Traffic safety meta-analysis of reversible lanes

Abstract

A meta-analysis was performed to review the potential effects of reversible lanes on traffic safety. A systematic review resulted in ten U.S studies, with 155 estimates of safety effects, that were selected for inclusion in the analysis. These studies employed either a simple before-after comparison or a cross-sectional comparison methodology. Study results were extracted by transforming the collision rates and frequencies of these studies into a log-odds ratio. The results of the meta-analysis suggest that the odds of a collision occurring are 30.9% higher for roads with reversible lanes when compared to roads without the treatment. The odds of a property-damage-only collision are predicted to be 16.6% higher, and injury collisions are predicted to be 34.9% higher with reversible lanes compared to no-treatment. The odds of a collision occurring during the peak period are expected to be 46.2% higher, while the odds of an off-peak period collision occurring are expected to be 12.8% higher. These results were statistically significant, with the exception of the off-peak period collision result. A meta-regression was performed, which regresses variables related to study type, collision type and operational characteristics of the study locations on the extracted log odds ratio. Peak period operations were positively and significantly correlated with an increase in crashes. Presence of left-turn restrictions and/or delineator/barrier were negatively correlated with crashes, whereas the presence of dynamic traffic control and static traffic control were positively correlated with crashes. Finally, cross sectional studies tend to find greater effects than before-after studies.

Keywords: Lane reversal; Meta-analysis; Meta-regression; Reversible lane.