EU Directive 2019/1936/EC on road infrastructure safety management introduced a network-wide road safety assessment approach (NWA) to improve the management of road safety across the member states. The methodology combines reactive (crash-based) and proactive (road design-based) analyses. However, its statistical foundation and national-scale application remain limited, particularly in handling statistical uncertainty, low crash frequencies, the integration of heterogeneous data sources and the inability to quantify differences in safety performance. To address these limitations, our study proposes three alternative approaches: the Rate Ratio test, the Bayesian Inference method and the Empirical Bayes method. These approaches were applied to the Czech national dataset of motorways and primary roads comprising 7826 road sections. The beneficial features of the proposed methods have been described. All three methods demonstrated improved performance when compared to the NWA approach. The Rate Ratio test provided formal hypothesis testing and reduced false positives. The Bayesian Inference provided a probabilistic framework, improved reliability for road sections with low crash counts and explicit estimation of the magnitude of safety performance differences. The Empirical Bayes method enabled the identification of the most influential design factors. In conclusion, the proposed methods provide statistically sound and robust alternatives to support road safety management at the national-network level.
Andrášik et al. (Mon,) studied this question.