Background

Pedestrians are among the most vulnerable road users because of their limited resistance to biomechanical forces and lack of mass, speed, and protection [1]. Pedestrians face relatively high risks of injury and death in traffic crashes, and such crashes have resulted in hundreds of thousands of pedestrian deaths annually worldwide [2]. From 2009 to 2019, pedestrian fatalities rose by 51% in the world. Moreover, a total of 1,300,000 pedestrians are killed in motor vehicle crashes annually [3]. Statistics revealed that approximately 75,000 pedestrians were injured in motor vehicle crashes in 2019 [4], accounting for 17% of all crash fatalities in that year. According to the US National Highway Traffic Safety Administration, most pedestrian crashes occur at intersections [5, 6]. Although the number of crashes occurring at non-intersection locations such as road segments is not as high as that of crashes occurring at road intersections, injuries sustained by pedestrians involved in road-segment crashes tend to be more severe than those sustained by pedestrians involved in intersection crashes [7].

Contraflow cycling schemes have been applied to improve cyclist safety and reduce crash risks [8]. The primary benefit of contraflow cycling schemes is that motor vehicles and cyclists can maintain continuous eye contact and perform evasive maneuvers to avoid crashes [9]. Such schemes have been adopted for pedestrians who must walk along road segments. The Finnish Road Traffic Act [10, 11] stipulates that pedestrians walk against traffic, and this mandate has been in effect for decades. Additionally, authorities in other countries encourage pedestrians to walk facing oncoming traffic in the absence of a sidewalk, pedestrian lane, or path [12].

Evidence from research conducted in Finland demonstrates that walking against traffic was beneficial in reducing pedestrian crash risks [13]. Similarly, a study conducted in Taiwan reported that pedestrians who walked with traffic appeared to have more severe injuries than did those who walked against traffic [14]. A possible explanation for the benefits of walking against traffic is that pedestrians walking against traffic are more visible than those walking with traffic, particularly at night [15].

Purpose

The aforementioned studies clearly demonstrated that walking against traffic is advantageous for reducing both crash risks and severity. According to our knowledge, relatively few studies have examined whether the beneficial effect of walking against traffic on injury severity also applies to a set of conditions such as lighting conditions, crash time, roadway characteristics (rural/urban areas), ages, and vehicle types. Accordingly, the primary objective of the current study was to fill this research gap by investigating whether the benefits of walking against traffic are consistent across various circumstances involving such factors.

Methods

Data source

This study collected data on vehicle crashes for the period between 1991 and 2020 from the STATS19 crash database, a national traffic database of the United Kingdom. The STATS19 database contains data on every accident that involves at least one vehicle, results in personal injuries, and is reported to the UK police within a 30-day timeframe after the incident [16]. The database was created in 1949 and is periodically updated. Its datasets include an accident report, a vehicle file, and a casualty file. The accident report documents the time and date of a crash and the corresponding weather, road, and lighting conditions; the vehicle file contains information regarding the vehicle and its driver; and the casualty file contains information about each injury. This database is publicly accessible on the Department of Transportation's website [16].

Figure 1 illustrates a flowchart of the data selection process in this study. We excluded casualties with missing data on sex, age, and speed limit. We used a complete case analysis approach, as proposed by Kang [17]. Our collected contained 44,488 pedestrian crashes, of which 16,889 (38.0%) involved pedestrians walking against traffic and 27,559 (62.0%) involved pedestrians walking with traffic.

Fig. 1
figure 1

Sample selection flowchart. aExclusion criteria listed below are not exhaustive, total cases may exceed 1,013

Full size image

Definitions of variables

We acquired the following demographic information on pedestrian casualties: sex, age, street lighting condition, crash time, speed limit, crash day and month, weather conditions, road surface conditions, crash partner (vehicle, sex, and age), and pedestrian walking direction (against traffic; with traffic). Age was categorized as follows: < 18, 18–40, 41–64, and ≥ 65 years. Moreover, the following variables were considered in the study: crash location (rural: roadways with speed limits of ≥ 51 km/h; urban: roadways with speed limits of ≤ 50 km/h), weather conditions (fine weather; adverse weather; or unknown), street lighting condition (daylight, lit streets in darkness, unlit streets in darkness, or unknown), road surface condition (dry; slippery), crash season (spring/summer: March–August; autumn/winter: September–February), crash day (weekday: Monday–Friday; weekend: Saturday–Sunday), and crash time (rush hour: 07:00–08:59 and 17:00–18:59.; non–rush hour: 09:00–16:59; evening: 19:00–23:59; and midnight/early morning: 00:00–06:59).

Statistical analysis

We determined the distribution of pedestrian injury severity according to a set of independent variables (Table 1). Chi-square tests were then employed to examine the relationships between the independent variables and pedestrian injury severity and to identify variables significantly associated with the outcome variables (p < 0.05). These variables were then incorporated into logistic regression models with multiple variables. Subsequently, the odds of fatal and nonfatal injuries were estimated using stepwise logistic regression models after for specific variables were controlled for. In addition, joint effect analysis was used to examine the beneficial effects of walking against traffic on injury severity to different situations.

Table 1 Distribution of pedestrian injury severity according to independent variables
Full size table

Results

As presented in Table 1, of the 44,488 pedestrian casualties, 1,409 were fatal (3.2%) and 43,079 (96.8%) were not. As many as 17,854 (40.1%) casualties were aged 19–40 years; although only 10.9% of the casualties were elderly pedestrians, their fatality rate was the highest among the age groups (4.5%). Regarding pedestrians’ movements, the number of fatal injuries was higher among pedestrians walking with traffic (27,559; 62.0%) than it was among those walking against traffic (16,889; 38.0%). Most of the pedestrian crashes occurred in daylight (28,521; 64.1%), during non–rush hour periods (09:00–16:59; 20,933; 47.1%), in urban areas (32,664; 73.4%), on weekdays (31,878; 71.7%), in autumn/winter (23,409; 52.6%), in fine weather (36,641; 82.4%), and on dry road surfaces (31,944; 71.8%); most of the pedestrian crashes also involved male pedestrians (27,670; 62.2%), cars as crash partners (32,361; 72.7%), crash partner drivers aged 19–40 years (16,603; 37.3%), and male crash partner drivers (26,781; 60.2%).

Table 2 presents the estimation results obtained from logistic regression models. The estimated parameter for walking with traffic was significant, suggesting that pedestrians in with-traffic crashes were 1.542 times more likely (adjusted odds ratio [AOR] = 1.542; confidence interval [CI] = 1.139–1.927) to sustain fatal injuries compared with those in against-traffic crashes. Other risk factors for fatal injuries include darkness-unlit condition (AOR = 2.349; CI = 1.949–2.939), midnight (0–6) (AOR = 3.023; CI = 2.369–3.859), rural roadways (AOR = 4.808; CI = 4.126–5.604), weekend (AOR = 1.202; CI = 1.062–1.360), elderly (AOR = 5.220, CI = 4.143–6.577), slippery road surfaces (AOR = 1.17; CI = 1.02–1.35), heavy goods vehicles as crash partners (AOR = 4.385, CI = 2.458–7.823) and male crash partner (AOR = 1.464; CI = 1.239–1.731).

Table 2 Multivariate logistic regression analysis results
Full size table

Figure 2 displays a forest plot demonstrating the interaction effects of walking with traffic and other variables on fatalities. Walking with traffic may interact with other variables synergistically to intensify pedestrian injury severity. For example, pedestrian injury severity may be intensified through the synergistic interaction of walking with traffic with elderly pedestrians (AOR = 2.65, CI = 2.16–3.26), rural roadways (AOR = 2.20, CI = 1.92–2.51), heavy goods vehicles as crash partners (AOR = 1.51, CI = 1.28–1.78), unlit streets in darkness (AOR = 1.48, CI = 1.29–1.70), and midnight crashes (AOR = 1.60, CI = 1.37–1.87).

Fig. 2
figure 2

Joint effects of walking with traffic and other variables on pedestrian fatalities

Full size image

Discussion

We analyzed accident data collected from the STATS19 database and determined that walking against traffic was more beneficial in reducing pedestrian fatalities during crashes than did walking with traffic. Our findings are in line with those of studies conducted in Finland and Taiwan [13, 14]. This study could be applied to these countries such as Japan and Spain where traffic compositions and urban design layout are similar to those in United Kingdom. These findings imply that mandating that pedestrians walk against traffic may not be practical in the United Kingdom; nevertheless, pedestrians should be encouraged to walk against traffic, especially on road sections without sidewalks, as suggested by Pai [14].

Our study contributes to research on pedestrian safety by demonstrating that the detrimental effect of walking with traffic on pedestrian fatalities was more pronounced under the following circumstances: in darkness-unlit condition, during midnight (00:00–06:59 am) hours, in rural areas, when the pedestrians were elderly (≥ 65 years old), when heavy goods vehicles were crash partners. Our findings support the notion that the benefits of walking with traffic are consistent across various circumstances, particularly in crashes involving elderly pedestrians.

Our study made several findings regarding the interaction effects of walking with traffic and other variables on fatalities. First, pedestrian crashes and injury severity could be increased by the interaction of walking with traffic with unlit streets in darkness, rural roadways, and midnight crashes [18,19,20]. This finding indicates that walking against traffic in these circumstances could help reduce pedestrian injury severity, and this suggestion can contribute to pedestrian safety. Although a previous study [19] recommended that pedestrians wear fluorescent or reflective garments to enhance their conspicuity, walking against traffic is also an effective countermeasure in rural areas, during midnight hours, or in unlit darkness.

Second, walking with traffic could interact with age to increase pedestrian fatalities, particularly among elderly pedestrians. Research [21, 22] has pointed out that clothing with retroreflective materials can make pedestrians, particularly elderly pedestrians, more conspicuous. Additionally, we recommend that pedestrians, especially older pedestrians, should walk against traffic, which may enhance the effectiveness of their reflective clothing and increases their conspicuity, consequently reducing their fatality risk.

Third, the detrimental effect of walking with traffic on fatalities could be accentuated by heavy goods vehicles as crash partners. This is because blind spots close to heavy goods vehicles may be exacerbated when pedestrians walk with traffic, allowing fewer opportunities for evasive action [23, 24]. The combination of walking against traffic and improved pedestrian conspicuity through pedestrian detection systems could be an intervention focus to reduce both crashes and injury severity.

Our study inevitably has several limitations. First, geometric characteristics, such as the presence of sidewalks and vertical/horizontal road curvatures, are not readily available in the STATS19 database; these factors affect pedestrian conspicuity, crash risks, and fatalities. Second, other factors such as alcohol use and fatigue, which may be more prevalent during midnight hours, were not considered in this study; they should thus be investigated further in future studies. Finally, although police-reported crash data generally contain no information on mobile phone use either by vehicle drivers or pedestrians, the impact of this factor on pedestrian crashes and fatalities should be explored further.

Conclusions

We determined that walking against traffic was beneficial in reducing pedestrian fatalities compared with walking with traffic. These benefits were particularly pronounced under circumstances involving unlit streets in darkness, midnight driving (00:00–06:59), rural roadways, elderly pedestrians, and heavy goods vehicles as crash partners.