Road safety data collection and storage are critical for a series of procedures to improve road safety. However, capturing accurate and complete road safety data is a difficult and complex issue. The main problems faced when recording road accidents are the unclear determination of road accident location, insufficient or incorrect recording of information and insufficient accident coverage.
Concerning road safety data availability, it is widely known that there is a serious lack of road safety data in African countries. Even when data are available, for example through international databases (WHO, IRF), little is known about data collection systems, data definitions, etc. Only few countries dispose suitable time series of road fatality data and especially for the latest available decade 2005-2014, only 21 African countries have available data for more than five years. The greatest lack in data concerns risk exposure and safety performance indicators.
Another key issue is the comparability of the data and the potential of using different databases in a complementary way. Concerning the fatality data, different definitions are used among African countries. WHO provides the primary data as received by the national sources, in the country profiles of its reports which adjust them to the 30-days definition and publish them. However, these data are not directly comparable because of differences in the quality of data collection processes among the countries. In order to take into account under-reporting issues and achieve comparability, WHO has developed statistical models to estimate the number of fatalities. While IRF uses the 30-days definition for the killed persons in road accidents, the data that publishes are those reported by the national sources, which use different definitions. Thus, the data from different countries are not comparable without being processed first and attention is needed when combining the two databases.
Concerning the data on exposure and road safety performance, the comparability of the countries with available data is not reliable, since the data refer to different years, with a difference of more than ten years in some cases (e.g. road network density). Moreover, there is not much information on the collection methods necessary for appropriate comparisons.
Within SaferAfrica, recommendations for a minimum set of harmonized data collection procedures and definitions that could be applied in the short- to medium term to improve African data collection systems (based on the WHO Safety Data Manual) were drafted. The recommendations for all types of data (accident, exposure and safety performance) consist of a minimum set of data elements and a common collection system. In addition, a two-step approach is proposed, including the improvement and harmonization of the existing data and methods the collection of new harmonised data.
As far as road accident data are concerned, the data collection form used by the police, is recommended to be revised frequently, include detailed information on the vehicles and road users involved in the accident, as well as adopt all existing standardized international definitions of variables and values. Concerning road fatalities, the international 30-days definition is recommended to be adopted by the African countries. On that purpose, the countries that are not currently utilizing such a definition should modify the data collection process and develop appropriate conversion factors. Underreporting is also an issue that should be tackled, so that the databases are further improved and comparability of the data among the countries is reached. The data are recommended to be adjusted by means of linking police data with hospital data.
When developing a common accident data system, the minimum data elements should be defined based on selection criteria, concerning the usefulness of the selected variables and values, the level of disaggregation and the difficulty of their collection. All variables and values should refer to casualty road accidents. Additionally, the accident data structure is suggested to comprise four categories of variables, which are related to crash, road, vehicle and road user characteristics.
Regarding the exposure and performance indicators, the respective variables and values are recommended to be defined in a way that they will be complementary and compatible to the accident data. The exposure measures concern two groups of data, the road traffic estimates and the road user at risk estimates. Collection processes concern travel surveys and traffic count systems, while national registers may also provide useful and commonly used exposure data, such as population, drivers’ population, vehicle fleet etc. Road safety performance indicators could be estimated either by using observational techniques or based on national statistics and data collected by national registers.