Reporting road victims: Assessing and correcting data issues through distinct injury scales.

INTRODUCTION: The most common measurement for road accidents relies in police reports; however, there is a high portion of underreporting and misclassification, mainly concerning elderly causalities, urban accidents, slightly injured, users of two-wheeled vehicles, and car occupants.

METHODS: In order to assess these issues, road accidents occurring in the Porto Metropolitan Area, Portugal, covering a 6-year period (2006-2011) were analyzed based on police and hospital datasets. By linking hospital data with police data, it is possible to evaluate the misclassification of the victims' severity by the police regarding the maximum abbreviated injury scale (MAIS) classification. Additionally, considering that 29% of the victims recorded by hospitals were not reported by the police, which is in line with the reality of other EU countries, underreporting is further investigated. Thus, we used econometric and statistics tools to measure the correlation between different available data to identify possible causes of underreporting and misclassification. In this sense, factors contributing to the misclassification of casualties by the police are identified using a univariate analysis. On the basis of the linked police-hospital data, and considering those factors and the police classification, a probabilistic model was developed to estimate a MAIS-based classification for all individuals included in the police accident records.

RESULTS: The results of misclassification indicate a significant over-classification of severe injury by the police. Additionally, a systematic police underreporting phenomenon of around 30% was found. Conclusions and Practical Applications: Finally, comparing estimated results and actual data, we were able to produce non-fatality adjustment coefficients to estimate the total casualties taking into account the underreporting and misclassification phenomena and to compare them with the Portuguese and European realities.