Detection of Geometric Risk Factors Affecting Head-On Collisions through Multiple Logistic Regression: Improving Two-Way Rural Road Design via 2+1 Road Adaptation.

Laura Cáceres,Miguel A Fernández,Alfonso Gordaliza,Aquilino Molinero

doi:10.3390/ijerph18126598

Laura Cáceres, Miguel A Fernández + Show 2 more

Open Access

https://doi.org/10.3390/ijerph18126598

Copy DOI

Abstract

This study aims to characterize locations on two-way rural roads where head-on crashes are more likely to occur, attending to geometric road design factors. For this purpose, a case-control study was carried out using multiple logistic regression models with variables related to road design parameters, considering several scenarios. The dataset corresponding to cases (places where crashes have occurred) was collected on Spanish “1+1” rural roads over a four-year period. The controls (places where no crashes have occurred in the period) where randomly selected through a specific ad hoc designed method. The obtained model identifies risk factors and allows the computation of the odds of a head-on collision on any specific road section: width of the pavement (when it exceeds 6 m), width of the lanes (for intermediate widths between 3.25 and 3.75 m) and tight curves (less than 250 m of radius) are identified as factors significantly increasing the odds of a crash, whereas a paved shoulder is a protective factor. The identified configurations on two-way rural roads may be susceptible to transformation into “2+1” roads to decrease the odds of a head-on crash, thus preventing possible serious injuries and enhancing transportation safety.

Highlights

In most countries around the world, the highest percentages of all road fatalities come from crashes on rural roads
“1+1” roads into the so-called “2+1” roads which are two-way rural roads, with one or two lanes in each direction in the same carriageway and where both ways are separated through the installation of a road restraint system or any device to avoid the presence of a vehicle in the opposite direction
In order to select a final model for each scenario, we considered two criteria: the proportion of correctly classified observations and the area under the ROC curves (AUC) [39] of each model compared using the DeLong

Summary

Introduction

In most countries around the world, the highest percentages of all road fatalities come from crashes on rural roads (for instance, in the European Union -EU-, fatalities on these roads represent 55% of the total number [1]) Among these interurban roads, the most frequent type is the two-way rural road with only one carriageway, in particular, the so-called “1+1” roads (two-way rural roads, with one lane for each direction in the same carriageway and where overtaking is possible). One possible solution to avoid head-on crashes and prevent the associated injuries is to transform these “1+1” roads into the so-called “2+1” roads which are two-way rural roads, with one or two lanes in each direction in the same carriageway and where both ways are separated through the installation of a road restraint system or any device to avoid the presence of a vehicle in the opposite direction (see Figure 1) This type of road is used as an intermediate solution between the common two-lane road and a dual carriageway road

Objectives

Methods

Results

Discussion

Conclusion