casualties. Their under-reporting is well established; the degree of under-reporting can be quite large, and furthermore it varies with characteristics such as injury severity and road user type. The police classification of injury severity into slight and serious casualties is also questionable. The validity of the data serving as a base for epidemiologic knowledge of road casualties should be assessed, and this should be done in each country as police definition of road casualties and of their injury severity differ across countries. We focus on the validity of police data of non-fatal casualties in France. The police definition of injury severity is based on length of hospital stay, and we first analyse how the police severity classification into ‘slight’ and ‘serious’ casualties compares with a widely used trauma scale, based on injury assessment, the Abbreviated Injury Scale (AIS). We further explore police severity classification biases: for instance, which road user types are more likely to be classified as ‘seriously injured’ at a given injury severity? and how often? In the second part, we analyse under-reporting of non-fatal casualties in the police files: the degree of under-reporting and how it varies according to some crash or casualty characteristics; in epidemiologic wording, it consists of identifying and quantifying selection bias factors. (Author/publisher)
Abstract