Abstract
Pedestrians account for 26% of all traffic fatalities worldwide. According to in-depth collision databases, around 3500 temporal variables can affect the outcome of a collision, making it crucial to establish the relationship between each variable and the outcome. To-date, there is no method defined to assess these temporal variables' relevance other than a statistical correlation, which can sometimes lead to reasonable conclusions, but only under specific circumstances. This article addresses this issue by first conducting a literature review to determine all relevant variables, followed by developing a variable selection criterion to select crucial variables, and then conducting a meta-analysis to combine statistical results. Epidemiological studies published between 1990 and 2022 were examined, including 93 papers from 19 different nations, considering 904,655 pedestrian collisions. Of the 204 variables that were extracted from these studies, 152 were examined using the variable selection criterion, and 68 were found to be significant. Of these, 31 were included in the meta-analysis, which combined odds ratio to aggregate the effect of a variable across various studies, thus removing study-specific conclusions. This study is innovative as it proves that statistical correlation alone is insufficient to determine the importance of a variable. The proposed method is an objective way to distinguish the variables for stakeholders and identify the relevant variables. This study provides for the first time the definitive list of the 68 variables that must be included in any pedestrian-to-vehicle accident databases, allowing appropriate actions for safer roads.
Original language | English |
---|---|
Article number | 100158 |
Number of pages | 27 |
Journal | Transportation Engineering |
Volume | 11 |
Early online date | 30 Dec 2022 |
DOIs | |
Publication status | Published - Mar 2023 |
Bibliographical note
This is an open access article under the CC BY-NC-ND licenseKeywords
- Pedestrian
- Injury severity
- Accident
- Meta-analysis
- Systematic review
- Influential parameters