Variation in traffic injury settings—same implication of hospital and police-based traffic injury data?

Introduction: Injury and death from traffic incidents are still a major global concern and are associated with significant costs. Traffic incidents are preventable, but this requires planning based on accurate and complete injury data, otherwise efforts bypass targets. Objectives: The objective was to compare the agreement between hospital and police records for parameters independently recorded by both datasets and to report the most frequent traffic incident setting reported by either dataset. Methods: We analyzed routinely collected hospital records from Odense University Hospital, Denmark, for traffic injuries from 2015 to 2021. These include patient demographics and incident-related data (time and place of incident, road users involved, and collision mode) for all patients involved in traffic incidents. These injuries were then linked with police records from Funen Police District, which covers the same geographical area of the island Funen, Denmark. Results: In total, 21,562 road users were injured and subsequently treated at the hospital, and 5176 (24.0%) of these suffered from severe injury. The agreement between hospital and police records was for means of transport (passenger car/bicycle/moped/pedestrian/motorcycle) 93.9% 95% confidence interval: 93.0%, 94.9%, role (driver/passenger/pedestrian) 97.7% 97.1%, 98.3%, and use of seat belt/helmet 76.9% 75.2%, 78.6%. The accuracy of police severity classification was 50.5% 47.0%, 54.0%, and the completeness of police-reported severe incidents was just 15.1% 14.1%, 16.0% compared with hospital records. The top-5 hospital-reported severe traffic incident setting was an injured cyclist with no counterpart on a city road (n = 1130), and 4 of 5 in this top-5 involved an injured cyclist. The top-5 police-reported severe traffic incident setting was an injured cyclist with a passenger car counterpart on a city road (n = 116), and the remaining 4 in this top-5 involved an injured passenger car driver. Only one setting overlapped between hospital- and police-reported top-5 severe injury scenarios. Conclusion: This study showed an extraordinarily high agreement between parameters independently recorded by hospital and police records, but police records misclassify severity and is not a reliable source for severe injuries. Thus, road safety policies may not have the intended effect if preventive efforts are based solely on police-reported severe injuries. With the documented agreement on means of transport, role, and use of safety equipment we suggest that planning of preventive efforts is based on hospital records in areas where registration practices are sufficiently thorough supplemented by police data.