Evaluating the role of speed reduction on the severity of vehicle-pedestrian crashes: a case study in Palermo, Italy

Abstract

Introduction:

This study examined the influence of real-world vehicle speeds on injury outcomes in police-reported vehicle-to-pedestrian crashes in Palermo, Italy. Specifically, the study quantified the relationship between percentile travel speeds and the risk of fatal and serious injuries (FSI) for pedestrians and assessed the potential effects of small reductions in travel speed using counterfactual scenarios.

Methods:

Pedestrian crash locations from police-reported vehicle-to-pedestrian crashes in Palermo, Italy, were linked with real-world travel speed data obtained from TomTom Move. Percentile travel speeds (15th, 50th, and 85th) were used as proxies for impact speed to estimate FSI risk using established risk functions. A counterfactual analysis was conducted to assess the potential effects of reducing travel speeds by 3 km/h and 5 km/h across the speed distribution on pedestrian injury risk.

Results:

A strong positive relationship was observed between percentile travel speeds and increased FSI risk, with substantially higher risks at upper percentile speeds. The counterfactual analysis indicated that reductions of 3 km/h and 5 km/h across the speed distribution could meaningfully reduce the risk of fatal and serious pedestrian injuries.

Discussion:

The findings suggest that even modest reductions in vehicle speed can substantially improve pedestrian safety, particularly in urban environments with high pedestrian activity. These results highlight the potential value of targeted speed management strategies for reducing the severity of vehicle-to-pedestrian crashes.

1 Introduction

Road traffic fatalities and serious injuries (FSI) remain a significant global public health challenge (WHO, 2023). According to the Global Status Report on Road Safety, road traffic crashes result in approximately 1.19 million fatalities each year worldwide (WHO, 2023). Similarly, according to the Road Safety Annual Report prepared by the International Traffic Safety Data and Analysis Group (IRTAD), based on road safety data from IRTAD member countries, road traffic fatalities decreased by 2.8% in 2024 compared with 2023; however road crashes still claimed nearly 80,000 lives (ITF, 2025). This persistent challenge is also evident in the European Union, where road crashes in the EU-27 resulted in 19,932 fatalities and approximately 1.15 million injuries in 2024 (European Commission, 2026). Within this broader road safety burden, pedestrians remain among the most vulnerable road users, accounting for 21% of global road traffic fatalities (WHO, 2023). In the European Union, 3,560 pedestrian fatalities were reported in 2024, confirming that pedestrian safety remains a major regional concern despite overall road safety improvements (European Commission, 2026). A large share of these pedestrian deaths is directly associated with interactions between pedestrians and cars. In 2024, 2,302 of the 3,560 pedestrian fatalities in the EU, corresponding to 65%, involved cars as the main collision opponent (European Commission, 2026). This high proportion reflects the severe injury consequences of car-pedestrian collisions, particularly because pedestrians lack physical protection and are therefore more susceptible to FSI during crashes (Levulytė et al., 2026).

This car-pedestrian pattern is particularly pronounced in Italy, where pedestrian safety remains a major concern within the wider European context. In 2024, Italy recorded 470 pedestrian fatalities, the second-highest number in the EU after Romania and higher than those recorded in France and Poland (European Commission, 2026). More importantly, 72% of pedestrian fatalities in Italy involved cars as the main collision opponent, exceeding the EU average of 65% and representing the third-highest proportion among EU Member States after Slovenia (83%) and Lithuania (74%) (European Commission, 2026). These figures indicate that pedestrian safety in Italy is not only a matter of overall pedestrian vulnerability, but is strongly linked to car-pedestrian interactions. National statistics further confirm the severity of this issue: according to the Italian National Institute of Statistics (ISTAT), pedestrians experience a fatality rate of 2.7 per 100 crashes, which is 4.4 times higher than that of car occupants; pedestrian fatalities also increased by 3.2% in 2022 compared with 2021 (ISTAT, 2022). Such risks are especially critical in tourist, historic, and densely populated urban areas, where high pedestrian activity and exposure to car traffic increase the likelihood of car-pedestrian conflicts (Congiu et al., 2019). Palermo represents one such urban context. Its historic urban layout and urbanisation-related mobility challenges contribute to pedestrian safety risks (Amoroso and Caruso, 2008; Potoglou et al., 2018). For instance, between 2015 and 2022, Palermo Municipality Police recorded 1,999 pedestrian crashes, highlighting the need for targeted interventions to mitigate pedestrian crashes in the city.

Several factors contribute to pedestrian injury severity, including socio-economic factors (Davis, 2001; Henary et al., 2006; Rosén and Sander, 2009; Tay et al., 2011), environmental characteristics (Li et al., 2017; Sanders et al., 2022; Siddiqui et al., 2006), roadway features (Moradi et al., 2019; Pulugurtha and Sambhara, 2011; Tulu et al., 2015) and vehicle attributes (Desapriya et al., 2010; Kim et al., 2010; Lefler and Gabler, 2004; Roudsari et al., 2004; Tefft, 2013). Among these, vehicle speed has consistently been identified as one of the most influential determinants of pedestrian injury severity. In this regard, previous studies have examined different aspects of speed, including impact speed, mean or operating speed, and posted speed limits. Impact speed is directly related to the biomechanical forces transferred to the pedestrian during a collision and has been strongly associated with the probability of severe and fatal injury outcomes (Davis, 2001; Kong and Yang, 2010; Rosén and Sander, 2009; Tefft, 2013). Mean or operating speed reflects the general traffic-speed environment and influences both the likelihood of pedestrian–vehicle conflicts and the time available for drivers and pedestrians to perceive, react, and avoid a collision (Kröyer, 2015; Leaf and Preusser, 1999). Posted speed limits, in contrast, represent the regulatory speed environment and have been used in previous studies as an indicator of the severity risk associated with crashes occurring in higher-speed road contexts (Ballesteros et al., 2004; Gårder, 2004; Islam, 2023).

The established relationship between vehicle speed and pedestrian injury severity has led to growing attention to interventions that reduce vehicle speeds before pedestrian–vehicle conflicts occur. Speed-limit reductions are one such intervention, as they influence the legal and operational speed environment in which drivers approach pedestrian activity areas. In this context, Islam (2023) examined pedestrian crash severity across different speed-limit zones and found that higher-speed environments were associated with more severe pedestrian injury outcomes across different roadway types. This finding suggests that pedestrian safety interventions should account for the speed environment in which crashes occur, rather than treating all pedestrian crash locations as having the same severity risk. In addition to speed-limit policies, pedestrian crossing control devices can influence driver behaviour at locations where pedestrian exposure is concentrated. Specifically, Ugan et al. (2022) evaluated Pedestrian Hybrid Beacons (PHBs) and Rectangular Rapid-Flashing Beacons (RRFBs) at midblock crossings and uncontrolled intersections and found that both treatments contributed to reductions in driver speed near pedestrian crossing locations. These devices are relevant because they increase the visibility and priority of pedestrians at conflict points, thereby encouraging earlier driver response and lower approach speeds. Similarly, street lighting has been identified as an important countermeasure for improving pedestrian safety, particularly during nighttime and low-visibility conditions. In this regard Li et al. (2023) analysed the effect of horizontal illuminance on nighttime pedestrian crashes using crash modification factors and found that increasing average light levels from low-to-median and median-to-high illuminance reduced the probability of nighttime pedestrian crashes on midblock segments by 77.5%, 81.2%, and 85.5%, respectively. In addition Pazzini et al. (2024) evaluated the effectiveness of enhanced pedestrian crosswalk lighting and found that enhanced lighting reduced vehicle approach speeds, with 72% of drivers travelling at a speed that allowed the vehicle to stop safely compared to standard street lighting (10%). In addition, under enhanced lighting, 85% of drivers stopped at pedestrian crossings compared with only 26% under standard street lighting. Moreover, traffic calming measures address the problem through a different mechanism by physically modifying the road environment to discourage high-speed driving. For example, Damsere-Derry et al. (2019) analysed the effect of speed tables in Ghana and found reductions in average vehicle speed, fewer vehicles exceeding the 50 km/h speed limit, and a lower probability of pedestrian fatalities in areas where traffic calming devices were installed., These findings indicate that speed-reduction measures can contribute to pedestrian safety through multiple pathways, including regulatory controls, improved driver awareness at crossings, enhanced street lighting, and physical modifications to the road environment.

The effectiveness of speed-reduction measures also depends on the broader road safety context in which they are implemented. Pedestrian risk is rarely determined by vehicle speed alone; it is also shaped by road design, crossing provision, driver compliance, enforcement intensity, land-use patterns, and the degree to which safety measures are coordinated at the local level (Kayisu et al., 2024; Mikusova et al., 2024). Infrastructure-based measures can reduce operating speeds by changing the physical driving environment, while enforcement can improve compliance with posted limits and discourage risky driving behaviour. At the same time, coordinated road safety management can support the selection and implementation of measures that respond to local crash patterns rather than applying isolated interventions without regard to site-specific conditions. Earlier studies on road safety management and transport policy similarly emphasize the importance of institutional coordination, local safety planning, and integrated management structures in improving road safety outcomes (Mikusova, 2016; Mikušová, 2011; Mikusova et al., 2024). This broader perspective is particularly important for pedestrian safety, since pedestrians have limited physical protection and depend strongly on the design and regulation of the surrounding road environment. Accordingly, the speed-reduction scenarios examined in this study are positioned within a wider body of evidence showing that pedestrian injury severity can be reduced through a combination of lower vehicle speeds, safer crossing environments, enforcement, and coordinated urban road safety management.

Among the different dimensions of vehicle speed, impact speed has been widely recognised as the most direct and critical factor influencing pedestrian injury severity, as illustrated by the S-shaped fatality risk curve (Davis, 2001; Kong and Yang, 2010; Rosén and Sander, 2009). However, despite its importance, real-world crash data often suffer from significant limitations, including biases toward severe and fatal injuries and inaccuracies in estimating actual impact speeds at crash sites (Hussain et al., 2019; Rosén et al., 2011). In addition, research (Kröyer, 2015) highlighted that integrating impact speed into speed management policies presents considerable challenges. For instance, while higher mean travel speeds at crash sites typically correlate with higher impact speeds, individual crash-involved vehicles may have been travelling at speeds significantly lower or higher than the mean traffic speed at that location (Kröyer, 2015).

To address these challenges, researchers have explored alternative measures, including mean travel speed (Kröyer, 2015; Leaf and Preusser, 1999), and speed limits and designated speed zones (Gårder, 2004; Leaf and Preusser, 1999; Mitchell et al., 2015; Mueller et al., 1988; Olszewski et al., 2015). While these single-dimensional speed indicators are positively associated with an increased risk and severity of pedestrian crashes, they fail to fully capture the effect of speed variability across different speed ranges. This oversight neglects the influence of speed fluctuations, which can introduce unpredictable and hazardous conditions for pedestrians. Prior research (Elvik, 2019) has highlighted this limitation, arguing that mean speed alone is a poor predictor of traffic injuries and that crash severity varies significantly with the speed distribution.

To bridge this gap, speed distribution, particularly percentile speeds (e.g., 15th, 50th, and 85th percentiles), has been proposed as a more effective metric for assessing pedestrian safety (Arias et al., 2021; Mahmoud et al., 2023). Unlike single-speed measures, percentile speeds capture variations in driver behaviour, offering more profound insights into the range of speeds at which pedestrians may be struck. Prior research (Arias et al., 2021; Mahmoud et al., 2023) has demonstrated that higher percentile speed ranges are associated with a higher frequency of pedestrian crashes than lower percentile ranges. This demonstrates that speed variability, especially at higher ranges, plays a critical role in crash risk. However, despite their potential, the relationship between percentile speeds and pedestrian injury severity remains underexplored, mainly representing a significant gap in the existing literature.

This study seeks to address this issue by investigating the relationship between vehicle speed, specifically percentile speeds, and pedestrian injury severity in the urban context of Palermo, Italy. To achieve this objective, the study used TomTom’s Move speed percentile data (15th, 50th, and 85th) as a proxy for travel speeds. The analysis evaluates how variations in speed, particularly in low-to-median and median-to-high ranges, impact the likelihood of FSI in vehicle-to-pedestrian crashes. In addition, the study incorporates a counterfactual analysis to assess the effects of reducing travel speeds by 3 km/h and 5 km/h to understand their effect on FSI severity risk reduction and vehicle speed reduction. Analysing the effect of speed reduction through a counterfactual analysis, particularly in understanding the changes in pedestrian crash severity and other crash severities before and after speed limit reductions, has proven effective, and studies such as (Fridman et al., 2020) found that reducing speed limits from 40 to 30 km/h resulted in a 28% reduction in pedestrian motor vehicle collisions, whereas Hu and Cicchino (2024) found that reducing speed limits on non-arterial streets (from 25 to 20 mph) and on arterial streets (from 30 to 25 mph) resulted in a 17.2% reduction in crashes across different injury severity levels, including fatal, disabling, and evident injuries. Therefore, the study’s findings aim to provide actionable insights for urban planners and policymakers seeking to implement targeted speed management strategies to enhance pedestrian safety in Palermo and similar urban settings.

2 Data description

The study is based on two separate datasets. The pedestrian crash dataset was obtained from the Palermo Municipality Police, and real-world speed data were sourced from TomTom Move, which provides traffic analytics and mobility data derived from vehicle-based GPS data collected anonymously from a range of devices.¹

2.1 Crash data

The Palermo Municipal Police (Polizia Municipale) provided the crash data used in this study, spanning 8 years from 2015 to 2022. The dataset provides detailed information on pedestrian-related crashes within the municipality, focusing on various crash attributes, pedestrian and driver characteristics, and environmental and traffic-related factors. These attributes include crash severity, crash locations, temporal factors (e.g., time of day, day of the week, seasonal variations), demographic characteristics of both pedestrians and drivers (e.g., age, gender), environmental conditions (e.g., weather, visibility), and roadway features (e.g., road type, pedestrian crossings, lighting, road conditions). In addition, vehicle-related variables, such as vehicle type and crash-at-fault status (e.g., driver at fault, pedestrian at fault), were recorded. It is important to note that the dataset does not include detailed crash reconstructions, such as the exact mechanisms of the crash or the relative positions of the vehicles and pedestrians involved. Consequently, the probable crash causes reported by the police are based on their initial assessments and should not be treated as definitive explanations of the events leading to the crashes.

In this study, each crash in the dataset was categorised by the severity of injuries sustained by the pedestrian. Injury severity was classified into four categories: “Died”, “Seriously Injured”, “Injured”, and “Not Injured.” This classification provided a clear framework for analysing the relationship between crash characteristics and injury severity. However, it is worth noting that to ensure the analysis focused on pedestrian crashes involving direct impact by a moving vehicle, crashes attributed to other factors were excluded. These excluded factors included cases in which pedestrians were struck by vehicles in reverse, fell for other reasons, or were involved in crashes outside regular traffic settings (such as non-traffic-related incidents). As a result, a final sample of 54 fatal crashes and 87 serious injury crashes was extracted for detailed analysis, providing a comprehensive dataset of pedestrian injury crashes with fatal and serious injury outcomes. Therefore, from the pedestrian crash dataset, only study-specific variables such as crash locations, injury severity, crash distribution of severity by each year, and roadway types were considered for analysis, as shown in Table 1. Considering the total number of crashes (141) and the differences among the crash locations, it was difficult to report each crash location in Table 1; therefore, crash locations were not included in Table 1. Among these variables, crash locations were used in the speed database to identify and acquire the speed profiles for the road segments where crashes were reported, whereas injury severity from Table 1 was used to divide the dataset into two sub-datasets, including fatal (54) and serious injury (87) crash locations, and Equations 1, 2 were then used to estimate FSI risk accordingly. In addition, the crash distribution of severity by each year in Table 1 was solely used to display the total number of crashes reported in each year and the severity of crashes throughout each year, whereas the road types in Table 1 were considered to understand the contextual background where crashes were reported and to understand the speed behaviour of traffic on these roadway types. Consequently, other variables, such as temporal factors (e.g., time of day, day of the week, seasonal variations), environmental factors, vehicle types, and crash-at-fault, etc., from the acquired pedestrian crash dataset were not considered in the analysis, as they were not relevant to the present study context.

Moreover, it is important to note that the variables obtained from the pedestrian crash dataset were not directly incorporated into Equations 1, 2 for estimating FSI risk. Instead, the crash dataset was primarily used to identify the crash locations, from which the corresponding speed profiles were obtained from the speed database and subsequently applied in the risk estimation process.

2.2 Speed data

To assess the relationship between travel speed and FSI in pedestrian crashes, crash locations and their geographic coordinates from the crash datasets were used to obtain speed profiles for each location from the TomTom Move speed database. The TomTom Move “Traffic Stat” tool was employed to generate comprehensive speed reports for each crash site at a particular road segment. These reports covered 1 month and assessed the crash site road segment over 24 h a day, across all days of the week, providing detailed data on vehicle speeds at the crash locations. Although the TomTom Move speed reports covered 1 month, 24 h per day, and all days of the week, the analysis did not use speed data aggregated across the entire day. Instead, for each crash location, only the 1-h interval corresponding to the recorded crash time was extracted from the TomTom Move speed report and used in the analysis. For example, if a crash occurred at 12:45 pm, only the speed data for the 12:00–1:00 pm interval was considered. This approach ensured that the percentile speeds represented the prevailing traffic speed behaviour during the same hourly period in which the crash occurred while being derived from 1 month of observations across all days of the week. In addition, the crash sites were identified using the tool’s search function, which enabled precise selection of the relevant road segments within the roadway network corresponding to each crash location.

The analysis considered various road types for speed profile collection to ensure an accurate representation of traffic flow conditions at each crash site across the road segment. For one-way streets, speed profiles were collected specifically for the direction of travel where the crash occurred. On two-way roads, speed profiles were collected for both directions of travel to account for potential differences in vehicle speed by direction. Subsequently, average speed data were collected for both directions. Particular attention was given to crashes occurring at intersections. In these cases, the leg of the intersection where the crash occurred was determined from statements provided by the individuals involved in the incident, as recorded in the police reports in the dataset. When crashes occur during turning movements, the relevant intersection leg was identified as the direction from which the vehicle entered the turn, ensuring the speed data reflected the specific road conditions and turning behaviour at the time of the crash. Moreover, in terms of route sample size, for the 54 fatal crash locations, the estimates were derived from a total of 29,208 route observations (mean = 541, median = 358 observations per location; IQR: 74–744), whereas for the 87 serious injury crash locations, the estimates were derived from a total of 41,938 route observations (mean = 482, median = 341 observations per location; IQR: 125–694).

The percentile speed data collected at FSI crash locations enabled the application of a model to forecast FSI risks. The analysis aimed to assess the impact of different percentile speed distributions on pedestrian FSI risks and to explore potential interventions, such as speed reductions, to mitigate the severity of pedestrian injuries at FSI crash locations in the study area. In addition, the interpretation of these speed profiles requires consideration of Palermo’s urban mobility and road-network context. According to the Palermo Sustainable Urban Mobility Plan (PUMS), mobility in the city is strongly characterised by private motorised travel, particularly by cars and motorcycles, which account for 69.2% of weekday trips, followed by public transport (15.6%) and active modes (15.2%) (Città Metropolitana di Palermo, 2023). The high volume of private vehicles circulating across the road network is particularly significant because pedestrian crashes are more likely to occur where intense vehicle flows, high pedestrian exposure, and constrained urban street layouts interact.

The analysed crash locations reflect this heterogeneous urban network. As reported in Table 1, they included one-way streets (37.6%), two-way roads (55.3%), dual carriageways (5.7%), and carriageways with service roads (1.4%). These roadway configurations represent a range of operating conditions, from local urban streets to higher-capacity urban roads. Palermo’s current mobility planning also includes several measures aimed at managing vehicle speeds and improving pedestrian conditions, such as limited traffic zones, pedestrian areas, 30 km/h zones, speed-control measures, traffic-calming interventions, and pedestrianisation strategies in historic areas (Città Metropolitana di Palermo, 2023). In particular, a limited traffic zone with road pricing policies was implemented in the historic city centre, and pedestrianised areas in Palermo have been extended to more than 340,000 square meters following the implementation of pedestrianisation policies in the historic urban area. These measures provide the broader policy and mobility context for the travel speed conditions analysed in this study.

Within this context, the TomTom percentile speeds should be interpreted as indicators of prevailing vehicle speed conditions at the selected pedestrian crash locations, rather than as direct measurements of impact speed for individual crashes. The 15th, 50th, and 85th percentile speeds provide a consistent basis for comparing baseline and reduced-speed scenarios across the observed crash locations. However, the speed data are not disaggregated by vehicle type or by detailed traffic composition at the exact time of each crash. Therefore, the findings are most transferable to comparable dense urban environments where pedestrian exposure, mixed traffic, and speed-management challenges are similar to those observed in Palermo.

3 Methods

3.1 Estimation of FSI risk

The representative percentile speeds (15th, 50th, and 85th) at the crash locations were applied to estimate FSI risks using the risk equations (Davis, 2001), as shown below. The Davis (2001) model was selected because it provides a continuous relationship between vehicle speed and pedestrian injury risk and can therefore be applied consistently across the baseline and counterfactual speed-reduction scenarios considered in this study. This suitability is particularly relevant because the Palermo crash dataset does not include detailed crash reconstruction variables, such as observed impact speed, vehicle characteristics, or collision configuration, which would be required to apply more recent injury risk curves without introducing additional assumptions. Therefore, the Davis (2001) model was considered appropriate for the comparative estimation of changes in FSI risk under different speed-reduction scenarios. In addition, the pedestrian age information in the acquired pedestrian crash dataset was recorded only as a broad binary category, distinguishing pedestrians below 18 years of age from those aged 18 years and above. Therefore, it does not provide exact pedestrian ages or detailed age-group classifications corresponding to the Davis (2001) age groups (60+, 15–59, and 0–14 years). Consequently, the available binary age category could not be aligned with the specific age ranges used in the Davis (2001) model, and pedestrian age data from the crash dataset were not incorporated into the model. Due to the unavailability of exact pedestrian age data, the parameter estimates for the 15–59 age group were used as a broad reference category for FSI risk estimation.

3.1.1 Fatal risk estimation

3.1.2 Serious injury risk estimation

Where is the impact speed (km/h), for which the percentile speed was used as a proxy in this study; is the slope parameter representing the average increase in damage for each unit increase in impact speed; and and are the two damage thresholds used in the ordered logit model proposed by Davis (2001). These equations were initially designed to estimate FSI risks across different age groups based on impact speed, using data from pedestrian crashes. Davis (2001) refined the injury risk curve, providing a more accurate model compared to earlier works by Ashton (1980) and Pasanen (1992). Specifically, Davis (2001) estimated model parameters for three distinct age groups: 60+, 15-59, and 0–14 years. However, since pedestrian age data were not available with exact pedestrian ages in the acquired crash dataset, the logit model parameter estimates for the 15-59 age group, such as (, , ) mentioned above, were used in both Equation 1 and Equation 2 for risk calculations. Subsequently, the FSI risk for each pedestrian crash was calculated at each of the speed percentiles (15th, 50th, and 85th).

3.2 Speed reduction scenarios

Two sample speed reductions were investigated for the counterfactual analysis, based on Davis (2001), to explore the potential benefits to pedestrian FSI risk from reduced speeds in vehicle-to-pedestrian crash scenarios. The percentile speed (15th, 50th, and 85th percentile) profiles for each crash were reduced by 3 and 5 km/h using a trial-and-error method, and the resultant FSI risks were estimated using Equations 1, 2 as before. After that, the risk ratio reduction between before and after FSI was examined using log-scale plotting to understand the impact of reduced speed on FSI risk reduction.

4 Results

4.1 Speed and risk

The analysis investigates the impact of speed reductions on pedestrian injury severity, focusing on FSI. The results are presented in Tables 2, 3, which summarise the mean speed and associated risks before and after speed reductions across three percentile speed categories: 15th, 50th, and 85th. Both Tables 2, 3 provide quantitative evidence of the relationship between travel speed distributions and FSI risk in vehicle-to-pedestrian crashes. The “mean speed” for each percentile category represents the average speed observed within that specific percentile range for all FSI crashes. In contrast, the “mean risk” reflects the average risk of FSI, as calculated using Equations 1, 2 for each crash.

The data presented in Tables 2, 3 reveal a clear trend: as percentile speed increases, both the mean speed and the associated risks increase. Specifically, the difference in mean speed and risk between lower and higher percentiles becomes more pronounced, with the gap between the 15th and 50th percentiles relatively small compared to the larger difference between the 50th and 85th percentiles. This trend strongly suggests that higher vehicle speeds are directly linked to an increased risk of serious pedestrian injury outcomes in FSI crashes, particularly in the context of pedestrian vulnerability.

4.2 Counterfactual analysis

Figures 1–3 present the estimated fatality risk distributions for the 15th, 50th, and 85th percentile speeds at fatal pedestrian crash locations before and after the 3 km/h and 5 km/h speed-reduction scenarios. The fatality risk distributions show a clear increase across the percentile speed ranges, with the lowest estimated fatality risk associated with the 15th percentile speed and the highest estimated fatality risk associated with the 85th percentile speed. Figure 1 shows the distribution at the 15th percentile speed, representing the lower range of observed travel speeds at fatal crash locations. Figure 2 presents the distribution at the 50th percentile speed, representing the median speed condition, where estimated fatality risk is higher than at the 15th percentile but lower than at the 85th percentile. Figure 3 presents the distribution at the 85th percentile speed, representing the higher-speed condition and the highest estimated fatality risk. Across all three percentile speed ranges, the 3 km/h and 5 km/h reduction scenarios shift the fatality risk distributions toward lower values, indicating that reduced vehicle speeds are associated with lower estimated fatality risk.

FIGURE 1

FIGURE 2

FIGURE 3

Figures 4–6 present the estimated serious injury risk distributions for the 15th, 50th, and 85th percentile speeds at serious injury pedestrian crash locations before and after the 3 km/h and 5 km/h speed-reduction scenarios. Similar to the fatality risk results, the estimated serious injury risk increases across the percentile speed ranges, with the lowest risk associated with the 15th percentile speed and the highest risk associated with the 85th percentile speed. Figure 4 shows the distribution at the 15th percentile speed, Figure 5 shows the distribution at the 50th percentile speed, and Figure 6 shows the distribution at the 85th percentile speed. The 3 km/h and 5 km/h reduction scenarios also shift the serious injury risk distributions toward lower values across all percentile speed ranges. However, the magnitude of the estimated serious injury risk and the extent of reduction differ from those observed for fatality risk, as the two outcomes are estimated using separate risk equations and different crash-location samples. Therefore, the results indicate a consistent direction of effect for both injury outcomes, while maintaining a distinction between fatality risk and serious injury risk.

FIGURE 4

FIGURE 5

FIGURE 6

5 Discussion

Pedestrians are among the most vulnerable road users and understanding the factors that influence crash severity is critical to improving road safety. Traditional speed measures, such as impact speed, mean travel speed, and speed limits, offer valuable insights but often fail to account for the variability in travel speed distributions, which significantly affects crash outcomes. Travel speed at the time of a collision depends on factors such as braking response, driver reaction time, and vehicle control, making it essential to analyse how speed changes across low-to high-impact crash severities. Although impact speed directly measures the collision force (Pitt et al., 1990), its variability across crashes limits its utility in shaping speed management policies (Kröyer, 2015).

To overcome these limitations, percentile speeds have emerged as a reliable proxy for capturing the full spectrum of travel speed distributions. Unlike one-dimensional speed measures, percentile speeds provide a comprehensive view of speed ranges, such as low to median (15th-50th percentile) and median to high (50th–85th percentile). The results of this study also demonstrate that variability in travel speed significantly affects vehicle-to-pedestrian crash outcomes, as crash risk increases with higher percentile speeds. Specifically, the median-to-high (50th-85th percentile) speed range is associated with higher FSI crash severity than the low-to-median (15th-50th percentile) range, as shown in Tables 2, 3 and Figures 1–6. For fatal crashes, the estimated crash risk at the 15th, 50th^, and 85th percentiles of speed is 0.7%, 2.6%, and 8.2%, respectively.

In contrast, for serious injury across each percentile speed, the estimated crash risk is 13%, 26.8%, and 43.6%, respectively, as reported in Tables 2, 3. This shows that across the upper percentile range (50th-85th), FSI crash risk increases compared to the lower percentile range (15th-50th). This is likely because drivers may struggle to adjust rapidly to sudden changes in traffic conditions, leading to higher-impact collisions. This finding is supported by recent studies (Arias et al., 2021; Mahmoud et al., 2023), which demonstrated significant influences of travel speed distributions on crash outcomes and indicated that the upper percentile speeds are associated with increased pedestrian crash frequency.

The current study further demonstrates that even modest speed reductions of 3–5 km/h can measurably decrease the risk of pedestrian FSI and support model efficiency (Davis, 2001) as reported in Tables 2, 3. While a 3–5 km/h reduction might appear minimal on an individual roadway, implementing such reductions across a broader network can lead to substantial improvements in pedestrian safety. This finding is particularly relevant in urban areas and high pedestrian-activity zones, where the likelihood of pedestrian-vehicle interactions is greatest. One practical approach to achieving these speed reductions is to adjust existing speed limits. Effective speed limit adjustments require a nuanced understanding of traffic dynamics and roadway design, including factors such as traffic volume, lane count, roadway type, and pedestrian exposure. Research indicates that targeted speed limit reductions can lead to significant decreases in both travel speeds (Heydari et al., 2014; Islam et al., 2014) and the severity of crashes (Bornioli et al., 2020; Cairns et al., 2015; Fridman et al., 2020; Hu and Cicchino, 2024; Islam and El-Basyouny, 2015).

For example, an analysis of TomTom Move speed datasets shows that most urban roadways in Palermo maintain a standard speed limit of 50 km/h, which is typical for built-up areas in Italy. However, Palermo’s unique characteristics, its historical urban layout, high pedestrian density, and significant tourist attractions suggest that a 50 km/h limit may be too high in specific contexts. In zones such as the historic city centre, residential areas, and school districts, where pedestrian volumes are particularly high, a lower speed limit may be more appropriate. Moreover, Palermo’s cultural heritage sites and popular tourist attractions further increase pedestrian density, thereby increasing the risk. In this regard D’Orso and Migliore (2020) considered the speed perspective in their assessment of the pedestrian infrastructure model in the Palermo context, emphasising that 30 km/h is considered safe for pedestrians compared to 50 km/h. Further, they found that high vehicle speeds can significantly affect pedestrians’ perceptions of safety. Although Palermo has already introduced a range of speed-management and pedestrian-oriented measures, these findings suggest that further opportunities may exist to improve pedestrian safety through additional reductions in travel speed, particularly in locations with high pedestrian activity.

In light of these observations, the appropriateness of a 50 km/h speed limit should be evaluated in the context of localised traffic conditions, pedestrian density, and urban design. In high-risk areas identified through crash data, reducing speed limits to 45 or 40 km/h could be a viable strategy, warranting further evaluation of its impact on travel speeds, crash severity, and crash frequency. Supporting this approach, a case study in Seattle demonstrated that reducing speed limits by five mph (approximately 8 km/h) led to a 17.2% decrease in crash severity and a 19.9% reduction in crash frequency (Hu and Cicchino, 2024). These findings highlight the potential benefits of targeted speed limit adjustments in urban environments.

In addition, in one illustrative case, the fatal crash site on “Viale Ercole” in Palermo, where the designated speed limit is 50 km/h, recorded 15th, 50th, and 85th percentile speeds of 52 km/h, 70 km/h, and 92 km/h, respectively, during a 1-month sample period. These findings indicate frequent speed limit violations and highlight the risks associated with higher driving speeds. To counter such behaviours, our study supports the installation of intelligent transport tools, such as speed cameras, which can effectively enforce compliance with reduced speed limits and reduce reliance on constant police monitoring.

Furthermore, the application of the “Movement & Place” framework (VicRoads, 2016) offers a promising approach to urban planning that can encourage pedestrian-friendly speeds. This framework integrates mobility needs (“Movement”) with social, commercial, and community functions (“Place”), allowing streets to be classified and redesigned based on their primary function. In areas that prioritise “Place” functions, such as city centres, tourist areas, and cultural or historical sites, speed limits can be lowered to enhance pedestrian safety and overall urban amenities. Conversely, streets that predominantly support high traffic volumes (“Movement”) may require tailored speed-reduction strategies to balance mobility with safety. In the historic city centre of Palermo, for example, applying the Movement & Place framework could guide the redesign of streets and neighbourhoods to prioritise pedestrian safety over vehicle mobility. However, the implementation of these enforcement and urban planning strategies should also consider the national regulatory framework, as the availability and application of such strategies vary across countries. In the Italian context, the current regulatory framework may limit the implementation of certain interventions. Therefore, these recommendations should be considered within the applicable regulatory context when applied to other urban settings.

6 Conclusion

The present study provides critical insights into the potential of speed management policies to mitigate the risk of pedestrian injuries. A key contribution of this research is the integration of police-reported pedestrian crash data with real-world percentile speed distributions derived from the TomTom Move database. Unlike approaches that rely mainly on posted speed limits or average travel speeds, the use of 15th, 50th, and 85th percentile speeds allows the analysis to capture lower, typical, and higher ranges of the observed speed environment at pedestrian crash locations. This provides a more detailed basis for examining how variations in real-world travel speeds may influence estimated pedestrian FSI risk. Using the historic city of Palermo as a case study, the research highlights the importance of considering changes in travel speed when evaluating vehicle-pedestrian crash severity, particularly through the lens of percentile speeds. The findings reveal that the high range of percentile speeds significantly influences crash outcomes, underscoring the need for targeted speed management strategies. In addition, the study demonstrates that even modest speed reductions (3–5 km/h) can substantially reduce the risk of FSI, particularly in areas with high pedestrian activity.

The research advocates for a comprehensive speed reduction framework that combines measures such as adjusting speed limits, implementing traffic-calming infrastructure, and deploying innovative enforcement technologies. These interventions can effectively lower vehicle speeds, reduce crash severity, and enhance overall pedestrian safety. The study supports the practical relevance of speed management measures for achieving sustainable and effective pedestrian safety improvements. The findings are particularly relevant for historic cities like Palermo, where pedestrian safety is a pressing concern, and they also have broader applicability to similar urban contexts worldwide.

However, the study is not without limitations. The absence of exact pedestrian ages in the crash dataset posed a significant constraint. To address this, the study utilised model parameter estimates for the 15–59 age group, which, while covering a broad demographic range, may not accurately reflect the FSI risks for younger or older pedestrians. Future research could address this limitation by incorporating more comprehensive datasets and exploring the differential impacts of speed on vulnerable age groups. In addition, the Davis (2001) injury risk model was not specifically calibrated using Palermo or Italian crash data. Therefore, the estimated FSI risks should be interpreted as comparative estimates within the counterfactual speed-reduction framework, rather than as locally calibrated absolute injury probabilities. Future research could improve this aspect by using local crash and injury data to calibrate pedestrian injury risk functions for the Palermo or broader Italian context. In addition, the relatively small sample of FSI pedestrian crashes analysed in this study should be considered when interpreting the findings, as a larger sample could provide a broader representation of crash conditions within the study area.

Furthermore, the study adopted percentile travel speeds (e.g., 15th, 50th, and 85th percentiles) to capture the effect of speed variability across different speed ranges, which may not be fully reflected through single-dimensional speed measures such as mean travel speed or speed limits. The findings of the study further demonstrated that higher percentile speed ranges were associated with increased pedestrian FSI risk, highlighting the importance of considering speed distribution in the assessment of pedestrian crash severity. Therefore, the use of percentile speeds provided useful insights into the relationship between travel speed distributions and pedestrian injury severity. However, although percentile travel speeds provided a useful representation of travel speed distribution, their use as proxies for impact speed may not fully capture the actual collision speed at the moment of impact due to factors such as braking behaviour, driver reaction, and evasive manoeuvres prior to impact. Future research could further investigate the relationship between percentile travel speeds and actual impact speeds in vehicle-to-pedestrian crashes to better understand how travel speed distributions correspond to pedestrian injury severity. Although the use of TomTom Move is supported by previous transportation studies employing TomTom Traffic Stats and by research comparing commercial FCD with independent traffic measurements (Pozzoni et al., 2023), the travel speeds used in this study and obtained from the TomTom Move database were not independently validated against direct on-site speed measurements at the crash locations. Moreover, travel speeds derived from the TomTom Move database are subject to inherent measurement uncertainty because they represent estimated travel speeds derived from Floating Car Data (FCD), rather than direct speed measurements of vehicles at the individual crash locations and times. In particular, Torrisi et al. (2026) compared TomTom FCD with fixed radar measurements in an Italian urban context and identified spatial and temporal variation in probe coverage and some underestimation of vehicle speeds. However, they also found that commercially available FCD can provide reasonable estimates of urban travel speeds, although accuracy may decrease when the number of observed trajectories is limited. These findings support the application of TomTom data in the present study while also reinforcing the need to interpret the estimated percentile speeds with appropriate caution.

References

• 1

  AmorosoS.CarusoL. (2008). Pedestrian Safety Indicators Study, 117. Southampton, United Kingdom: WIT Press, 455–464. 10.2495/SC080431

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  AriasD.EdererD.RodgersM. O.HunterM. P.WatkinsK. E. (2021). Estimating the effect of vehicle speeds on bicycle and pedestrian safety on the Georgia arterial roadway network. Accid. Anal. Prev.161, 106351. 10.1016/j.aap.2021.106351

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  AshtonS. J. (1980). “A preliminary assessment of the potential for pedestrian injury reduction through vehicle design,” in SAE Technical Papers. 10.4271/801315

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  BallesterosM. F.DischingerP. C.LangenbergP. (2004). Pedestrian injuries and vehicle type in Maryland, 1995-1999. Accid. Anal. Prev.36, 73–81. 10.1016/S0001-4575(02)00129-X

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  BornioliA.BrayI.PilkingtonP.ParkinJ. (2020). Effects of city-wide 20 mph (30km/hour) speed limits on road injuries in Bristol, UK. Inj. Prev.26, 85–88. 10.1136/injuryprev-2019-043305

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  CairnsJ.WarrenJ.GarthwaiteK.GreigG.BambraC. (2015). Go slow: an umbrella review of the effects of 20 mph zones and limits on health and health inequalities. J. Public Health (United Kingdom)37, 515–520. 10.1093/pubmed/fdu067

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7

  CongiuT.SotgiuG.CastigliaP.AzaraA.PianaA.SaderiL.et al (2019). Built environment features and pedestrian accidents: an Italian retrospective study. Sustain. Switz.11, 1064. 10.3390/su11041064

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  Damsere-DerryJ.EbelB. E.MockC. N.AfukaarF.DonkorP.KalowoleT. O. (2019). Evaluation of the effectiveness of traffic calming measures on vehicle speeds and pedestrian injury severity in Ghana. Traffic Inj. Prev.20, 336–342. 10.1080/15389588.2019.1581925

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  DavisG. A. (2001). Relating severity of pedestrian injury to impact speed in vehicle-pedestrian crashes: simple threshold model. Transp. Res. Rec.1773, 108–113. 10.3141/1773-13

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  DesapriyaE.SubzwariS.SasgesD.BasicA.AlidinaA.TurcotteK.et al (2010). Do light truck vehicles (LTV) impose greater risk of pedestrian injury than passenger cars? A meta-analysis and systematic review. Traffic Inj. Prev.11, 48–56. 10.1080/15389580903390623

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  D’OrsoG.MiglioreM. (2020). A GIS-Based method for evaluating the walkability of a pedestrian environment and prioritised investments. J. Transp. Geogr.82, 102555. 10.1016/J.JTRANGEO.2019.102555

  • CrossRef
  • Google Scholar

• 12

  ElvikR. (2019). A comprehensive and unified framework for analysing the effects on injuries of measures influencing speed. Accid. Anal. Prev.125, 63–69. 10.1016/j.aap.2019.01.033

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  European Commission (2026). ROAD SAFETY Annual Statistical Report on Road Safety in the EU 2026.

  • Google Scholar

• 14

  FridmanL.LingR.RothmanL.CloutierM. S.MacarthurC.HagelB.et al (2020). Effect of reducing the posted speed limit to 30 km per hour on pedestrian motor vehicle collisions in Toronto, Canada - a quasi experimental, pre-post study. BMC Public Health20:20, 56. 10.1186/S12889-019-8139-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  GårderP. E. (2004). The impact of speed and other variables on pedestrian safety in Maine. Accid. Anal. Prev.36, 533–542. 10.1016/S0001-4575(03)00059-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  HenaryB. Y.IvarssonB. J.CrandallJ. R. (2006). The influence of age on the morbidity and mortality of pedestrian victims. Traffic Inj. Prev.7, 182–190. 10.1080/15389580500516414

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  HeydariS.Miranda-MorenoL. F.LipingF. (2014). Speed limit reduction in urban areas: a before-after study using Bayesian generalized mixed linear models. Accid. Anal. Prev.73, 252–261. 10.1016/j.aap.2014.09.013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  HuW.CicchinoJ. B. (2024). Effects of lowering speed limits on crash severity in Seattle. J. Saf. Res.88, 174–178. 10.1016/j.jsr.2023.11.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  HussainQ.FengH.GrzebietaR.BrijsT.OlivierJ. (2019). The relationship between impact speed and the probability of pedestrian fatality during a vehicle-pedestrian crash: a systematic review and meta-analysis. Accid. Anal. Prev.129, 241–249. 10.1016/j.aap.2019.05.033

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  IslamM. (2023). An exploratory analysis of the effects of speed limits on pedestrian injury severities in vehicle-pedestrian crashes. J. Transp. Health28, 101561. 10.1016/J.JTH.2022.101561

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  IslamM. T.El-BasyounyK. (2015). Full Bayesian evaluation of the safety effects of reducing the posted speed limit in urban residential area. Accid. Anal. Prev.80, 18–25. 10.1016/j.aap.2015.02.026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  IslamM. T.El-BasyounyK.IbrahimS. E. (2014). The impact of lowered residential speed limits on vehicle speed behavior. Saf. Sci.62, 483–494. 10.1016/j.ssci.2013.10.006

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  ISTAT (2022). Road Accidents.

  • Google Scholar
  • View reference in article

• 24

  ITF (2025). Road Safety Annual Report 2025. Paris.

  • Google Scholar
  • View reference in article

• 25

  KayisuA. K.BahnasawiM.ElMikušováM.EgbineK.AlsisiM.KambaleW. V.et al (2024). Navigating chaos: a qualitative system Dynamics-based analysis of road safety challenges for vulnerable road users in a major developing city (Kinshasa). WSEAS Transactions on Environment and Development20, 1085–1097. 10.37394/232015.2024.20.99

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  KimJ. K.UlfarssonG. F.ShankarV. N.ManneringF. L. (2010). A note on modeling pedestrian-injury severity in motor-vehicle crashes with the mixed logit model. Accid. Anal. Prev.42 (6), 1751–1758. 10.1016/j.aap.2010.04.016

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  KongC.YangJ. (2010). Logistic regression analysis of pedestrian casualty risk in passenger vehicle collisions in China. Accid. Anal. Prev.42, 987–993. 10.1016/j.aap.2009.11.006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  KröyerH. R. G. (2015). Is 30km/h a “safe” speed? Injury severity of pedestrians struck by a vehicle and the relation to travel speed and age. IATSS Res.39, 42–50. 10.1016/j.iatssr.2014.08.001

  • CrossRef
  • Google Scholar
  • View reference in article

• 29

  LeafW. A.PreusserD. F. (1999). Literature Review on Vehicle Travel Speed and Pedestrian Injuries. Washington, DC: NHTSA.

  • Google Scholar
  • View reference in article

• 30

  LeflerD. E.GablerH. C. (2004). The fatality and injury risk of light truck impacts with pedestrians in the United States. Accid. Anal. Prev.36, 295–304. 10.1016/S0001-4575(03)00007-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  LevulytėL.BaranyaiD.SokolovskijE.TörökÁ. (2026). Pedestrians’ role in road accidents. 10.7708/ijtte.2017.7(3).04

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  LiD.RanjitkarP.ZhaoY.YiH.RashidiS. (2017). Analyzing pedestrian crash injury severity under different weather conditions. Traffic Inj. Prev.18, 427–430. 10.1080/15389588.2016.1207762

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  LiQ.WangZ.KollaR. D. T. N.LiM.YangR.LinP. S.et al (2023). Modeling effects of roadway lighting photometric criteria on nighttime pedestrian crashes on roadway segments. J. Saf. Res.86, 253–261. 10.1016/J.JSR.2023.07.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  MahmoudN.Abdel-AtyM.AbdelraoufA. (2023). The impact of target speed on pedestrian, bike, and speeding crash frequencies. Accid. Anal. Prev.192, 107263. 10.1016/j.aap.2023.107263

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  MikušováM. (2011). Value of networking in transport policy related to the road safety. Commun. Comput. Inf. Sci.239 CCIS, 70–77. 10.1007/978-3-642-24660-9_8/SAVE-RESEARCH

  • CrossRef
  • Google Scholar

• 36

  MikusovaM. (2016). Experiences with the implementation of measures and tools for road safety improvement. Univ. Politecnica Valencia1632–1638. 10.4995/cit2016.2016.2555

  • CrossRef
  • Google Scholar

• 37

  MikusovaM.KyamakyaK.GnapJ. (2024). Sustainable development through strategic road safety management: a regional approach. WIT Trans. Ecol. Environ.262, 379–390. 10.2495/SDP240321

  • CrossRef
  • Google Scholar

• 38

  MitchellR. J.BambachM. R.FosterK.CurtisK. (2015). Risk factors associated with the severity of injury outcome for paediatric road trauma. Injury46, 874–882. 10.1016/j.injury.2015.02.006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39

  MoradiA.AmeriP.RahmniK.NajafiM.JamshidiE.FakhriY.et al (2019). Factors affecting the severity of pedestrian traffic crashes. Arch. Trauma Res.8, 46. 10.4103/atr.atr_6_19

  • CrossRef
  • Google Scholar

• 40

  MuellerB. A.RivaraF. P.BergmanA. B. (1988). Urban-rural location and the risk of dying in a pedestrian-vehicle collision. J. Trauma - Inj. Infect. Crit. Care28, 91–94. 10.1097/00005373-198801000-00013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  OlszewskiP.SzagałaP.WolańskiM.ZielińskaA. (2015). Pedestrian fatality risk in accidents at unsignalized zebra crosswalks in Poland. Accid. Anal. Prev.84, 83–91. 10.1016/j.aap.2015.08.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  PasanenE. (1992). Driving Speeds and Pedestrian Safety: A Mathematical Model.

  • Google Scholar

• 43

  PazziniM.CameliL.VignaliV.SimoneA.LantieriC. (2024). Video-based analysis of a smart lighting warning system for pedestrian safety at crosswalks. Smart Cities7, 2925–2939. 10.3390/SMARTCITIES7050114

  • CrossRef
  • Google Scholar

• 44

  PittR.GuyerB.HsiehC. C.MalekM. (1990). The severity of pedestrian injuries in children: an analysis of the Pedestrian injury causation Study. Accid. Anal. Prev.22, 549–559. 10.1016/0001-4575(90)90027-I

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45

  PotoglouD.CarlucciF.CiràA.RestainoM. (2018). Factors associated with urban non-fatal road-accident severity. Int. J. Inj. Contr. Saf. Promot.25, 303–310. 10.1080/17457300.2018.1431945

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  PozzoniM.CeccarelliG.GorriniA.ManentiL.SanfilippoL., (2023). TomTom data applications for the assessment of tactical urbanism interventions: the case of Bologna. Sustainability15, 15. 10.3390/SU151712716

  • CrossRef
  • Google Scholar

• 47

  PulugurthaS. S.SambharaV. R. (2011). Pedestrian crash estimation models for signalized intersections. Accid. Anal. Prev.43, 439–446. 10.1016/j.aap.2010.09.014

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  RosénE.SanderU. (2009). Pedestrian fatality risk as a function of car impact speed. Accid. Anal. Prev.41, 536–542. 10.1016/j.aap.2009.02.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  RosénE.StigsonH.SanderU. (2011). Literature review of pedestrian fatality risk as a function of car impact speed. Accid. Anal. Prev.43, 25–33. 10.1016/j.aap.2010.04.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  RoudsariB. S.MockC. N.KaufmanR.GrossmanD.HenaryB. Y.CrandallJ. (2004). Pedestrian crashes: higher injury severity and mortality rate for light truck vehicles compared with passenger vehicles. Inj. Prev.10, 154–158. 10.1136/ip.2003.003814

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  SandersR. L.SchneiderR. J.ProulxF. R. (2022). Pedestrian fatalities in darkness: what do we know, and what can be done?Transp. Policy (Oxf).120, 23–39. 10.1016/j.tranpol.2022.02.010

  • CrossRef
  • Google Scholar

• 52

  SiddiquiN. A.ChuX.GuttenplanM. (2006). Crossing locations, light conditions, and pedestrian injury severity. Transp. Res. Rec. 1982, 141–149. 10.3141/1982-19

  • CrossRef
  • Google Scholar

• 53

  TayR.ChoiJ.KattanL.KhanA. (2011). A multinomial logit model of pedestrian-vehicle crash severity. Int. J. Sustain. Transp.5, 233–249. 10.1080/15568318.2010.497547

  • CrossRef
  • Google Scholar

• 54

  TefftB. C. (2013). Impact speed and a pedestrian’s risk of severe injury or death. Accid. Anal. Prev.50, 871–878. 10.1016/j.aap.2012.07.022

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  TorrisiV.CalabròG.OliveiraT. D. S.InturriG.CavalieriS.IgnaccoloM. (2026). Exploring the reliability of floating Car data (FCD) through penetration rate prediction in urban contexts. Res. Transp. Bus. and Manag.64, 101565. 10.1016/J.RTBM.2025.101565

  • CrossRef
  • Google Scholar

• 56

  TuluG. S.WashingtonS.HaqueM. M.KingM. J. (2015). Investigation of pedestrian crashes on two-way two-lane rural roads in Ethiopia. Accid. Anal. Prev.78, 118–126. 10.1016/j.aap.2015.02.011

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  UganJ.Abdel-AtyM.CaiQ. (2022). Estimating effectiveness of speed reduction measures for pedestrian crossing treatments using an empirically supported speed choice modeling framework. Transp. Res. Part F. Traffic Psychol. Behav.89, 276–288. 10.1016/J.TRF.2022.07.002

  • CrossRef
  • Google Scholar

• 58

  VicRoads (2016). Movement and Place in Victoria.

  • Google Scholar

• 59

  WHO (2023). Global Status Report on Road Safety 2023.

  • Google Scholar