How to reduce pedestrian fatalities: a case-by-case study to evaluate the potential of vehicle and road infrastructure interventions

Descriptive analysis of pedestrian accidents

A descriptive statistical analysis including frequencies and proportions was conducted on all available variables to identify characteristics of the fatal accidents. The accident location was divided into rural and urban according to the Swedish definition. An urban area is defined as an area with houses with at least 200 inhabitants unless the distance between houses is more than 200 m. Rural area denotes a geographic area not defined as urban with low population density located outside towns and cities. The collisions were also classified and divided according to the direction of the pedestrian and the vehicle prior to the crash (Figure 1).

Projected potential of countermeasures

A case-by-case analysis was performed to retrospectively determine the potentials of vehicle and infrastructure safety countermeasures. The analysis considered the entire chain of events leading to the fatal crash (Figure 2). This method has previously successfully been used on different data sets (Kullgren et al., 2019; Strandroth et al., 2016; Strandroth, 2015). The method is also used by the Swedish Transport Administration to manage the Swedish national road traffic safety work and to prioritize among future interventions aimed to achieve Vision Zero.

For each crash, estimations of the potential effects of interventions were made by studying crashes from a baseline (normal) and apply assumptions regarding the effectiveness of a specific intervention. The interventions included in the analysis are listed in Appendix A. The included interventions could be considered as independent but the combined potential of the two areas was assessed without double counting (i.e. a fatality cannot be prevented twice with different interventions).

The assumptions of the potential of the infrastructure to avoid each fatality were based on previous evaluations (Elvik & Vaa, 2004) and the design principles in Vision Zero (Johansson, 2008). The road infrastructure interventions (e.g. speed management (roundabouts, speed cameras, and speed bumps), and physical separation between vehicles and unprotected road users) were analysed and assessed depending on the road width, traffic flow and other road characteristics for each of the cases. In the analysis of future potential of road infrastructure interventions, they were assumed to be implemented immediately and thereby impacting the outcome.

The future potentials of the vehicle safety technology interventions (e.g. ESC, AEB, LDW and LKA) were based on previous evaluations (Cicchino, 2016; Cicchino, 2018; Cicchino, 2022; Kullgren et al., 2022; Lie et al., 2006; Rizzi et al., 2014; Sternlund, 2017) and analysed based on the estimated prognoses on their implementation rates in the Swedish vehicle fleet (Appendix B). The model year when a close to 100% fitment rate of each safety technology in new sold vehicles was predicted by a Swedish group of experts lead by the Swedish Transport Administration (STA) including car manufacturers and senior researchers. The implementation rate of each safety technology in new sold vehicles was based on current technology development and previous implementation rates of various safety technologies in Sweden. See for example implementations rates of AEB in (Ydenius & Kullgren, 2019). For each accident, based on the model year of the vehicles involved in the fatal crashes, it was assessed whether the vehicle would be fitted with a certain safety technology in 2030 or 2050 (Appendix B). The effects of two different implementation rates were analysed, one with an expected rate (denoted normal) and the other with a five-year faster implementation rate (denoted fast).

Characteristics of the fatalities

Out of the 226 fatally injured pedestrians (95 women and 131 men) 86 occurred on rural roads and 140 on urban roads (Appendix C). On urban roads an equal distribution between men and women was found, while on rural roads 70% of the fatalities were men. Half of the pedestrians (51%) were 65 years or older. Fifteen were under the age of 15, where of 7 under the age of 10. In 68% the pedestrian was hit by a passenger car, in 10% by a Heavy Goods Vehicle (HGV) and in 8% by a Light Goods Vehicle (LGV). The distribution was similar in both rural and urban areas apart from a higher proportion of crashes involving buses in urban areas (11%) compared to rural areas (3%). In urban areas the most common accident scenario was that the pedestrian was hit while crossing the road (71%), in daylight (53%), and with a posted speed limit of 50 km/h at the location of the accident. Only 10% occurred on roads with a speed limited of 30 km/h or below. In rural areas it was more likely that the pedestrian was hit while walking along the road (53%), with the majority (63%) occurring in darkness (63%), and most often with a speed limit of 70 to 90 km/h. Most of the accidents on rural roads (71%) occurred under conditions in which a combination of light, weather, and/or sight was assessed to complicate pedestrian detection by the vehicle safety sensors. On urban roads only 47% of the accidents occurred under such conditions. In 19% of the accidents the pedestrian was partly obscured by other vehicles or objects. Regarding intoxicants, 15% of the pedestrians on urban roads were under the influence of alcohol, while the corresponding figure for rural roads was 37%. Six per cent of the drivers of the hitting cars were under the influence of alcohol.

The potential of countermeasures

It was assessed that 93% of the pedestrian fatalities could be addressed and thereby have been avoided with relevant and known interventions regarding road infrastructure, vehicle safety systems, or a combination of the two (Table 1). It was assessed that road infrastructure interventions could have prevented 67% of the pedestrian fatalities and vehicle-related intervention could have prevented 80%. Seventy percent of the pedestrian fatalities on rural roads and 66% on urban roads would potentially have been prevented if relevant vehicle safety systems had been implemented in the whole vehicle fleet. For 11 (5%) of the fatalities, no known intervention could address the fatal accident (3 on rural and 8 on urban roads). Of these 11, two were hit by a motorcycle, one by a bicycle, one by a large truck, one by a bus, one by a construction vehicle, one by an unknown vehicle and four were hit by a passenger car. Among these four, one driver was under the influence of drugs and without a driving license, one was sitting on the hood of a car, one was run over by an MPV, and one was hit on a 2+1 road with the speed limit of 100 km/h.

Interventions related to the road infrastructure

In general, infrastructural interventions related to speed were found to address a larger proportion of fatal accidents on urban roads compared to rural roads, while separated pedestrian paths outside the carriageway were found to address a larger proportion on rural roads compared to urban roads (see Table 2). The intervention with the largest total potential was pedestrian crossings with speed calming measures for the motor vehicles to ensure a vehicle speed below 30 km/h, which had the potential to address 49% the fatalities on urban roads and 15% on rural roads (in total 36% of the fatalities). A reduced speed limit to 30 km/h in combination with speed calming interventions had the potential to prevent 44% of the fatalities on urban road, while only 5% on rural roads (in total 29% of the fatalities). Separate pedestrian paths outside the carriageway had the potential to prevent approximately 35% of fatalities on rural roads and 17% on urban roads (in total 15% of the fatalities), see Table 2.

Interventions related to the vehicles

It was estimated that the vehicle safety technology with the highest potential was AEB with pedestrian detection for passenger cars (Table 3). With this system available, 63% of the fatalities on urban roads and 50% of those on rural roads could potentially be prevented. Furthermore, autonomous emergency steering system (AES) with detection of pedestrians was found to have potential to prevent 21% of the fatalities on urban roads and 57% on rural roads. Furthermore, it was assessed that especially on urban roads AEB reverse, back cameras/sensors, and side radars for heavy vehicles (trucks and buses) have the potential to prevent 11% and 9% of the fatalities on urban and rural roads respectively. However, it should be noted that no projections related to the implementation rate of these systems were made in this study.

Future potential of interventions

The full potential of vehicle interventions was estimated to occur far in the future (in 2050), assuming expected implementation rates (Figure 2). In 2030, between 90% and 95% of the accident population would not be addressed by the estimated improvements of the vehicle fleet according to the estimates made. An increased implementation rate (5 years faster) would mean an additional 79 saved lives on urban roads and 44 on rural roads during a 30-year period compared to the expected implementation rate (Figure 2). Expected and increased implementation rates for the various vehicle safety systems are presented in Appendix A.

The impact of road infrastructure countermeasure

Regarding road infrastructure, two main approaches for controlling crash severity were identified; either to keep the relative velocity between road users within acceptable levels or to separate road user groups from each other. This study shows that infrastructural interventions related to the speed were found to address a large proportion of fatal accidents on urban roads, while separated pedestrian paths outside the carriageway were found to address a large proportion on rural roads. Only 10% of the fatal accidents occurred in areas with a speed limit of 30 km/h or below. A lower speed entails a higher possibility to ensure sufficient time to be able to detect a pedestrian, which means a lower risk of fatal injury (Rosén & Sander, 2009). According to recommendation #8 in 'Saving Lives Beyond 2020: The Next Steps' as well as the recommendations based on Vision Zero, the vehicle speed should be limited to 30 km/h in areas where pedestrians and vehicle co-exist (STA, 2019; Stigson & Kullgren, 2010). Several cities in Europe (e.g. Paris, Bilbao, and Brussels) have implemented 30 km/h and it will be very interesting to follow the consequences of these implementations. The present study also shows that many fatalities occurred in crossings, highlighting the need for controlling and reducing the speed in crossings or to implement level crossings. Previous studies have shown that for example speed bumps are effective (Agerholm et al., 2017; Lee et al., 2013; Naci et al., 2009; Pucher et al., 2010). In the present study no accidents occurred on a road with speed reducing interventions. It was also identified that the intervention with the largest total preventive potential was pedestrian crossings with speed calming measures for the motor vehicles. Several studies have shown that there is an increased risk for serious and fatal injury for pedestrians above 60 years age (Rosén et al., 2011; Stigson & Kullgren, 2010). The result from this study show that half of the pedestrian fatalities were above 65 years age. Both the road network and the road environment need to be designed to ensure the safety for this age category. To Achieving equality in the transportation systems is also recognized in the UN sustainability goals. Another risk group identified was pedestrians 18 years or younger in or on their way to school (22 of the 226). The public and private sectors have an important role to increase their safety and this role is highlighted in recommendation #4 in ‘Saving Lives Beyond 2020: The Next Steps’, in which they are suggested to ensure that children have a safe road environment that enables them to walk and ride a bicycle.

The impact of vehicle safety technology countermeasures

In addition to changes in the road infrastructure, improvements of the vehicle fleet to protect pedestrians will be very important. Recent studies show that existing AEB aimed for pedestrians only has a 20–25% crash reduction, and no reduction in darkness (Cicchino, 2022; Kullgren et al., 2022). It is in the present study assumed that AEB for vulnerable road will have a large positive effect for avoiding or mitigating pedestrian fatalities. With this system available, 63% of the pedestrian fatalities on urban roads and 50% of those on rural roads could potentially be prevented. However, it was identified that the full potential for vehicle interventions would not occur until many years ahead, not until 2050 with the current expected implementation rates of vehicle safety technologies. In Sweden the average lifetime of vehicles is approximately 18 years, that entails a slow implementation rate. In this study the average age of the vehicles in collision with pedestrians was approximately 10 years. Actions aimed to speed up the implementation rate are required to reach the 2030 sustainability goals. Vehicle procurement policies can be an effective tool to increase the implementation rate. It was estimated that in 2021, 58% of all new vehicles in Sweden were purchased by a business in private or public sector. A recent study shows that it is more common to include environmental requirements than safety requirements in procurement policies (Klingegård et al., 2022).

It was found that most of the accidents on rural roads (71%) and about half on urban roads (47%) occurred under conditions in which a combination of light, weather, and/or sight distance would complicate the detection of the pedestrian by the vehicle safety sensors. In approximately one out of five accidents (19%) the pedestrian was partly obscured by other vehicles or objects, which highlights current design challenges for AEB systems. Since 2018 Euro NCAPs test protocol cover tests of AEB in darkness and with an obscured view. However, studies of real-world pedestrian collisions clearly show that the AEB systems need to be improved, especially in darkness (Cicchino, 2022; Kullgren et al., 2022). With the current performance of pedestrian AEB many of the accidents would be difficult to mitigate.

Methodological considerations

For each accident, predictions were made whether the accident would lead to a fatality if it occurred in 2030 or 2050. The predictive methodology has successfully been used in previous studies (Kullgren et al., 2017; Strandroth et al., 2012). Fatal accidents that were estimated to be avoided by 2030 or 2050 were removed from the accident sample for the next step, thus providing a population of future crashes that will require further actions. However, there are some limitations with this method.

The study included all fatal crashes with pedestrians in Sweden that occurred during the sampling period (except suicides). The study does not cover serious injuries. Further research should be performed regarding the protection of pedestrians sustaining serious injuries. Preventive interventions aimed for serious injury could differ from those aimed for fatalities. It could also be the case that an avoided fatality will lead to a serious injury. This should also be studied further.

If each fatally injured pedestrian, especially older ones, could have used other transport modes was not considered. Each pedestrian was assumed to still be a pedestrian. Also, the age of the pedestrian was not considered as a factor in the analysis (e.g., there may be a difference in effectiveness of interventions dependent on age group (children vs. adults).

Moreover, the method does not account for the influence of changes of post-crash interventions, such as rescue, hospital care, and rehabilitation on fatality outcomes. It was assumed that rescue, emergency care and rehabilitation would have the same standard in Sweden during the analysis period. Furthermore, it is difficult to account for future trends which may impact the prognoses made in this study. For instance, it is predicted that geo-fencing could have large impact but there is no data so far that can be used to account for the effect of this technology. Other measures such as pedestrian use of reflective material (Lahrmann et al., 2018) and use of illuminated crossings (Høye & Elvik, 2012) have not been included as it to date is difficult to estimate their effects. In the analysis of the future potential, it was assumed that changes to the infrastructure was introduced immediately because implementations rates could not be assessed for these interventions. However, the implementation of these will take long time. And it should be noted that a systematic implementation of the analysed countermeasures of the whole road network is necessary to achieve the safety potentials shown in the present paper.

While the material used was fully representative for Swedish conditions, in such retrospective studies it may be difficult to consider the possible behavioural adaption that could follow the implementation of certain countermeasures. The results apply to Swedish conditions as is based on the current infrastructure and vehicle fleet in Sweden and also the expected developments in Sweden. For example, differences in implementation rates within European countries have been seen. However, recent EU directives have made the differences smaller over the years. One should therefore be careful to generalize the results and proposed interventions to other countries.