Crashes Involving Electric Personal Mobility Devices (ePMD) Reported by the French Police: Types of Crashes, Users Involved, Injuries and Explanatory Factors

doi:10.21203/rs.3.rs-2801940/v1

The growth of mobility via Electric Personal Mobility Devices (ePMD) has been associated with an increased number of crashes. The French police were asked to report crashes resulting in injury or death in a post-crash survey. We have retrieved the databases of crashes reported by the French police in 2019, 2020 and 2021. This article reports on the contribution of the various categories of vehicles to the crashes recorded in France, and the parameters of ePMD crashes in France: types of crashes, periods, contexts and departments involved; age, gender, protective equipment, injuries of the users concerned by these ePMD crashes, etc. Compared to 2019, ePMD was the main category of users with an increasing number of crashes in 2020 and in 2021. Nearly three quarters of all ePMD crashes were a collision with a car, ePMD users crashed alone in 14% of cases, or hit a pedestrian (9.2%). ePMD users involved in crashes without helmets were younger on average than those wearing helmets. Males and females wore helmets with a similar frequency. When they collided with another user, ePMDs usually injured vulnerable road users such as pedestrians. When they collided with a vehicle not driven by a vulnerable user (protected by their vehicle body), ePMD users were more likely to be injured than the non-vulnerable user. The contexts of ePMD crashes, differences between crash types and user categories are discussed. Finally, results provide guidance for public policy and prevention campaigns (e.g., geographical areas that should be targeted).

Road traffic accidents

electric Personal Mobility Device

ePMD

police-reported data

types of ePMD crashes

France

ePMDs were the user category with the most dramatic increase in the number of crashes in 2020 and 2021
ePMD users involved in crashes mostly hit another road user at an intersection rather than crashing alone
ePMD users who struck another user were generally younger than those who crashed alone
ePMD users involved in crashes without helmets were younger than those with helmets
ePMD users were injured or killed primarily by colliding with another vehicle

Overview and objectives. The electric Personal Mobility Device (ePMD) is attracting attention worldwide as an increasingly popular mode of transport. This rise of ePMDs has led to an increase in ePMD crashes resulting in injuries and deaths, contributing to healthcare system costs, insurance costs, and revenue leakage for society (e.g., in New Zealand, Campbell et al., 2019; in the US, Lavoie-Gagne et al., 2021; Tischler et al., 2021). In general, road traffic accidents are an “economic burden”, especially for countries with low or moderate incomes (Zakeri et al., 2021). Systematic observations may help to prevent road traffic accidents and to design effective public policies: “statistics on road safety are essential for understanding the mechanisms and explanatory factors” of crashes, according to the French Minister of the Interior in April 2017 (Fekl, 2017, p. 1). However, in France, few detailed reports of ePMD crashes are available either at local, national or road level. In France, ePMDs are e-scooters, e-monowheels and gyropods, according to the Ministry of the Interior (Ministère de l’Intérieur, 2019). The ePMD category includes other small and speed-restricted motorized vehicles, with the exception of e-bikes and wheelchairs - respectively considered as cyclists and pedestrians[1], according to the French Office of Legal and Administrative Information (Direction de l’information légale et administrative, 2021, 2022). We explore here the database that is completed by French police after any intervention in a crash resulting in injury or death, in order to report on the characteristics of ePMD crashes. First, we describe the road traffic accidents reported in France, as regards user categories and the types of crashes in which they are involved, in order to report on the characteristics of ePMD crashes. Next, we report on the contexts of ePMD crashes, and the characteristics and injuries of the users involved. For example, we will see the contribution made to the number of road traffic accidents in France by ePMD crashes, the periods and hourly distribution of ePMD crashes, their spatial distribution, road contexts, weather conditions, age and gender of road users involved in ePMD crashes, their injuries and the use of helmets among ePMD drivers. We also present age and gender differences among ePMD users involved in crashes (e.g., according to the type of crash, helmet use or not); and differences in injuries, gender, and age among various user categories involved in ePMD crashes.

Context. The rise in e-scooter rental services between 2018 and 2019 expanded ePMD mobility in several major cities in the USA, Europe, Asia, Australia and New Zealand (Akter et al., 2021; Boglietti et al., 2021; Dias et al., 2021; Puzio et al., 2020; Schneeweiss et al., 2021; Zagorskas & Burinskienė, 2019). Today, the majority of ePMD users ride e-scooters, usually a rental one (e.g., in Paris, France, Christoforou et al., 2021; Zagorskas & Burinskienė, 2019). The introduction of a dockless e-scooter hire service in a large city usually leads to an increase in crashes involving ePMD users, and a rise in hospitalized ePMD users. For example, an increase in hospitalized e-scooter users following the introduction of e-scooter rental services in 2018 was reported in several major US cities (Alwani et al., 2020; Kobayashi et al., 2019; Namiri et al., 2020), in Paris, France (Hennocq et al., 2020); in Berlin, Germany (Uluk et al., 2021); and Auckland and Christchurch in New-Zealand (Brownson et al., 2019; Campbell et al., 2019; McGuinness et al., 2021). ). With the increase in hospitalized ePMD users, there has been an increase in people hospitalized after being struck by an e-scooter, particularly pedestrians (e.g., in the US, Lavoie-Gagne et al., 2021). In 2019, the “rapid influx of scooter-sharing services in U.S. cities has caught many municipal governments by surprise” (McKenzie, 2019, p. 27). In 2020, the city of Paris (France) asked the government to regulate e-scooter rental services and e-scooter traffic because their use on sidewalks - which has been banned in France since 2019 - was “a source of insecurity for pedestrians” (Ville de Paris, 2020).

In France, 10 ePMD users died in 2019, and 7 in 2020, according to the official report on road traffic accidents published each year by the French National Interministerial Observatory for Road Safety (ONISR, 2020, 2021). For comparison, Belgium had “only” 1 ePMD user fatality in 2019, and 0 in 2020; in the United Kingdom, 12 ePMD users died in 2019, and “only” 2 in 2020; South Korea had 12 ePMD fatality in 2019, and 17 in 2020, according to the Organisation for Economic Co-operation and Development (OECD, 2022). In France, 24 ePMD users died in 2021, accounting for 0.8% of road users killed in 2021 on the French roads (ONISR, 2022b, p. 8). In 2021, the main categories of users killed on French roads were drivers of light vehicles such as cars (48% of road deaths in 2021), motorcyclists (19.4%), pedestrians (14.1%), and cyclists (7.7%). The other categories of users killed on French roads in 2021 were users of commercial vehicles (3.5%), light moped drivers (3.3%), truck drivers (1.5%), etc.

In France, injuries following a crash involving ePMDs in urban areas increased by 40% between 2019 and 2020 (ONISR, 2021, p. 12). Contrary to ePMD users, other categories were generally involved in fewer crashes between 2020 and 2019: there was a decrease in road traffic accidents during COVID-19 lockdown periods. In 2021, the number of ePMD users injured or killed increased by 140%, compared to 2019 (ONISR, 2022b, p. 82).

In France in 2020, 781 ePMD users were injured or killed after a crash (ONISR, 2021). They were usually injured or killed by hitting a car (56% of ePMD users injured or killed in 2020). Other ePMD users crashed alone (17% of ePMD users injured or killed in 2020), hit a utility vehicle (10%), a motorcyclist (4.6%), a cyclist (4.1%), another ePMD user (1.9%), or a pedestrian (1.3%), etc. In addition to injuries among ePMD users, ePMD crashes resulted in 157 injuries to non-ePMD users, including 78 pedestrians, 34 cyclists, 27 motorcyclists, and 13 motorists. This order of importance was repeated in 2021: two-thirds of the ePMD users killed in 2021 collided with a vehicle (car, truck, commercial vehicle, bus), or they crashed alone in one third of the cases (ONISR, 2022b, p. 82). Other studies have revealed that a collision between an ePMD user and another vehicle was the most common cause for the hospitalization of an ePMD user. For example, J. Y. Kim et al. (2021) pointed out that, in two thirds of cases, ePMD users hospitalized in South Korea had collided with another vehicle. Specifically, some South Korean emergency services reported that hospitalized e-scooter users mostly crashed alone (Choi et al., 2022); while, in other places, they are mostly hospitalized after colliding a vehicle (e.g., ePMD users hospitalized for facial or dental injuries, M. Kim et al., 2021). Thus, the data are contradictory on this topic of the main type of crash leading to the hospitalization of ePMD users. For example, Brownson et al. (2019) revealed that the 180 e-scooter users hospitalized in Auckland (New Zealand) in 2019 typically reported falling off alone (91.7% of those hospitalized in Auckland in 2019 following an e-scooter crash), due to a malfunctioning e-scooter (3.3%), or a collision with a car (2.8%). Other studies support this finding of e-scooter users being hospitalized mostly after crashing alone: this was the case, for example, amongst the ePMD users hospitalized in 2018 in Indianapolis (Ind.) emergency departments (Alwani et al., 2020), while Southern California trauma centers reported that the majority of hospitalized ePMD users had collided with a vehicle during this period (Dhillon et al., 2020).

Lack of a helmet is one of the factors that increases the medical costs of an e-scooter crash (Kobayashi et al., 2019; Lavoie-Gagne et al., 2021). Consequently, between 66% and 97% of e-scooter users admitted to a U.S. emergency department in 2017 and 2020 were not wearing a helmet (Dhillon et al., 2020; Kobayashi et al., 2019; Lavoie-Gagne et al., 2021; Schneeweiss et al., 2021; Trivedi et al., 2019). Similarly, 98% of the 180 e-scooter users admitted to Auckland emergency rooms in New Zealand in 2019 were not wearing a helmet (Brownson et al., 2019). The same was true for 99% of the 248 e-scooter users admitted to a Berlin emergency room in Germany in 2019 (Uluk et al., 2021), 85% of the 108 e-scooter users admitted to an emergency room in South Korea between 2018 and 2021 (Choi et al., 2022), and 88% of the 123 e-scooter users hospitalized in two Paris (France) emergency rooms for facial or dental trauma in 2019 (Hennocq et al., 2020), etc. Regarding helmet use, of the 10 ePMD users killed in 2019 on French roads, “8 were not wearing helmets, like 71% of those who were injured” in 2019 (ONISR, 2020, p. 66). A few studies claim that the majority of e-scooter users wear a helmet, which is contradictory to most studies (Tian et al., 2022, p. 12).

In France to date, the characteristics of ePMD crashes have not been fully investigated. For example, medical research is sometimes based “only” on ePMD users hospitalized after a major crash in a city and/or on a sample of users hospitalized for a specific type of surgery (e.g., Hennocq et al., 2020; M. Kim et al., 2021). Regarding the official report on road traffic accidents in France, “only” a single page is dedicated to ePMD crashes (ONISR, 2020, 2021, 2022b). Most of these results are specific to a subsample of crashes in which the ePMD user was seriously injured or killed (e.g., with regard to vehicles involved in crashes with ePMD users during 2019, only the crashes that resulted in the hospitalization or death of an ePMD user, ONISR, 2020, p. 66). Regarding ePMD crashes, this article conducts a secondary analysis that expands on the official reports of road traffic accidents in France - based on the same data as these reports (ONISR, 2020, 2021, 2022b). For example, this article offers comparisons between the types of ePMD crashes and between the types of users involved in those crashes (vs. official road safety reports which do not provide comparisons).

[1] Unlike ePMDs, users of wheelchairs, rollerblades, skateboards or scooters (without a motor) are considered to be pedestrians and must travel on the sidewalk, without using the cycle lanes or the road. In France, ePMDs and cyclists (e-bikes or not) belong to two distinct categories but share the same mobility areas: they must use the cycle lane or - in the absence of a cycle lane - the road.

2.1 Survey completed by the French police

This study was based on the surveys completed by the French police after major crashes (“Traffic Accident Analysis Report” / “Bulletin d’Analyse des Accidents de la Circulation” or BAAC). Several police forces may complete this survey (e.g., municipal, national or military police). These surveys are used to populate the database of French road traffic accidents. After a crash resulting in injury or death, the French police completed a survey and answered several multiple-choice question (MCQ) about

the general situation (e.g., date, time, location, type of road, weather condition, general question about the type of crash);
the vehicles involved (e.g., type of vehicle and actions of these vehicles such as hitting a pedestrian, another vehicle or an animal);
the users involved (e.g., age, gender, injuries, driver / passenger or pedestrian, safety equipment).

In this study, “crashes” means “crashes recorded by the French police”. Crashes that did not result in injury, and collisions between pedestrians were not supposed to be reported by the French police. Nevertheless, a small fraction of ePMD police-reported crashes did not produce injuries.

2.2 Data collection and exploitation

In October 2022, we retrieved data from the survey that the French police had completed following road traffic accidents in France in 2019, 2020, and 2021 (including French overseas territories). The data was retrieved from the website dedicated to French government open data, from the French Ministry of the Interior (Ministère de l’Intérieur, 2022). Over the 3-year period considered (2019, 2020, 2021), the French traffic accidents database included 163,102 crashes. These crashes involved 279,091 vehicles and 367,425 users (drivers, passengers or pedestrians). This database contained missing values: details of the characteristics of the crashes, vehicles or users were not systematically provided by the police (e.g., the users had fled and the police reported the crash from a witness statement or video recording).

Because of the strong linear relationship between the number of inhabitants and the number of road traffic accidents in a geographical area, the number of crashes in this area is usually related to its population. This is the case for the reports of fatalities per million inhabitants within French departments (e.g., ONISR, 2022b, p. 40). French population census data was used to relate the number of crashes recorded in a department to the population of the department in 2019: statistics per million residents of the department were calculated using the 2019 census data. These census data were obtained from the National Institute of Statistics and Economic Studies for the population in metropolitan France (INSEE, 2022), or other official sources for the overseas territories[2] (e.g., www.isee.nc/population/recensement for New Caledonia in 2019). Geographical data were used for spatial analyses[3].

The R programming language (version 4) and the Rstudio software (R Core Team, 2021) were used for all the processes and analyses. Several R packages were used to import data, conduct analyses or produce tables, maps and figures[4]. All the results presented hereafter issued the dataset delivered by the French Ministry of the Interior: a computer code in R language calculated all the numbers, figures, tables, and some of the texts presented afterwards (e.g., lists of users, dates, lists of geographical areas).

2.3 Data analysis approach (assigned vehicle category and type of crash)

Vehicle category. This study considered 23 vehicle categories[5] (vs. the original database which had 32 categories). In France, two-wheeled motorized and three-wheeled motorized vehicles are categorized according to engine cylinder capacity, depending on whether they are (a) greater than 125cm³, (b) between 50cm³ and 125cm³, or (c) under 50cm³ (e.g., 3-wheeled motorized < = 50cm³). Other vehicles are classified by gross vehicle weight[6] - in metric tons, abbreviated hereafter as “T” (e.g., commercial vehicle over 1.5T and under 3.5T). In this study, vulnerable users were pedestrians, cyclists (riding an e-bike or a non-motorized bike), PMD (ePMD or non-motorized PMD), quads users and motorcyclists (two-wheelers and three-wheelers).

Type of crash. For each crash reported by the police, we have created a type of crash with the details of the vehicles. This definition of a crash type was not based on the general “type of crash” question. First, crashes that simultaneously involved a multi-vehicle crash and a collision with a pedestrian were identified (823 crashes over the 3-year period considered). Then, the remaining crashes were separated into pedestrians struck by a vehicle (25,744 crashes over the 3-year period considered); vehicle-to-vehicle crashes - with no pedestrians hit (100,818 crashes) and single-user crashes - without third party (35,717 crashes). Thus, this latter category of “single-user” crashes included some vehicles striking a parked vehicle - even though the parked vehicle would have suffered property damage (1,965 crashes reported for the 3-year period in question), and collisions with domestic or wild animals (560 crashes).

Inconsistencies between overall type of crash and details of vehicle actions. There were sometimes differences between the overall variables attributed to the crash and the details given about the vehicles. For example, the police recorded ‘non-collision’ crashes (in response to a general question about the crash) yet the details about the vehicles indicated that a vehicle had collided with a pedestrian (1,042 crashes between 2019 and 2021, or 3.9% of crashes involving a pedestrian who was hit during this period). Over the same 3-year period, the police identified 493 ‘non-collision’ crashes (in response to a general question about the crash) in which the details provided indicated that one vehicle had collided with another (0.5% of the crashes in which a vehicle hit another moving vehicle).

Lack of detail about vehicle actions. A few crashes did not have enough detail - in the “vehicles details” section of the survey - to categorize them. These crashes all involved multiple vehicles identified by the police, which did not provide details about the actions of these vehicles (e.g., no details about a vehicle hitting another user). For example, we deduced that 3,271 crashes were collisions between mobile vehicles, without the police explicitly reporting it (2% of all the 2019–2021 crashes). These 3,271 crashes without details about vehicles actions were generally reported by the police via the “general type of crash” MCQ:

as a vehicle-to-vehicle collision (2,425 crashes);
as an ‘other type of collision’ (432 crashes);
or without responding to this overal variable about the type of crash (265 crashes, or 0.16% of the 2019–2021 crashes).

This latter case of a non-explicitly reported collision between vehicles - either via an MCQ or by reporting that one vehicle hit another - concerned 6 crashes involving an ePMD user between 2019 and 2021 (0.2% of the ePMD crashes in France between 2019 and 2021).

2.4 Analysis

First, section 3.1 reports on the importance of the various user categories - including ePMDs - in French road traffic accidents, and the types of crashes they are involved in. Next, section 3.2 reports the types of ePMD crashes, the times of year and the French departments (county) involved in these crashes or contributing to an increase in ePMD crashes between 2019 and 2021. Spatial analysis, for example, allows us to specify the number of French departments that reported ePMD crashes, or an increase in ePMD crashes between 2019 and 2021. Section 3.3 reports the contexts of ePMD crashes: hours of day, mobility areas and road conditions, obstacles encountered by ePMD users in France, and mobility areas where pedestrians were struck by ePMD users - with police responses dedicated to describing the actions and location of pedestrians who were struck. Then, section 3.4 reports on the age, gender and protective equipment of ePMD users involved in crashes in France. Section 3.5 reports the age, gender and mode of transport of users who collided with ePMD users, as well as the mobility areas in which these users collided with ePMDs - with police responses dedicated to describing the mobility areas of the crashes. Then, we report on injuries to ePMD users or to users who collided with them (section 3.6).

The article also reports several comparisons (section 3.4, section 3.5 and section 3.6). We conducted ANOVAs, in order to compare:

the age of ePMD users involved in crashes by
- year (2019 vs. 2020 vs. 2021)
- types of crash (crash without a third party vs. pedestrian struck vs. collision between vehicles)
- whether they were wearing a helmet or not
severity of injuries for ePMD users involved in crashes (injury severity is expressed from 0 for “uninjured” to 3 for “killed”), by
- types of crash
- whether or not they were wearing a helmet
severity of injuries for users colliding with ePMD users, by type of user (vulnerable vs. other users)
the age and the severity of injuries for ePMD users involved in crashes with one type of user with the age and injury severity of that type of user, considering only those who crashed against an ePMD user.

We performed a Chi-square test, to compare the rate of male and female ePMD users injured with and without helmets. A correlation analysis was conducted on the 107 French departments, in order to identify the association between the number of ePMD users involved in crashes in a department and the number of users involved in crashes without wearing a helmet in that department.

[2] The population of the French overseas territories is counted every 5 years.

[3] The boundaries of the Paris districts and French departments were provided by the Database of Global Administrative Areas (GADM, https://gadm.org). The map of the city of Paris - in appendix B - are from OpenStreetMap (© OSM contributors), under Open Database License (ODbL), delivered by Stamen Design under a Creative Commons license (CC BY 3.0).

[4] Example of R packages that have been used are Allaire et al. (2019); Ben-Shachar et al. (2020); Bivand et al. (2015); Dowle et al. (2019); Gohel (2021); Kahle et al. (2019); Schauberger & Walker (2021); Wickham (2021); Wickham & Seidel (2020); Xie (2021b); Xie (2021a).

[5] The 6 categories of motorcycles and scooters are grouped into three categories (i.e., less than 50cm³, between 50 and 125 cm³, and greater than 125 cm³). The light and heavy quads are merged into one category, such as “autocar” [coaches] and “autobus” [buses]; trucks; and tractors (agricultural and road tractors, with or without trailers).

[6] Gross vehicle weight is “total of the weight of the vehicle (or combination of vehicles) including its load, […] the weight of the driver and the maximum number of persons permitted to be carried” (European-Commission, 2022).

The next section 3.1 provides an overall view of French police-reported types of crashes reported between 2019 and 2021, the main vehicle categories involved in these crashes and in an increasing number of crashes between 2019 and 2021. Then, the study focuses on ePMD crashes (section 3.2 to section 3.6).

3.1 Overview of road traffic accidents in France

Types of crashes. The 58,840 crashes reported by the French police in 2019 involved 100,710 vehicles, including 636 ePMDs (some crashes involved multiple vehicles). The 47,744 crashes reported in 2020 involved 81,066 vehicles, including 883 ePMD. The 56,518 crashes reported during 2021 involve 97,315 vehicles, including 1,510 ePMD.

The crashes reported in France between 2019 and 2021 can be divided into:

collisions between vehicles, without a pedestrian being struck (62% of the crashes reported between 2019 and 2021);
single-user crashes (22%);
a pedestrian being struck, without a collision between vehicles (16%);
a collision between vehicles during which a pedestrian was struck (0.5%).

User categories.The main vehicle category involved in crashes reported in 2019, 2020 or 2021 - combined - was light vehicles (e.g., cars, vans under 1.5T). Approximately 76% of the crashes reported by the French police involve one or more light vehicles (and possibly other vehicle categories).

Table 1 reveals that 7 of the 23 vehicle categories identified in this study were involved in an increase in the number of crashes between 2019 and 2020: ePMDs, other vehicles, e-bikes, quads, non-motorized PMDs, 3-wheeled motorized < = 50cm3, not specified. In 2021, 12 categories were involved in more crashes, compared to 2019 (motorcycles < 50cm3, commercial vehicles over 1.5T and under 3.5T, bicycles, ePMDs, 3-wheeled motorized > 125 cm3, other vehicles, e-bikes, quads, non-motorized PMDs, trams, trains, not specified). Thus, ‘only’ 6 categories of users were involved in more crashes in 2020 and 2021, than in 2019 (ePMDs, other vehicles, e-bikes, quads, non-motorized PMDs, not specified).

Of the 23 vehicle categories considered in this study, ePMD users were the 8th highest vehicle-user category involved in crashes resulting in injury and/or death in France between 2019 and 2021. In the user categories involved in more than 100 crashes over the 3-year period, ePMD was the category with the highest increase in crashes reported in 2020 and in 2021, compared to crashes reported in 2019.

In addition to the increase in ePMD crashes in 2020 and 2021, the French police reported an increase in e-bike crashes in 2020, compared to 2019 (+ 29.5%), and a major increase in 2021 (+ 109.2% of e-bike crashes in 2021, compared to 2019). Then, non-motorized cyclists are involved in almost as many crashes in 2019 and 2020 (-0.7% of crashes during 2020, compared to 2019); but they were involved in more crashes in 2021, compared to 2019 (+ 10.5%). Compared to 2019, other user categories recorded an increase in the number of crashes reported in 2020, but crashes did not increase in 2021 (e.g., three-wheelers under 50cm³). Conversely, some categories were involved in fewer crashes in 2020, but there was a significant increase in 2021 (e.g., three-wheeled motorized vehicles over 125cm³). Table A.1 (in Appendix A) provides detailed statistics for 2019, 2020 and 2021.

3.2 Overall description of ePMD crashes

The next section 3.2 reports the types of ePMD crashes, the times of year and the departments involved in these crashes.

3.2.1 ePMD crash types

Table 1 reveals that ePMD crashes:

were mostly collisions between a vehicle and an ePMD user, without a pedestrian being struck (76% of ePMD crashes between 2019 and 2021);
involved only the ePMD user and no pedestrian or another moving vehicle (14%);
involved a pedestrian struck by an ePMD - without another vehicle being involved (9.2%);
involved a collision between vehicles following which a pedestrian was hit (1%).

In 2019, ePMD users were involved in 22 “vehicle-to-vehicle crashes in which a pedestrian is struck” and they hit 70 pedestrians. These 92 crashes that simultaneously involved an ePMD user and a struck pedestrian included 68 cases of pedestrians struck by ePMD users that were explicitly reported by the police under “vehicle action details”. In 2020, ePMD users were involved in 3 “vehicle-to-vehicle crashes in which a pedestrian is struck” and they hit 80 pedestrians. For each of these 83 crashes, the police systematically reported that it was the ePMD users who hit a pedestrian. In 2021, ePMD users were involved in 6 “vehicle-to-vehicle crashes in which a pedestrian is struck” and they hit 126 pedestrians, including 125 cases where the police explicitly reported that it was the ePMD user who hit a pedestrian, under “vehicle action details”.

In 2020, the increase in ePMD crashes affected almost all crash types: ePMD users were involved in more crashes with another vehicle - without a pedestrian being hit (+ 224 crashes in 2020 compared to 2019), in more single-user crashes without a third party (+ 32), and they hit more pedestrians in 2020 than in 2019 (+ 10). However, they were involved in fewer crashes involving multiple vehicles - including an ePMD - and a struck pedestrian (-19 crashes). Similarly, when comparing the year 2019 with the year 2021, collisions between an ePMD and another vehicle increased (+ 728 crashes in 2021), as did the number of pedestrians struck by ePMDs (+ 56 crashes), and the number of ePMD users who crash alone (+ 96). As for 2020, the number of crashes that involved multiple vehicles - including an ePMD - and a struck pedestrian decreased in 2021, compared to 2019 (-16 crashes).

3.2.2 Scale of ePMD crashes, by week

In France, 632 of the 58,840 police-reported crashes in 2019 involved an ePMD (1.1% of all 2019 crashes). In 2020, 879 of the 47,744 crashes involved an ePMD (1.8% of the crashes of 2020) and 1,496 of the 56,518 crashes in 2021 (2.6% of the crashes in 2021). Figure 1 shows the proportion of crashes involving an ePMD, in the total road traffic accidents per week from 2019 to 2021.

At most, crashes involving an ePMD accounted for 4% of the crashes reported in any week from 2019 to 2021 (e.g., the week of 30/08/2021). In France, only one of the weeks in the 3-years period considered did not involve a police-recorded ePMD crash, at the end of the first COVID-19 lockdown period (the week of 27/04/2020).

3.2.3 Spatial distribution of ePMD crashes between 2019 and 2021

Spread of ePMD crashes in France. There are 107 departments in France. Between 2019 and 2021, the police recorded ePMD crashes in 94 French departments (88% of the French departments reported ePMD crashes during 2019–2021). In 2019, 55 departments reported no ePMD crashes. In 2020, 35 departments reported no ePMD crashes. In 2021, “only” 21 departments reported no ePMD crashes.

On average during the 3-year period in question, the police reported 28 ePMD crashes in each French department - including those with no ePMD crashes (SD = 105 crashes). By relating the number of crashes involving an ePMD to the population census of each department surveyed in 2019, ePMD users were involved on average in 26 crashes per million inhabitants of a department, over 3 years (SD = 53 crashes per million inhabitants of a department between 2019 and 2021). In 2019, each French department reported an average of 4.1 ePMD crashes per million inhabitants of the department (SD = 13). In 2020, departments reported an average of 6.6 ePMD crashes per million inhabitants (SD = 18). In 2021, French departments reported an average of 16 ePMD crashes per million inhabitants (SD = 27).

Between 2019 and 2020, 52 departments reported an increase in the number of ePMD crashes, including 24 departments that reported no ePMD crashes in 2019. On the other hand, ePMD crashes decreased in 2020 in 17 departments, including 6 departments with no ePMD crashes in 2020 (but with some in 2019). Comparing 2019 and 2021, ePMD crashes increased in 80 departments, including 37 departments that reported no ePMD crashes in 2019 and reported some ePMD crashes in 2021. Of these 37 departments, 15 reported no ePMD crashes during 2020 and 2019. Compared to 2020, ePMD crashes increased in 75 departments in 2021, including 41 departments that reported an increase in crashes for the second consecutive year (number of ePMD crashes in 2019 < number of ePMD crashes reported in 2020 < number of ePMD crashes reported in 2021).

Remarkable departments, regarding ePMD crashes. In 11 French departments, ePMDs represented more than 2% of the crashes recorded in the department[7] between 2019 and 2021, and ePMD crashes represented between 1% and 2% of crashes that occurred in 22 departments. The police reported ePMD crashes between 2019 and 2021 in 56 other departments, yet these ePMD crashes did not represent more than 1% of the crashes that occurred in those departments. In addition, 7 other departments had no ePMD crashes in 2019, 2020 and 2021. Table A.2 (in Appendix A) provides statistics for each French department.

Over a 3-year period (2019 and 2021), the 12 departments in which police reported the most ePMD crashes per million inhabitants were Paris (465 crashes involving an ePMD user per million inhabitants), Rhône (175.3), Saint-Pierre-and-Miquelon (167.1), Hauts-de-Seine (141.4), Val-de-Marne (126.7), Marne (82.8), Seine-Saint-Denis (80.2), Pyrénées-Orientales (67.2), French Guiana (65.2), Somme (57.8), Indre-et-Loire (54.3) and Meurthe-et-Moselle (49).

The French departments with the most ePMD crashes were Paris (34% of the ePMD crashes reported in France), Rhône (10.8%), Hauts-de-Seine (7.6%), Val-de-Marne (5.9%), Seine-Saint-Denis (4.4%), Bouches-du-Rhône (2%), Gironde (1.9%), Nord (1.6%), Marne (1.6%) and Maine-et-Loire (1.3%). The 8 departments of the Paris region accounted for 54% of the ePMD crashes reported in France. For comparison, the Paris region has 19% of the population of France, and it accounted for 32% of all the road traffic accidents reported by the police between 2019 and 2021 (regardless of whether they involved an ePMD or other categories).

Four of the 8 departments of the Paris region accounted for 96% of ePMD crashes reported in the Paris region (i.e., Paris, Hauts-de-Seine, Val-de-Marne, Seine-Saint-Denis). These 4 departments accounted for 38% of the inhabitants of the Paris region in 2021, and reported 78% of all the crashes reported by the French police in the Paris region between 2019 and 2021. These results reveal that ePMD crashes occurred particularly in Paris and in the surrounding departments. Table B.1 and Figure B.1 (in Appendix B) show the police-recorded ePMD crashes in each of the 20 Paris districts. For example, Table B.1 shows that some of the Parisian districts were not affected by an increase in ePMD crashes in 2020 (i.e., the 6th, 5th, 9th and 20th districts) or in 2021, compared to 2019 (i.e., the 5th, 6th and 2nd districts). One district of Paris reported no pedestrians hit by an ePMD in 2019, 2020 or 2021 (i.e., the 2nd district).

3.3 Contexts of ePMD crashes in France (2019–2021)

The next section 3.3 reports on time distribution, lane types and road conditions of ePMD crashes, details of obstacles leading to falls, and the actions of pedestrians who were hit.

3.3.1 Hourly distribution of ePMD crashes

Regarding the time of day when these crashes occurred, most of the ePMD crashes occurred between 8 a.m. and 8 p.m.: 74% of the ePMD crashes reported during the 3-year period occurred during this time period. Figure 3 shows the time distribution of road traffic accidents in France.

In France, 20% of the ePMD crashes were reported between 8 a.m. and 12 p.m., 22% were reported between 12 p.m. and 4 p.m., and 33% between 4 p.m. and 8 p.m. The remaining ePMD crashes were divided between those occurring between 8 p.m. and midnight (13% of the ePMD crashes) or between midnight and 8 a.m. (13%). On average, ePMD crash occurred at 14.07 p/m. (SD = 5:29’). Interestingly, other crashes - not involving an ePMD - occurred within a comparable time frame (M = 13.57’, SD = 5:26’).

3.3.2 Mobility areas and traffic conditions

Mobility areas involved. In France, 98% of ePMD crashes recorded by the police occurred in urban areas. These ePMD crashes usually took place on the roadway (72% of ePMD crashes identified between 2019 and 2021). To a lesser extent, ePMD crashes took place on a cycle path (16% of ePMD crashes), a sidewalk (5.8%), a roadside shoulder (0.5%), or another type of road - listed under “other special track” or “other” categories (5.2%). No ePMD crashes recorded in 2019, 2020 or 2021 occurred in an emergency lane. Regarding the slope of the road, ePMD crashes generally occurred on a flat road (88% of ePMD crashes), a slope (11%), or even on hilltops or downhill areas (1.7%).

Most of ePMD crashes occurred at an intersection (59% of ePMD crashes). These ePMD crashes that occurred at intersections typically occurred at a crossroads intersection (23% of ePMD crashes) or a “T-junction” (16%). The remaining ePMD crashes that occurred at an intersection are divided into those that occurred at a roundabout (6.8% of ePMD crashes); a public square (3.4%); a “Y intersection” (3.3%); an intersection with more than 4 exits (1.5%); a grade crossing (0.4%) or other crossings - reported as “other” (3.9%).

Traffic conditions. Regarding the time of day, ePMD crashes generally occurred in daylight, according to the answers of the police to a MCQ (68% of ePMD crashes). At night, ePMD crashes usually occurred when street lights were turned on (23%), were non-existent (1.3%) or turned off (0.5%). Other ePMD crashes occurred at dusk or dawn (6.5% of ePMD crashes).

Regarding weather conditions, ePMD crashes generally occurred in conditions that police reported as “normal” (83% of ePMD crashes). The remaining ePMD crashes occurred during light rain (9.3%), overcast weather conditions (4.7%) or heavy rain (1.4%). Other weather conditions accounted for 1.9% of ePMD crashes[8]. In addition to these meteorological explanations, the police generally stated that the surface on which an ePMD crash took place was “normal” (84% of ePMD crashes), or wet (15% of ePMD crashes). The other road conditions accounted for 0.7% of the ePMD crashes[9].

3.3.3 Obstacles encountered by ePMD users who crash alone and actions of struck pedestrians

Single-user crashes (14% of ePMD crashes). In France, the police reported that 36% of ePMD users who crashed alone had hit an obstacle. In addition to these obstacles hit by ePMD users, crashes involving only the ePMD user could occur in bad weather conditions (13% of crashes involving only an ePMD user, without third parties). Regarding obstacles, ePMD users who crashed alone hit:

a curb (14% of ePMD users who crash alone);
an obstacle on the roadway (5.8%);
a parked vehicle (5.1%);
a pole (4.1%);
an obstacle on the sidewalk (3.1%);
an item of street furniture or a parapet, a tree, a wall, a barrier along a road (3.1%);
or other forms of obstacles such as a ditch, an embankment, or even a run-off road crash without any particular obstacle (1.2%).

Pedestrians struck (9.2% of ePMD crashes). Table 2 shows the circumstances in which pedestrians were struck by ePMD users in France.

Table 2: Circumstances in which pedestrians were struck by ePMD users in 2019, 2020 and 2021

Pedestrians struck by ePMD users can be separated into those struck at a crosswalk, roadway, or sidewalk. In the majority of cases, pedestrians were struck by ePMDs while crossing a road, cycle path or, more rarely, a sidewalk (63% of pedestrians struck by ePMD users between 2019 and 2021).

3.4 Characteristics of ePMD users involved in crashes in France

The next section 3.4 reports on the age and gender of ePMD users involved in crashes, their protective equipment, and comparisons of the age of ePMD users involved in crashes according to (a) the types of crashes they were involved in, (b) the year, (c) whether they were wearing a helmet. Notable differences are summarized at the end of the section. The police do not systematically provide the characteristics of the users involved in a crash. The statistics presented in this section are mainly calculated based on the crashes for which the police provided the characteristics of the users[10] (i.e., ages, gender, protective equipment, or injuries).

3.4.1 Age and gender of ePMD users involved in a crash in France

Age and gender of ePMD users involved in a crash. In France, the average age of ePMD users involved in a crash was 30 years (SD = 13), and 30% of them were female.

In France, 76% of ePMD users involved in crashes were under 40 years old. The ePMD users involved in crashes who were under 40 years old are divided between those under 18 (13% of the ePMD users involved in a crash), between 18 and 25 (26%), between 25 and 30 (16%), between 30 and 35 (12%) and between 35 and 40 (9.4%). The other ePMD users involved in crashes were between 40 and 45 years old (7.1%), between 45 and 50 (5.4%), between 50 and 60 (7%), between 60 and 70 (2.5%) or over 70 (0.57%). According to the surveys completed by the French police, the six oldest ePMD users involved in crashes between 2019 and 2021 were 118, 89, 85, 82, 81 and 78 years old. Thus, the age reported by the police suggests that ePMD users involved in a crash could be over 100 years old - which may be a typing mistake.

In 2019, ePMD users involved in crashes were on average 31 years old (SD = 13); in 2020, they were on average 32 years old (SD = 13) and 29 years old in 2021 (SD = 13). Thus, ePMD users involved in crashes in 2019 were approximately the same age as those involved in crashes in 2020, F(1, 1,542) = 3.4, p > .05. However, ePMD users involved in a crash in 2019 were slightly younger than those who crashed in 2021, F(1, 2,221) = 4.1, p = 0.043, d = 0.09.

3.4.2 Safety equipment and helmet use by ePMD users involved in crashes

In nearly two-thirds of ePMD crashes, police reported that the ePMD users were not wearing any safety equipment (68% of the ePMD crashes reported between 2019 and 2021). When they had safety equipment, it was usually a helmet (29% of ePMD crashes). Other safety equipment was much less common, whether gloves (1.3%), reflective jackets (1.1%), or some other equipment - reported as “other” (0.2%). Thus, ePMD users involved in a crash were wearing a helmet in nearly one out of three crashes. The departments with a high number of ePMD crashes were those with a high number of ePMD users who had crashed without helmets, with two exceptions, r(105) = 0.97; p < .001.

An ANOVA[11] revealed that ePMD users involved in crashes wearing helmets were generally older than those without helmets (respectively M = 34 years, SD = 14 years; M = 29 years, SD = 13 years), F(1, 3,111) = 86, p < .001, d = 0.37.

In France, 28% of male ePMD users involved in crashes between 2019 and 2021 wore a helmet, and 30% of female ePMD users wore a helmet. Thus, a Chi-square[12] test revealed that male and female ePMD users who had crashed in France between 2019 and 2021 had the same tendency to wear a helmet, χ² (1) = 0.04, p > .05.

3.4.3 Comparison of the age of ePMD users involved in a crash in France, by type of crash

Between 2019 and 2021, ePMD users who hit occupied vehicles - without a pedestrian being struck - were on average 31 years old (SD = 13). Over the same time period, ePMD users who hit a pedestrian - without a vehicle being hit - were on average 30 years old (SD = 13). Lastly, ePMD users who were involved in a crash without third parties were on average 34 years old (SD = 14).

A series of ANOVAs[13] revealed that ePMD users who crashed alone were older on average than those who hit a pedestrian, F(1, 676) = 15, p < .001, d = 0.31. Similarly, ePMD users who crashed alone were older on average than those who hit a vehicle, F(1, 2,851) = 30, p < .001, d = 0.29. On average, the age of ePMD users who hit a pedestrian was comparable to the age of those who hit a vehicle, F(1, 2,651) = 0.39, p > .05, d = 0.04. Overall, there was a difference in the age of ePMD users who crashed alone, hit a pedestrian, or hit another vehicle, F(3, 3,109) = 11, p < .001. A Tukey post hoc test (HSD) at the threshold p < .05 - given the series of three comparisons performed - confirmed these age differences according to the type of crash: although ePMD users who hit a vehicle were roughly the same age as those who hit a pedestrian (mean difference of 0.55 years, p > .05); ePMD users who crashed alone were older than those who hit pedestrians (average difference of 4.3 years, p < .001) or those who hit another vehicle (average difference of 3.8 years, p < .001).

Summary of differences (section 3.4).

ePMD users who crashed in 2021 were slightly older than users who crashed in 2019.
ePMD users who crashed wearing helmets were generally older than those who did not wear helmets.
Male and female ePMD users involved in crashes were equally likely to wear helmets.
ePMD users who crashed alone were older than those who hit a vehicle or those who hit a pedestrian.

3.5 Characteristics of users involved in collisions with ePMD users

The next section 3.5 reports on the categories of users involved in crashes with an ePMD, the mobility areas in which these categories of users collided with an ePMD user, the ages and genders of these users involved in a crash with an ePMD user, and comparisons between the characteristics of these users and those of ePMD users.

3.5.1 Types of users and mobility areas involved

Light vehicles users (e.g., cars, vans with a gross vehicle weigh under 1.5T) were the main user category involved in crashes with ePMD users (1,646 crashes between 2019 and 2021, or 55% of the ePMD crashes during this period). Table 3shows the vehicles most frequently involved in crashes with ePMD users, which does not necessarily imply that these users hit or were hit by an ePMD user in the case of crashes that involved more than two vehicles.

Table 3

Types of users involved in a crash with an ePMD user between 2019 and 2021 and mobility areas involved
Type of user	Crashes involving an ePMD user (and % of the ePMD crashes for this year)			Changes		Mobility area (% of all crashes between this category of users and an ePMD user between 2019 and 2021)
Type of user	2019	2020	2021	2019 vs. 2020	2019 vs. 2021	Cycle lane	Sidewalk	Roadway	Roadside shoulder	Other
Light vehicles	329 (52.1%)	453 (51.5%)	864 (57.8%)	+ 37.7%	+ 162.6%	11.1%	3%	81.6%	0.3%	3.7%
Pedestrian	92 (14.6%)	83 (9.4%)	132 (8.8%)	-9.8%	+ 43.5%	23.8%	17.9%^**	51.1%	0.7%	5.9%
Commercial vehicle over 1.5T and under 3.5T	45 (7.1%)	81 (9.2%)	111 (7.4%)	+ 80%	+ 146.7%	19%	3%	72.6%	0.8%	4.6%
Bicycle	25 (4%)	53 (6%)	51 (3.4%)	+ 112%	+ 104%	62%	3.1%	29.5%	-	5.4%
Motorcycle < 50cm3	13 (2.1%)	21 (2.4%)	28 (1.9%)	+ 61.5%	+ 115.4%	14.5%	1.6%	75.8%	-	8.1%
Motorcycle > 50cm3 and < = 125cm3	11 (1.7%)	17 (1.9%)	24 (1.6%)	+ 54.5%	+ 118.2%	7.7%	1.9%	86.5%	-	3.8%
Bus	12 (1.9%)	12 (1.4%)	23 (1.5%)	=	+ 91.7%	2.1%	-	70.2%	-	27.7%
Motorcycle > 125cm3	14 (2.2%)	8 (0.9%)	24 (1.6%)	-42.9%	+ 71.4%	10.9%	-	84.8%	-	4.3%
Truck	5 (0.8%)	6 (0.7%)	17 (1.1%)	+ 20%	+ 240%	7.1%	-	78.6%	-	10.7%
Voiturette	3 (0.5%)	6 (0.7%)	9 (0.6%)	+ 100%	+ 200%	11.1%	5.6%	83.3%	-	-
Tram	1 (0.2%)	3 (0.3%)	8 (0.5%)	+ 200%	+ 700%	-	-	41.7%	-	58.3%
3-wheeled motorized > 125 cm3	3 (0.5%)	3 (0.3%)	5 (0.3%)	=	+ 66.7%	9.1%	-	90.9%	-	-
Other vehicle	2 (0.3%)	3 (0.3%)	3 (0.2%)	+ 50%	+ 50%	-	-	100%	-	-
Indeterminable	3 (0.5%)	4 (0.5%)	-	+ 33.3%	-100%	-	-	100%	-	-
^** In the general question about the mobility area of a crash, the police underestimated the number of pedestrians hit by ePMD users on the sidewalk, compared to the answers to the specific question about pedestrian actions - only for cases of pedestrians hit - with 25.7% of the pedestrians struck by ePMD that are on a sidewalk
Only the categories involved in more than 5 crashes with an ePMD user between 2019 and 2021 are presented in this table. Some of these users were not hit by ePMD users, but were still involved in crashes involving ePMDs (e.g., an ePMD hit a vehicle which in turn hit a pedestrian). For example, among the 307 pedestrians that are presented in the table, 276 were actually struck by ePMD users.

Crashes between a road user and an ePMD user generally involved light vehicles. To a lesser extent, ePMD users collided with pedestrians, commercial vehicles, bicycles and motorcycles with an engine under 50 cm³. We have seen previously that collisions between an ePMD and another user increased in 2020 and 2021, compared to 2019. However, some exceptions of user categories were involved in an equivalent number of crashes with an ePMD in 2019 and 2020 (i.e., buses; 3-wheeled motorized > 125 cm3). Some other user categories were involved in fewer crashes with an ePMD in 2020 than in 2019 (i.e., pedestrians, motorcycles > 125cm3). Considering only crashes in which an ePMD user struck a pedestrian (not considering crashes that involved both multiple vehicles and a pedestrian struck by one of those vehicles), section 3.2.1 reports that ePMD users struck more pedestrians in 2020 than in 2019. Furthermore, all user categories were involved in more ePMD crashes in 2021 compared to 2019, with the exception of the category “indeterminable”.

More than 50% of crashes between an ePMD user and a cyclist occurred on a cycle path or lane. Most other categories of users were involved in an ePMD crash on the road in more than 50% of the cases, except for a few categories of users who hit ePMDs on the road in less than 50% of ePMD crashes (i.e., trams, other vehicles, indeterminable, non-motorized PMDs, tractors, e-bikes, 3-wheeled motorized < = 50cm3, special vehicles, trains). During the first two years of the period considered (2019 and 2020), ‘only’ 3 categories of users were hit by an ePMD user on sidewalks in 2019 or 2020 (i.e., pedestrians and, to a lesser extent, cyclists and motorists). During the entire 3-year period considered (2019, 2020 and 2021), 8 categories of users were hit by an ePMD user on sidewalks.

3.5.2 Age and gender of users hit by ePMD users, or who hit ePMD users

Light vehicle drivers who collided with ePMD users were male in 32% of cases. The average age of light vehicle drivers who collided with ePMDs was 44 years old (SD = 16). Thus, ePMD users who were involved in crashes were usually male, hitting a motorist in nearly one out of every two ePMD crashes. In two thirds of cases, motorists who crashed with an ePMD user were also a man (see Table 4). When an ePMD user hit another vehicle, this latter vehicle was usually driven by a male. For example, the average age of cyclists hit by ePMD users was 37 years (SD = 15) and 26% of those cyclists hit by an ePMD user were female. However, this did not apply to pedestrians: pedestrians struck by ePMD users were on average 51 years old (SD = 24), and 63% of these collisions concerned female pedestrians - struck by an ePMD user.

Table 4

Vehicle categories involved in a crash with an ePMD between 2019 and 2021, genders and ages (M, SD, and comparison with the average age of ePMD users involved in these user-to-user crashes)
Type of crashes involving ePMD users (first four rows) and users hit by an ePMD	Crashes between 2019 and 2021^A	Gender (% of ePMD users who crashed under these conditions [first four rows], or gender of users struck by an ePMD user)		Age of users colliding with an ePMD		Age of ePMD users involved in this type of crash (first four rows) or colliding with this user category
	Crashes between 2019 and 2021^A	Female	Male	M	SD	M	SD
Collision between vehicles	2,275	28.0	72.0			30.7	12.8
Only the ePMD user (no third parties)	412	33.7	66.3			34.5	14.2
Struck pedestrian	255	20.4	70.2			29.8	12.6
Collision between vehicles and struck pedestrian	21	19.0	85.7			31.4	16.1
Light vehicles^***	1,540	30.1	64.9	43.9	16.4	30.1	12.8
Pedestrian^***	270	63.0	37.4	51.3	24.3	29.8	12.6
Commercial vehicle over 1.5T and under 3.5T^***	219	9.1	87.7	40.8	13.8	32.0	12.9
Bicycle	122	26.2	73.8	37.1	14.8	34.1	12.3
Motorcycle < 50cm3	54	16.7	68.5	29.7	15.1	30.9	12.8
Motorcycle > 50cm3 and < = 125cm3^*	50	10.0	88.0	37.9	12.7	31.8	13.7
Motorcycle > 125cm3⁺	46	4.3	97.8	38.3	13.8	33.0	12.7
Bus^***	45	17.8	84.4	43.2	13.4	31.3	11.0
Truck^**	26	3.8	96.2	45.5	11.1	32.9	14.5
Voiturette	17	23.5	76.5	36.7	15.5	34.9	10.8
Tram^**	12	16.7	83.3	42.4	10.7	24.7	12.6
3-wheeled motorized > 125 cm3^***	11	9.1	90.9	46.0	13.4	26.6	7.7
Other vehicle^**	6	16.7	83.3	42.2	12.0	22.4	8.5
Non-motorized PMD	3		100.0	32.7	22.1	26.0	4.0
Tractor⁺	3	33.3	66.7	46.3	11.4	26.0	7.2
E-bike	3	33.3	66.7	27.7	7.0	33.3	17.1
^A Crash for which the police provided information on the characteristics of the users involved (most of the ePMD crashes)
Comparison between the average age of users in this category hit by ePMD users and the average age of ePMD users involved in these crashes: ^*p < .001; ^p < .01; ^* p < .05; ⁺ p < .06
Only categories involved in more than 2 crashes with an ePMD user between 2019 or 2021 are shown. Crashes for which police did not provide user characteristics are not represented. Vehicle passengers are not included: these statistics are only for pedestrians, drivers of ePMDs and drivers of vehicles involved in crashes with ePMD users.

The comparisons presented in Table 4 focus on the ages of the ePMD users involved in crashes with a road user, only for the crashes for which the police recorded the characteristics of the users involved. These comparisons do not include user categories involved in a single crash with an ePMD user between 2019 and 2021 (i.e., 3-wheeled motorized < = 50cm3, special vehicle, train). These comparisons reveal that ePMD users who collided with other users were generally younger than those other users, with some exceptions of user categories involved in collisions with ePMD users who were approximately the same age, on average (i.e., bicycles, motorcycles < 50cm3, non-motorized PMDs, motorcycles > 125cm3, tractors, e-bikes, voiturettes).

3.6 Injuries and deaths caused by ePMD crashes

The next section 3.6 reports injuries of ePMD users and injuries of other users who collided with an ePMD user, comparisons of ePMD users’ injuries (by crash type, by helmet use or not, by type of user struck); and comparisons between the injuries of users colliding with ePMD users, depending on their mode of travel and whether or not they were vulnerable users.

Over the 3-year period considered, 16% of ePMD crashes reported by the French police resulted in hospitalization or death. In the majority of cases, the users hospitalized or killed following an ePMD crash were the ePMD users themselves (15% of ePMD crashes resulted in the hospitalization or death of the ePMD user). None of these crashes in which an ePMD user was seriously injured or killed led to the hospitalization or death of a user other than the ePMD user. To a lesser extent, other road users who were hospitalized or killed were pedestrians (0.96% of the ePMD crashes), cyclists (0.23%), drivers of motorcycles with an engine capacity over 125cm³ (0.17%) or between 50 and 125cm³ (0.17%), drivers of light vehicles (0.1%), etc.

3.6.1 Injuries of ePMD users

Type of crash resulting in the death or hospitalization of an ePMD user. The primary cause of hospitalization or death for ePMD users was a collision with another vehicle (75% of crashes leading to the hospitalization or death of an ePMD user). Other crashes leading to the hospitalization or death of an ePMD user were generally ‘single-user crashes’ (24% of crashes leading to the hospitalization or death of an ePMD user). Some ePMD users were hospitalized or killed after colliding with a pedestrian (0.67% of crashes leading to the hospitalization or death of an ePMD user) or as part of a multi-vehicle/pedestrian crash (0.45%). We will then see that the “most frequent” crashes were not necessarily the most dangerous for the ePMD user.

Injuries of ePMD users, by type of crash. Considering only police-recorded crashes, a crash involving only an ePMD user generally resulted in minor injuries to that user (72% of single-user ePMD crashes), or the ePMD user was hospitalized (22%), or killed (3.6%). The police recorded a few crashes that did not involve a user other than the ePMD user, in which the ePMD user remained unharmed - this was not supposed to be recorded by the police (2.4% of crashes involving only a single ePMD user). Similarly, following a crash between an ePMD user and another vehicle, the ePMD user was slightly injured (80% of crashes between an ePMD user and another vehicle), or the ePMD user was hospitalized (14%), or even died (1.2%). Following a crash between an ePMD user and another vehicle, the ePMD user was rarely unharmed (5.9% of crashes between an ePMD user and another vehicle).

Between 2019 and 2021, ePMD users who hit pedestrians were generally uninjured (83% of crashes in which an ePMD user struck a pedestrian). In the latter case, it was the pedestrian hit by an ePMD who was injured, since the police only report - in theory - crashes that result in injuries. After a collision with a pedestrian, some of the ePMD users were slightly injured (16% of the crashes between an ePMD user and a pedestrian), or hospitalized (1.2%). In France, no ePMD user died after hitting a pedestrian in 2019, 2020 or 2021. Except for pedestrians hit by ePMD users, ePMD users were injured in most other police-reported cases (i.e., an ePMD user crashed alone, hit a vehicle, or was involved in a vehicle-to-vehicle crash that also involved hitting a pedestrian).

Types of ePMD crashes and injury severity. Injury statistics are reported in the next section 3.6.2 (see Table 5). A comparison[14] of injuries by type of crash shows that ePMD users who crashed alone were more severely injured than those who hit a pedestrian, in police-reported crashes, F(1, 702) = 679, p < .001, d = 2. Second, ePMD users who crashed alone were slightly more severely injured than those who crashed into a vehicle, on average, F(1, 2,862) = 33, p < .001, d = 0.3. Finally, ePMD users who hit a pedestrian were less seriously injured than those who hit a vehicle, on average, F(1, 2,688) = 862, p < .001, d = 1.9. These comparisons highlight the significant variations in injuries of ePMD users involved in crashes, depending on whether these users crashed alone, hit a pedestrian, or hit another vehicle, F(2, 3,126) = 450, p < .001. A Tukey post hoc test (HSD) at the threshold p < .05 - given the series of three comparisons - confirmed these differences in injuries of ePMD users, according to the three types of conditions in which they crasheds. Among the crashes recorded by the police, ePMD users who crashed alone were more seriously injured than those who hit pedestrians (mean difference of 1.0 out of 3, p < .001) or, in a lesser extent, than those who hit another vehicle (average difference of 0.15 out of 3, p < .001). ePMD users who hit a vehicle were injured more severely than those who hit pedestrians (average difference of 0.9 out of 3, p < .001).

Helmet wearing and injury severity. Only 12% of the 42 ePMD users killed in 2019, 2020 or 2021 were wearing a helmet; 23% of the 412 hospitalized ePMD users were wearing a helmet, and 30% of the 2,229 slightly injured ePMD users were wearing a helmet. The ePMD users who were not wearing a helmet were generally more severely injured than ePMD users that were wearing a helmet, F(1, 3,149) = 7.6, p = 0.006, d = 0.11.

3.6.2 Injuries of road users colliding with ePMD users

Table 5shows the injuries of ePMD users involved in crashes in France between 2019 and 2021, and the injuries of users involved in crashes with ePMD users.

Table 5

Injuries of users involved in a crash with an ePMD between 2019 and 2021, and comparison
Type of crashes involving ePMD users (first four rows) and collisions between an ePMD and another category	Crashes between 2019 and 2021^A	Injury rates (% of ePMD users involved in this type of crash [first four rows], and % of users of this category struck by an ePMD user)^B				Severity of injuries from 0 to 3
						For the users colliding with ePMD users		For the ePMD user
		Death	Hospitalization	Minor injury	Uninjured	M	SD	M	SD
Collision between vehicles	2,275	1.2	13.6	79.8	5.9			1.1	0.5
Only the ePMD user (no third parties)	412	3.6	22.1	71.8	2.4			1.3	0.6
Struck pedestrian	255		1.2	16.5	82.7			0.2	0.4
Collision between vehicles and struck pedestrian	21		9.5	85.7	4.8			1.0	0.4
Light vehicles^***	1,540		0.2	1.3	98.7	0.0	0.1	1.1	0.4
Pedestrian ^{† ***}	270	1.5	8.1	84.4	7.0	1.0	0.5	0.2	0.4
Commercial vehicle over 1.5T and under 3.5T^***	219		0.5		99.5	0.0	0.1	1.1	0.4
Bicycle ^†	122		5.7	66.4	27.9	0.8	0.5	0.7	0.6
Motorcycle < 50cm3^***	54		3.7	27.8	68.5	0.4	0.6	0.9	0.4
Motorcycle > 50cm3 and < = 125cm3 ^†	50		10.0	44.0	46.0	0.6	0.7	0.8	0.6
Motorcycle > 125cm3^***	46		10.9	34.8	54.3	0.6	0.7	1.0	0.7
Bus^***	45		2.2		97.8	0.0	0.3	1.2	0.7
Truck^***	26				100.0			1.5	0.8
Voiturette^***	17		5.9		94.1	0.1	0.5	1.2	0.4
Tram^***	12				100.0			1.4	0.7
3-wheeled motorized > 125 cm3 ^†	11			63.6	36.4	0.6	0.5	0.5	0.5
Other vehicle^***	6				100.0			1.2	0.4
Non-motorized PMD	3		33.3	66.7		1.3	0.6	0.7	1.2
Tractor^***	3				100.0			1.7	0.6
E-bike ^†	3			66.7	33.3	0.7	0.6	0.7	1.2
^A Crashes for which the police provide information on the characteristics of the users involved (most of the ePMD crashes)
^B Distribution of injuries, whether those of the ePMD user (first four rows) or the user involved in a collision with an ePMD user
^† Injury rate greater than 50% of crashes with an ePMD user
Comparison between the injuries of users in this category hit by ePMD users and the injuries of ePMD users involved in these crashes: ^*p < .001; ^p < .01; ^* p < .05
Only categories involved in more than 2 crashes with an ePMD user between 2019 or 2021 are shown. Crashes for which police did not provide user characteristics are not represented. Vehicle passengers are not included: these statistics are only for pedestrians, drivers of ePMDs and drivers of vehicles involved in crashes with ePMD users.

Table 5 reveals that only some of the vulnerable road users were injured in more than half of the crashes with an ePMD user (i.e., pedestrian, bicycle, motorcycle > 50cm3 and < = 125cm3, 3-wheeled motorized > 125 cm3, e-bike). On the other hand, the other categories of users were unharmed in more than 50% of the crashes with an ePMD user[15]. Crashes with ePMD users were mainly dangerous for the physical integrity of vulnerable users. Therefore, users protected by their vehicle were unharmed in more than 50% of the collisions with ePMD users, and it was the ePMD user who was usually injured (the police only report crashes with injuries, in theory). An ANOVA[16] confirmed that a vulnerable user involved in a crash with an ePMD user (but not the ePMD user himself/herself) would generally be more severely injured than other “non-vulnerable” users struck by an ePMD, F(1, 2,343) = 1,422, p < .001, d = 2.3.

The injury severity comparisons shown in Table 5 reveal that ePMD users were generally more severely injured than most of the users they collided with, except for some exceptions of user categories more severely injured than the ePMD user they collided with (i.e., pedestrians); or with equivalent injury severity (i.e., bicycles, motorcycles > 50cm3 and < = 125cm3, 3-wheeled motorized > 125 cm3, non-motorized PMDs, e-bikes). Following a crash with an ePMD user, the most vulnerable category of users were pedestrians: 1.5% of pedestrians struck by an ePMD between 2019 and 2021 were killed, 8.2% had injuries requiring hospitalization, and 84% were slightly injured. Regarding the crashes recorded by the police, a pedestrian hit by an ePMD user was rarely unharmed (“only” 7% of pedestrians struck by ePMD users were unharmed). As a result, ePMD users were usually unharmed after a collision with a pedestrian (the police only record crashes that lead to injuries, in principle). Similarly, cyclists were usually injured after a collision with an ePMD user, although no cyclist died after a crash with an ePMD user between 2019 and 2021. In general, cyclists involved in collisions with ePMD users suffered minor injuries (66% of cyclists involved in crashes with ePMD users). They sometimes required hospitalization (5.7% of cyclists involved in crashes with ePMD users). The cyclist was unharmed - and the ePMD user was injured - in 28% of the collisions between a cyclist and an ePMD user. To a lesser extent, other vulnerable road users were sometimes injured following a crash with an ePMD user. For example, users of motorcycles with an engine size under 50 cm³ were uninjured in 69% of the collisions between a motorcyclist with an engine size under 50 cm³ and an ePMD user. When they were injured by a collision with an ePMD user, motorcyclists with an engine under 50cm³ were generally slightly injured (28% of crashes with an ePMD user), or hospitalized (3.7%). No motorcyclist riding a motorcycle with an engine size under 50 cm³ died after a crash with an ePMD user in 2019, 2020 or 2021.

Summary of differences (section 3.5 and section 3.6).

ePMD users who collided with another road user were generally younger than the other road user, with a few categories of road users with similar ages.
Among police-recorded crashes, ePMD users who crashed alone were more seriously injured than those who hit a pedestrian or, to a much lesser extent, those who hit a vehicle.
ePMD users involved in crashes who did not wear helmets were generally more seriously injured than those who were wearing helmets.
Vulnerable road users colliding with an ePMD user were more seriously injured than other road users who collided with ePMD users.
Users colliding with an ePMD user were generally less severely injured than the ePMD user himself/herself, with a few exceptions of categories that were generally more severely injured (i.e. pedestrians) or had approximately the same injuries.

[7] ePMDs represented more than 2% of the crashes recorded in Maine-et-Loire, Marne, Meurthe-et-Moselle, Pyrénées-Orientales, Haut-Rhin, Rhône, Sarthe, Paris, Somme, Hauts-de-Seine, Val-de-Marne.

[8] Occasionally, ePMD crashes occurred in a context that police reported as “dazzling” (1.3%), during a storm or windy episode (0.1%), in the case of fog or smoke on the road (0.17%) or other weather conditions, identified by the police under the “other” category (0.27%).

[9] Exceptionally, the police reported that the pavement was covered with ice (0.07% of ePMD crashes), puddles (0.07%), greasy substances such as oil (0.1%), a flood (0.03%) or other conditions, identified under the “other” category (0.43%).

[10] Police reported the characteristics of the ePMD users involved in 412 crashes without a third party (100% of non-third party crashes); 2,275 crashes between an ePMD user and another vehicle - without a pedestrian being struck (99% of crashes between an ePMD and another vehicle); 255 crashes in which a pedestrian was struck - without a vehicle being struck (92% of pedestrians struck by an ePMD user); and 21 crashes in which a vehicle-to-vehicle collision resulted in a pedestrian being struck (68% of these crashes).

[11] This analysis only considers the 3,151 ePMD users for whom the police indicated their ages and whether they were wearing a helmet or not.

[12] This analysis only considers the 3,113 ePMD users for whom the police reported both their gender and whether they are wearing a helmet or not.

[13] These analyses do not consider crashes that simultaneously involved a struck pedestrian and several vehicles.

[14] These analyses did not considered crashes involving a vehicle-to-vehicle collision and a pedestrian being struck, simultaneously.

[15] The user categories who were injured in less than half of the collisions with an ePMD user are: light vehicles, commercial vehicles over 1.5T and under 3.5T, motorcycles < 50cm3, motorcycles > 125cm3, buses, trucks, voiturettes, trams, other vehicles, tractors, 3-wheeled motorized <= 50cm3, special vehicles, trains.

[16] This analysis focuses on collisions between an ePMD user and other vehicles or pedestrians: vehicle passengers were not included in this analysis, and the crashes between two ePMD users are not included in this analysis. This analysis considers the following categories as vulnerable users: pedestrians, cyclists (e-bikes and non-motorized bikes), non-motorized PMDs, quads, motorcyclists and motorized three-wheelers.

Rise in ePMD crashes in France. This article provides information about ePMD crashes and their contexts for researchers and policy makers, from surveys completed by the French police after major crashes. Between 2019 and 2021, ePMD crashes accounted for 1.8% of all crashes resulting in injury or death on French roads. French police officers reported ePMD crashes every week over a 3-year period (2019–2021), apart from a single week at the end of the first COVID-19 lockdown in 2020.

The number of ePMD-related crashes increased between 2019 and 2021, as well as their relative importance in the total number of crashes recorded by the police: they accounted for 1.1% of the 2019 crashes in France, 1.8% of 2020 crashes, and 2.6% of 2021 crashes. In 2020, ePMDs were the main user category involved in more crashes than in 2019. Similarly, ePMDs were the main user category involved in more crashes in 2021, compared to 2019. Other categories than ePMD were involved in more crashes in 2020 compared to 2019 (other vehicles, e-bikes, quads, non-motorized PMDs, 3-wheeled motorized < = 50cm3, not specified) or in 2021, compared to 2019 (motorcycles < 50cm3, commercial vehicles over 1.5T and under 3.5T, bicycles, 3-wheeled motorized > 125 cm3, other vehicles, e-bikes, quads, non-motorized PMDs, trams, trains, not specified). These results are in line with studies that indicate an increase in the number of hospitalized ePMD users, without specifying the relative importance of these ePMD crashes in the total number of road traffic accidents (e.g., in the US, Kobayashi et al., 2019; Tischler et al., 2021).

In France, the majority of ePMD crashes take place in the Paris region (54% of ePMD crashes reported between 2019 and 2021). The Paris department - which includes only the city of Paris - accounted for 34% of the ePMD crashes in France between 2019 and 2021. Between 2019 and 2021, ePMD crashes were reported in 88% of French departments. On one hand, more and more departments reported ePMD crashes between 2019 and 2021: “only” 49% of the French departments reported at least one ePMD crash in 2019, 67% in 2020 and 80% in 2021. On the other hand, the rise in ePMD crashes in 2020 concerned 52% of French departments; and 78% of French departments reported an increase in the number of ePMD crashes in 2021, compared to 2019. Likewise, police officers reported ePMD crashes in all Paris districts between 2019 and 2021; and almost all Paris districts contributed to the increase in ePMD crashes in 2020 and 2021. These results reflecting a rise and a spread of ePMD crashes in France complement the official reports, which do not specify the types of crashes and the geographical areas concerned by this increase in ePMD crashes (ONISR, 2020, 2021, 2022b).

Types of ePMD crashes and contexts. The majority of ePMD crashes are collisions between vehicles (76% of ePMD crashes between 2019 and 2021). The remaining ePMD crashes involved only the ePMD user (14%) or involved a pedestrian struck by an ePMD user (9.2%). Police identified a few vehicle-to-vehicle crashes - including an ePMD - in which a pedestrian was struck (1% of the ePMD crashes).

Almost all (98%) ePMD crashes reported by the French police between 2019 and 2021 occurred in urban areas. Thus, it is not surprising that the majority of crashes leading to the hospitalization or death of an ePMD user take place in an urban areas, or that the official report on road traffic accidents indicated an increase in ePMD users injured in urban areas in 2020 (ONISR, 2021). Other categories of users are usually seriously injured or killed in urban areas, especially vulnerable users. For example, in the European Union, “around 70% of road fatalities in urban areas involve vulnerable road users which includes pedestrians, motorcyclist and cyclists”, non-motorized PMDs and ePMDs (European-Commission, 2021, p. 1).

In France, ePMD crashes typically occur during normal business hours, between 8 a.m. and 8 p.m. (74% of the reported ePMD crashes between 2019 and 2021 occurred in this time frame). Most of these crashes take place under ‘normal’ weather conditions - without rain or other bad weather conditions (83% of ePMD crashes). Importantly, ePMD crashes occur mainly at an intersection (59% of ePMD crashes). In France, ePMD crashes occur mainly on the roadway (72% of ePMD crashes), on a cycle path (16%), or even on a sidewalk - prohibited to ePMDs since 2019 (5.8%).

Characteristics and injuries of users involved in ePMD crashes. The majority of injured ePMD users are males under the age of 40 (76% of ePMD users involved in crashes in France between 2019 and 2021). In two-thirds of the cases, these ePMD users did not wear a helmet, regardless of whether the user was male or female: male and female ePMD users are equally inclined not to wear a helmet. ePMD users involved in crashes who don’t wear helmets tend to be younger than those with helmets. These results are consistent with previous studies suggesting that risky behavior is more common among younger e-scooter users (e.g., in Paris, France, Gioldasis et al., 2021; in Trondheim, Norway, Pazzini et al., 2022). However, the equal tendency of male and female ePMD riders involved in crashes to not wear helmets provides a major contribution to the research claiming that there is a difference between male and female risky behaviors. For example in Paris (France), Gioldasis et al. (2021, p. 11) argues - without considering helmet use - that “male e-scooter riders are more likely to get engaged in risky behavior such as to drive after having consumed alcohol, after having consumed drugs, and to drive the e-scooter while using the smartphone”. On the other hand, a survey of 210 e-scooter users in the US revealed that females are more likely to commit risky behaviors such as riding on a sidewalk (Tian et al., 2022). Our results are consistent with research that identifies no difference between male and female risky behavior (e.g., the speed of male e-scooter users was comparable to that of female users in Trondheim, Norway, Pazzini et al., 2022). More generally than road crashes, gender differences in risky behaviors are variable, depending on the domain (Byrnes et al., 1999).

ePMD users who crash alone are generally older than those who hit another user (whether pedestrian or vehicle). To our knowledge, these age differences have not yet been identified by research. In any case, they are not presented in French road traffic accidents reports (ONISR, 2020, 2021, 2022b).

Crashes between an ePMD and another road user are mostly collisions between an ePMD user and a light vehicle (car or commercial vehicle under 3.5T), a pedestrian, a commercial vehicle or a cyclist. The number of crashes between an ePMD and a cyclist doubled between 2019 and 2020. This increase in crashes between ePMDs and cyclists helped to maintain stability in the overall number of crashes involving cyclists in France between 2019 and 2020 - whether or not they involved an ePMD. Similarly, in 2021, the increase in crashes between a cyclist and an ePMD user contributed to the increase in the number of crashes involving cyclists.

After light vehicles, pedestrians, commercial vehicles and cyclists, the other user categories most often involved in a collision with an ePMD are riding/driving motorcycles < 50cm3, motorcycles > 50cm3 and < = 125cm3, buses, motorcycles > 125cm3 or, to a lesser extent, trucks. Apart from collisions between an ePMD and a light vehicle or a commercial vehicle, the other user categories most frequently involved in a collision with an ePMD are generally vulnerable users (i.e., pedestrians, cyclists and motorcyclists).

When they collide with a pedestrian, ePMD users generally strike a female pedestrian. However, crashes between ePMD users and other users generally involve males, either the ePMD user or the other user involved. Next, ePMD users are generally younger than the users they crash into.

In general, ePMD users who crash alone are more seriously injured than those who collide with another user, especially when that other user is a pedestrian. However, crashes between an ePMD user and another vehicle are much more frequent than crashes involving only the ePMD user. Nearly three-quarters of ePMD users hospitalized or killed between 2019 and 2021 collided with a vehicle, and about one-quarter of these hospitalized or killed ePMD users crashed alone. These findings are consistent with studies highlighting the fact that ePMD users are generally hospitalized after a vehicle collision (e.g., in South Korea, J. Y. Kim et al., 2021; M. Kim et al., 2021; in France, ONISR, 2022b). However, some studies have pointed out that ePMD users mostly crash alone, for example in the US (Alwani et al., 2020), in Berlin - Germany (Uluk et al., 2021) or in New Zealand (Brownson et al., 2019).

Our study reveals that ePMD users who crash without wearing a helmet are generally more severely injured than those who do wear one. This result extends previous findings from the ONISR reports (2020, p. 66) and a previous study on head injuries following e-scooter crashes in Paris (Hennocq et al., 2020). According to studies based on hospitalized e-scooter users, this result of ePMD crashes being more severe for users who don’t wear a helmet concerns several countries, such as New-Zealand (Brownson et al., 2019), Germany (Uluk et al., 2021), South Korea (Choi et al., 2022), and the US (Dhillon et al., 2020; Kobayashi et al., 2019; Trivedi et al., 2019), etc. The protective function of helmet is acknowledged, even though some confounding variables may intervene such as a possible link between lack of personal protection equipment and risk proneness. For example, in Brisbane (Australia) e-scooter users who ride on sidewalks are less likely to wear a helmet than other e-scooter users on the road (Haworth et al., 2021); but, in other locations, e-scooter users have the same inclination not to wear a helmet, regardless of whether they are riding on the sidewalk, a cycle path or the road (e.g., in Los Angeles, Todd et al., 2019).

In general, the vehicle type has been identified as a determinant of injury and medical cost following a crash (e.g., Zakeri et al., 2021). Our study extends this finding to ePMD crashes, highlighting the fact that injuries depend on the type of user struck by the ePMD. Users protected by a vehicle body are generally unharmed after a collision with an ePMD user, the latter being generally injured. Therefore, vulnerable road users are more likely to be injured or killed after a collision with an ePMD user than other road users (not in the vulnerable road user category). Pedestrians are the category of road user who are the most frequently injured or killed after a collision with an ePMD. Pedestrians are also the primary user category struck by an ePMD user on a sidewalk: between 2019 and 2021, details provided by the police about the actions and location of pedestrians who had been struck indicate that 26% of pedestrians struck by ePMDs were on a sidewalk; and during the same period, the general question about the location of the crash indicates that 17.9% of pedestrians struck by ePMDs were on a sidewalk. These results confirm the dilemma faced by the French public authorities concerning the widespread use of ePMDs, as in other locations around the world: should ePMDs be made to travel exclusively on the road, or should they be allowed to share the areas of vulnerable users such as cycle paths or sidewalks ? For example, the spread of e-scooter travel in the US “has led to a dilemma of where scooter riders should be as they have the potential to seriously injure pedestrians in collisions, but on the road, riders are relatively unprotected” (Lavoie-Gagne et al., 2021, p. 1).

Limitations

Period under consideration. Given that ePMDs are a new mobility device, a three-year perspective allows trends to be analyzed. However, the French data available so far do not allow a wider period to be considered: the ePMD category was added to the police-completed survey in 2019, as this category was previously combined with the category “other” vehicles (ONISR, 2022a, p. 22). Likewise, the e-bike category was added in 2019 to the survey completed by the police after a road traffic accident (e-bike users were previously classified as cyclists).

Unrecorded crashes. Crashes that result in injury are not systematically reported to the police. For example, “crashes involving an injured cyclist are not always known to the police who contribute to the BAAC file; they are not always called before a transfer to hospital, especially if no motorized user is involved” (ONISR, 2020, p. 64). It is reasonable to assume that this limitation is particularly relevant for users with minor injuries (e.g., some slightly injured users do not call the police or emergency services, even when users seek medical attention following the crash). In any case, the data used for this study and the official report on road crashes in France (ONISR, 2020, 2021, 2022b) underestimate the number of crashes on French roads.

Improvement of the survey completed by the French police after a crash

Officially, the surveys completed by the French police should not include crashes without injuries. However, the police don’t necessarily provide information on user injuries, among other characteristics that are not systematically reported. Next, the police report that ePMD users are unharmed in 2.4% of the crashes involving only an ePMD user (such crashes should not be reported by the police). Moreover, it is possible that some of the crashes reports completed between 2019 and 2021 are affected by typing errors, in addition to inconsistencies between the general questions about the crashes and the details about the vehicle actions (see section 2.3).

Some questions answered by the French police after a crash are not included in the open-data provided by the Ministry of the Interior, such as the exact model of the vehicles involved. For example, the official report on road traffic accidents indicates that the ePMD category includes wheelchairs, although “these cases are rare: of the 24 deaths [of ePMD users in 2021], 2 were on a vehicle adapted for disability” (ONISR, 2022b, p. 82). However, French law states that people in wheelchairs (motorized or not) belong to the category of pedestrians, and share the same mobility areas than pedestrians: in these conditions, people in wheelchairs should not be considered as ePMDs, according to the French Office of Legal and Administrative Information (Direction de l’information légale et administrative, 2021). To avoid this imprecision about the various vehicles grouped in the “ePMD” category, French crash data issued on an open-data basis should specify the type of vehicle (e.g., manufacturer and model), just like the open data issued by the British authorities (UK Department for Transport, 2022). These details would make it possible to identify similarities and differences between the various ePMDs, in terms of crash and user characteristics, injuries, etc. For example, J. Y. Kim et al. (2021) finds differences between the injuries of hospitalized South-Korean ePMD users, depending on their ePMD type (i.e., e-scooter, e-wheel, and e-board); and observations by Haworth et al. (2021) in Brisbane (Australia) emphasize that users of “shared” (rental) e-scooters are less inclined to wear a helmet than users of private e-scooters.

Regarding the location of crashes, there are inconsistencies between the departments indicated by the police and the coordinates provided (longitude and latitude of the crash). For example, 5 crashes involving an ePMD between 2019 and 2021 were associated by the police with the department of Paris (75) and with coordinates locating these crashes in neighboring towns - near the border between Paris and these towns. In section 3.2.3, these crashes are considered to be in the Paris department - according to the question about the department where the crash occurred, but they were not in any of the Paris districts (and not in the statistics describing each Paris district presented in Table B.1 - Appendix B).

Applications

Road safety campaigns and trainings for ePMD users would benefit from targeting some geographical areas where the police record the most ePMD crashes. In addition to these spatial conditions, trainings and prevention campaigns aimed at ePMD users should target males under 40 years old, who represent the majority of ePMD users injured in France between 2019 and 2021. Finally, campaigns aimed at younger users should emphasize the importance of wearing a helmet, given the fact that younger ePMD users who are injured in France are less likely to wear a helmet than older ePMD users. Tailored messages toward this population may be useful, assuming that the arguments are convincing enough and well-conceived (Teeny et al., 2020). Mandatory helmet use may be considered by public policy makers.

In France, 14% of ePMD users involved in crashes between 2019 and 2021 were hospitalized and 1.4% died, considering only ePMD police-reported crashes (injuring the ePMD user and/or another user involved in the crash). Vulnerable users struck by ePMD users were generally injured, while the ePMD user was uninjured. According to this finding, research that considers only hospitalized ePMD users excludes the majority of crashes between a vulnerable user and an ePMD user. On the other hand, crashes between an ePMD and a user protected by the body of their vehicle are over-represented by research based on hospitalized ePMD users only. Such bias may be corrected by more systematic records of the context of road crashes.

Funding. This work was supported by the Délégation à la Sécurité Routière (DSR), Paris, France [grant number: RP1- J20100.100LAB5CPR].

Corresponding author. Clément Laverdet ([email protected])

The authors have no financial or non-financial interests that are directly or indirectly related to the work submitted for publication. As such, there are no interests to disclose that could potentially influence the content or findings of the work.

Akter, S., Mamun, M. M. H., Mwakalonge, J. L., Comert, G., & Siuhi, S. (2021). A Policy Review of Electric Personal Assistive Mobility Devices. Transportation Research Interdisciplinary Perspectives, 11. https://doi.org/10.1016/j.trip.2021.100426
Allaire, J., Horner, J., Xie, Y., Marti, V., & Porte, N. (2019). Markdown: Render Markdown with the c Library Sundown [R Package].
Alwani, M., Jones, A. J., Sandelski, M., Bandali, E., Lancaster, B., Sim, M. W., Shipchandler, T., & Ting, J. (2020). Facing Facts: Facial Injuries from Stand-up Electric Scooters. Cureus. https://doi.org/10.7759/cureus.6663
Ben-Shachar, M. S., Lüdecke, D., & Makowski, D. (2020). Effectsize: Estimation of Effect Size Indices and Standardized Parameters. Journal of Open Source Software, 5(56), 2815. https://doi.org/10.21105/joss.02815
Bivand, R., Keitt, T., Rowlingson, B., Pebesma, E., Sumner, M., Hijmans, R., & Rouault, E. (2015). Package « Rgdal ». Bindings for the Geospatial Data Abstraction Library. Available online: https://cran. r-project. org/web/packages/rgdal/index.html [R package].
Boglietti, S., Barabino, B., & Maternini, G. (2021). Survey on E-Powered Micro Personal Mobility Vehicles: Exploring Current Issues towards Future Developments. Sustainability, 13(7), 3692. https://doi.org/10.3390/su13073692
Brownson, A. B., Fagan, P. V., Dickson, S., & Civil, I. D. (2019). Electric Scooter Injuries at Auckland City Hospital. The New Zealand Medical Journal, 132(1505), 62‑72. PMID: 31697664.
Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender Differences in Risk Taking: A Meta-Analysis. Psychological Bulletin, 125(3), 367‑383. https://doi.org/10.1037/0033-2909.125.3.367
Campbell, A., Wong, N., Monk, P., Munro, J., & Bahho, Z. (2019). The Cost of Electric-Scooter Related Orthopaedic Surgery. New Zealand Medical Journal, 132(1501), 7.
Choi, J. H., Oh, S. S., Kim, K. S., Hwang, J. H., & Lee, S. Y. (2022). Electric Scooter-Related Trauma in Korea. Journal of Korean Medical Science, 37(22), e181. https://doi.org/10.3346/jkms.2022.37.e181
Christoforou, Z., de Bortoli, A., Gioldasis, C., & Seidowsky, R. (2021). Who Is Using E-Scooters and How? Evidence from Paris. Transportation Research Part D: Transport and Environment, 92, 102708. https://doi.org/10.1016/j.trd.2021.102708
Dhillon, N. K., Juillard, C., Barmparas, G., Lin, T.-L., Kim, D. Y., Turay, D., Seibold, A. R., Kaminski, S., Duncan, T. K., & Diaz, G. (2020). Electric Scooter Injury in Southern California Trauma Centers. Journal of the American College of Surgeons, 231(1), 133‑138. https://doi.org/gp8cfj
Dias, G., Arsenio, E., & Ribeiro, P. (2021). The Role of Shared E-Scooter Systems in Urban Sustainability and Resilience during the Covid-19 Mobility Restrictions. Sustainability, 13(13), 7084. https://doi.org/10.3390/su13137084
Direction de l’information légale et administrative. (2021). Un Fauteuil Roulant Électrique n’est Pas Un Véhicule à Moteur. Www.Service-Public.Fr/Particuliers/Actualites/A14918.
Direction de l’information légale et administrative. (2022). Circulation En Trottinette Électrique, Rollers Ou Skateboard. Www.Service-Public.Fr/Particuliers/Vosdroits/F308.
Dowle, M., Srinivasan, A., Gorecki, J., Chirico, M., Stetsenko, P., Short, T., Lianoglou, S., Antonyan, E., Bonsch, M., & Parsonage, H. (2019). Package « Data. Table ». [R package].
European-Commission. (2021). Press Release (April 20, 2021) on Road Safety: 4,000 Fewer People Lost Their Lives on EU Roads in 2020 as Death Rate Falls to All-Time Low. https://ec.europa.eu/commission/presscorner/detail/en/ip_21_1767.
European-Commission. (2022). Glossary: Gross Vehicle Weight. https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Gross_vehicle_weight.
Fekl, M. (2017). Guide de Rédaction Du Bulletin d’analyse Des Accidents Corporels de La Circulation (BAAC), in Www.Onisr.Securite-Routiere.Gouv.Fr/Sites/Default/Files/2019-02/Guide%20BAAC%202017.Pdf.
Gioldasis, C., Christoforou, Z., & Seidowsky, R. (2021). Risk-Taking Behaviors of e-Scooter Users: A Survey in Paris. Accident Analysis & Prevention, 163. https://doi.org/10.1016/j.aap.2021.106427
Gohel, D. (2021). Flextable: Functions for Tabular Reporting [R Package].
Haworth, N., Schramm, A., & Twisk, D. (2021). Comparing the Risky Behaviours of Shared and Private E-Scooter and Bicycle Riders in Downtown Brisbane, Australia. Accident Analysis & Prevention, 152. https://doi.org/gjkd4j
Hennocq, Q., Schouman, T., Khonsari, R. H., Sigaux, N., Descroix, V., Bertolus, C., & Foy, J.-P. (2020). Evaluation of Electric Scooter Head and Neck Injuries in Paris, 2017-2019. JAMA Network Open, 3(11). https://doi.org/10.1001/jamanetworkopen.2020.26698
INSEE. (2022). Population de 1999 à 2022 : Comparaisons Départementales [Population from 1999 to 2022: Departmental Comparisons]. In Institut national de la statistique et des études économiques.
Kahle, D., Wickham, H., & Jackson, S. (2019). Ggmap: Spatial Visualization with Ggplot2 [R Package].
Kim, J. Y., Lee, S. C., Lee, S., Lee, C. A., Ahn, K. O., & Park, J. O. (2021). Characteristics of Injuries According to Types of Personal Mobility Devices in a Multicenter Emergency Department from 2011 to 2017: A Cross-Sectional Study. Medicine, 100(6). https://doi.org/10.1097/MD.0000000000024642
Kim, M., Lee, S., Ko, D. R., Kim, D.-H., Huh, J.-K., & Kim, J.-Y. (2021). Craniofacial and Dental Injuries Associated with Stand-up Electric Scooters. Dental Traumatology, 37(2), 229‑233. https://doi.org/10.1111/edt.12620
Kobayashi, L. M., Williams, E., Brown, C. V., Emigh, B. J., Bansal, V., Badiee, J., Checchi, K. D., Castillo, E. M., & Doucet, J. (2019). The E-Merging e-Pidemic of e-Scooters. Trauma Surgery & Acute Care Open, 4(1), e000337. https://doi.org/10.1136/tsaco-2019-000337
Lavoie-Gagne, O., Siow, M., Harkin, W. E., Flores, A. R., Politzer, C. S., Mitchell, B. C., Girard, P. J., Schwartz, A. K., & Kent, W. T. (2021). Financial Impact of Electric Scooters: A Review of Injuries over 27 Months at an Urban Level 1 Trauma Center (Cost of e-Scooter Injuries at an Urban Level 1 Trauma Center). Trauma Surgery & Acute Care Open, 6(1), e000634. https://doi.org/10.1136/tsaco-2020-000634
McGuinness, M. J., Tiong, Y., & Bhagvan, S. (2021). Shared Electric Scooter Injuries Admitted to Auckland City Hospital: A Comparative Review One Year after Their Introduction. The New Zealand Medical Journal, 134(1530), 21‑29.
McKenzie, G. (2019). Spatiotemporal Comparative Analysis of Scooter-Share and Bike-Share Usage Patterns in Washington, D.C. Journal of Transport Geography, 78, 19‑28. https://doi.org/10.1016/j.jtrangeo.2019.05.007
Ministère de l’Intérieur. (2019). Décret N 2019-1082 Du 23 Octobre 2019 Relatif à La Réglementation Des Engins de Déplacement Personnel, Article 3. (p. www.legifrance.gouv.fr/jorf/article_jo/JORFARTI000039272669).
Ministère de l’Intérieur. (2022). Bases de données annuelles des accidents corporels de la circulation routière - Années de 2005 à 2021 [Databases of road traffic injuries in France]. www.data.gouv.fr/fr/datasets/bases-de-donnees-annuelles-des-accidents-corporels-de-la-circulation-routiere-annees-de-2005-a-2021.
Namiri, N. K., Lui, H., Tangney, T., Allen, I. E., Cohen, A. J., & Breyer, B. N. (2020). Electric Scooter Injuries and Hospital Admissions in the United States, 2014-2018. JAMA Surgery, 155(4), 357. https://doi.org/10.1001/jamasurg.2019.5423
OECD. (2022). IRTAD-group Data Questionnaire: Road Fatalities by Age. https://stats.oecd.org/Index.aspx?DataSetCode=IRTAD_CASUAL_BY_AGE#.
ONISR. (2020). La Sécurité Routière En France: Bilan de l’accidentalité de l’année 2019. OBSERVATOIRE NATIONAL INTERMINISTÉRIEL DE LA SÉCURITÉ ROUTIÈRE in Www.Onisr.Securite-Routiere.Gouv.Fr/Sites/Default/Files/2020-09/Bilan_2019_version_site_internet_24_sept.Pdf (ISBN: 978 211 077525-2).
ONISR. (2021). La Sécurité Routière En France: Bilan de l’accidentalité de l’année 2020. OBSERVATOIRE NATIONAL INTERMINISTÉRIEL DE LA SÉCURITÉ ROUTIÈRE in Www.Onisr.Securite-Routiere.Gouv.Fr/Sites/Default/Files/2021-09/ONISR_Bilan_Accidentalit%C3%A9_2020.Pdf (ISBN: 978 211 077535-1).
ONISR. (2022a). Guide de production du fichier BAAC. OBSERVATOIRE NATIONAL INTERMINISTÉRIEL DE LA SÉCURITÉ ROUTIÈRE in www.onisr.securite-routiere.gouv.fr/sites/default/files/2022-03/ONISR_Guide%20de%20production%20du%20fichier%20BAAC_0.pdf. 32.
ONISR. (2022b). La Sécurité Routière En France: Bilan de l’accidentalité de l’année 2021. OBSERVATOIRE NATIONAL INTERMINISTÉRIEL DE LA SÉCURITÉ ROUTIÈRE in Www.Onisr.Securite-Routiere.Gouv.Fr/Sites/Default/Files/2022-09/ONISR_Bilan_Accidentalit%C3%A9_2021_0.Pdf (ISBN: 978 211 077549-8).
Pazzini, M., Cameli, L., Lantieri, C., Vignali, V., Dondi, G., & Jonsson, T. (2022). New Micromobility Means of Transport: An Analysis of E-Scooter Users’ Behaviour in Trondheim. International Journal of Environmental Research and Public Health, 19(12), 7374. https://doi.org/10.3390/ijerph19127374
Puzio, T. J., Murphy, P. B., Gazzetta, J., Dineen, H. A., Savage, S. A., Streib, E. W., & Zarzaur, B. L. (2020). The Electric Scooter: A Surging New Mode of Transportation That Comes with Risk to Riders. Traffic injury prevention, 21(2), 175‑178. https://doi.org/gp8cf3
R Core Team. (2021). R: A Language and Environment for Statistical Computing [Manual]. R Foundation for Statistical Computing.
Schauberger, P., & Walker, A. (2021). Openxlsx: Read, Write and Edit Xlsx Files [R Package].
Schneeweiss, M., Hassan-Ali, M., & Kam, A. (2021). Safety and Risk Factors Associated with Electric Scooter Use Globally: A Literature Review. McMaster University Medical Journal, 18(1), 13. https://doi.org/10.15173/mumj.v18i1.2588
Teeny, J. D., Siev, J. J., Briñol, P., & Petty, R. E. (2020). A Review and Conceptual Framework for Understanding Personalized Matching Effects in Persuasion. Journal of Consumer Psychology, 31(2), 382‑414. https://doi.org/10.1002/jcpy.1198
Tian, D., Ryan, A. D., Craig, C. M., Sievert, K., & Morris, N. L. (2022). Characteristics and Risk Factors for Electric Scooter-Related Crashes and Injury Crashes among Scooter Riders: A Two-Phase Survey Study. International Journal of Environmental Research and Public Health, 19(16), 10129. https://doi.org/10.3390/ijerph191610129
Tischler, E. H., Laurent Tsai, S. H., Wolfert, A. J., Suneja, N., Naziri, Q., & Tischler, H. M. (2021). Orthopedic Fracture Hospitalizations Are Revving up from E-Scooter Related Injuries. Journal of Clinical Orthopaedics and Trauma, 23, 101607. https://doi.org/10.1016/j.jcot.2021.101607
Todd, J., Krauss, D., Zimmermann, J., & Dunning, A. (2019). Behavior of Electric Scooter Operators in Naturalistic Environments. WCX SAE World Congress Experience, 2019-01-1007. https://doi.org/10.4271/2019-01-1007
Trivedi, T. K., Liu, C., Antonio, A. L. M., Wheaton, N., Kreger, V., Yap, A., Schriger, D., & Elmore, J. G. (2019). Injuries Associated With Standing Electric Scooter Use. JAMA Network Open, 2(1). https://doi.org/ggfnzd
UK Department for Transport. (2022). UK Road Safety Data. www.data.gov.uk/dataset/cb7ae6f0-4be6-4935-9277-47e5ce24a11f/road-safety-data.
Uluk, D., Lindner, T., Dahne, M., Bickelmayer, J. W., Beyer, K., Slagman, A., Jahn, F., Willy, C., Möckel, M., & Gerlach, U. A. (2021). E-Scooter Incidents in Berlin: An Evaluation of Risk Factors and Injury Patterns. Emergency Medicine Journal. https://doi.org/10.1136/emermed-2020-210268
Ville de Paris. (2020). Paris Renforce La Régulation Des Trottinettes Électriques En Libre-Service. Www.Paris.Fr/Pages/Paris-Renforce-La-Regulation-Des-Trottinettes-Electriques-En-Libre-Service-6647 [Institutionnel]. In www.paris.fr.
Wickham, H. (2021). Tidyverse: Easily Install and Load the Tidyverse [R Package].
Wickham, H., & Seidel, D. (2020). Scales: Scale Functions for Visualization [R Package].
Xie, Y. (2021a). Bookdown: Authoring Books and Technical Documents with r Markdown [R Package].
Xie, Y. (2021b). Knitr: A General-Purpose Package for Dynamic Report Generation in r [R Package].
Zagorskas, J., & Burinskienė, M. (2019). Challenges Caused by Increased Use of E-Powered Personal Mobility Vehicles in European Cities. Sustainability, 12(1), 273. https://doi.org/10.3390/su12010273
Zakeri, R., Nosratnejad, S., Sadeghi-Bazargani, H., Dalal, K., & Yousefi, M. (2021). The Economic Burden of Road Traffic Injuries until One-Year after Hospitalization: A Survey Study. Accident Analysis & Prevention, 163, 106459. https://doi.org/10.1016/j.aap.2021.106459

No competing interests reported.

Appendix.docx

Crashes Involving Electric Personal Mobility Devices (ePMD) Reported by the French Police: Types of Crashes, Users Involved, Injuries and Explanatory Factors

Status:

Version 1

Abstract

Figures

Highlights

1 Introduction

2 Methods