Abstract
Objective: The study aimed to (1) assess the magnitude of road traffic injuries in a country missing a formal linkage system of police with hospital data, (2) quantify the underreporting, and (3) produce a convenient algorithm exploring its constituent components.
Methods: Linkage of disaggregate (individual) data collected by the road traffic police (RTP) with those by the Emergency Department Injury Surveillance System (EDISS) on the Greek island of Corfu and coded with different classification systems was carried out. The applied four-step methodology, also comprising the calculation of underreporting coefficients of the variation by basic demographic variables, mode of transport, and injury outcome, led to the identification of the overall underreporting from either registry.
Results: RTP data captured 96.6% (coefficient: 1.035), whereas EDISS captured only 54.4% of total fatalities (overall concordance: 51.1%). On the contrary, EDISS captured 94.6% of nonfatal injuries, whereas RTP only captured 16% (coefficient: 6.238), resulting in a low overall concordance (10.6%). Considering severity of injury assessed by EDISS, by using the ISS as the gold standard, RTP data misclassified 20.3% of severe injuries as less severe, and a statistically significant difference in the underreporting by gender was also noted.
Conclusion: Relatively simple methodologies can provide essential coefficients to assess the actual numbers, severity, and components of road casualties by complementing routinely collected RTP with sentinel emergency department reporting systems.
ACKNOWLEDGEMENTS
This study has been supported in part by the European Home and Leisure Injury Surveillance project co-funded by the European Commission and by the Greek Ministry of Health and Welfare. Part of the work carried out by the NTUA team (George Yannis, Petros Evgenikos, Antonis Chaziris and Natalia Tselenti) was partly financed by the EC, within the co-funded “SafetyNet” Integrated Project.
Notes
*p Value derived from z-test comparing the underreporting coefficients of the two genders in each age group.