See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266949934

ATV Rollover, Rider Response, and Determinants of Injury: In-depth Analysis of

Video-documented ATV Rollover Events

Article in Traffic Injury Prevention · September 2014

DOI: 10.1080/15389588.2014.935940 · Source: PubMed

CITATIONS READS

7 235

5 authors, including:

Chris Van Ee Daniel E Toomey

Wayne State University Design Research Engineering
35 PUBLICATIONS 973 CITATIONS 15 PUBLICATIONS 82 CITATIONS

SEE PROFILE SEE PROFILE

Barbara Moroski-Browne
Design Research Engineering
3 PUBLICATIONS 21 CITATIONS

SEE PROFILE

All content following this page was uploaded by Barbara Moroski-Browne on 13 December 2016.

The user has requested enhancement of the downloaded file.

 Traffic Injury Prevention

ATV Rollover, Rider Response, and Determinants of

Injury: In-depth Analysis of Video-documented ATV
Rollover Events
ab ab a a a
C. A. Van Ee , D. E. Toomey , B. A. Moroski-Browne , M. Vander Roest & A. Wilson
a
Design Research Engineering, Novi, Michigan
b
Biomedical Engineering, Wayne State University, Detroit, Michigan
Published online: 11 Oct 2014.

To cite this article: C. A. Van Ee, D. E. Toomey, B. A. Moroski-Browne, M. Vander Roest & A. Wilson (2014) ATV Rollover, Rider
Response, and Determinants of Injury: In-depth Analysis of Video-documented ATV Rollover Events, Traffic Injury Prevention,
15:sup1, S190-S196, DOI: 10.1080/15389588.2014.935940

To link to this article: http://dx.doi.org/10.1080/15389588.2014.935940

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Traffic Injury Prevention (2014) 15, S190–S196
Copyright  C Taylor & Francis Group, LLC

ISSN: 1538-9588 print / 1538-957X online

DOI: 10.1080/15389588.2014.935940

ATV Rollover, Rider Response, and Determinants of Injury:

In-depth Analysis of Video-documented ATV
Rollover Events
C. A. VAN EE1,2, D. E. TOOMEY1,2, B. A. MOROSKI-BROWNE1, M. VANDER ROEST1, and A. WILSON1
1
Design Research Engineering, Novi, Michigan
2
Biomedical Engineering, Wayne State University, Detroit, Michigan

Received 19 March 2014, Accepted 14 June 2014

Objective: All-terrain vehicle (ATV) rollover events can lead to serious and fatal injuries. Crush protection devices (CPDs) are
intended to reduce injury by reducing the frequency of significant contact between an inverted ATV and rider. Currently, field data
on real-world ATV rollovers are primarily limited to injury causing events and lack ATV and rider dynamics necessary to evaluate
injury mitigation effectiveness and possible unintended consequences of CPDs. Unlike restrained automobile occupants, ATV rider
posture and positioning are highly variable and scant data are available to define the dynamically changing rider position in a roll
scenario. Additional data on the complex real-world dynamics and interactions of riders and vehicles are needed to further develop
and evaluate the effectiveness of rollover injury prevention strategies.
Methods: Using YouTube videos of real-world rollover events, vehicle, environment, and rider factors were categorized with a focus
on vehicle dynamics and rider responses, including dismount kinematics.
Results: One hundred twenty-nine ATV rollover events were coded, with side rolls representing 47%, rear 44%, and forward rolls
9%. The speed at onset of roll was relatively low, with 86% of the rolls occurring at speeds of 10 mph or less and 53% occurring
at less than 3 mph. No injury was identified for 79% of the events; 16% resulted in injury due to ATV contact and 5% resulted in
injury unrelated to ATV contact. Active dismount of the ATV was a commonly employed strategy, with 63% of the riders attempting
active dismount, resulting in successful separation from the ATV in 72% of the attempts. The overall injury rate for riders attempting
active dismount was 15% and the injury rate for riders not attempting active dismount was 32%. This investigation confirmed the
importance of active rider movements, including active dismount and subsequent separation in determining the outcome of ATV roll
events.
Conclusions: Rider active dynamics need to be considered when introducing new injury prevention strategies that may obstruct,
impede, or otherwise contact riders during an attempted separation. To the authors’ knowledge, this is the first systematic use of
real-world video-documented ATV rollover events to quantify and analyze ATV rollover dynamics and rider responses. These data
and techniques can guide effective design and implementation of injury mitigation strategies.
Keywords: ATV, all-terrain vehicle, CPD, crush protection device, roll, YouTube, quadbike

Introduction ATVs in use in the United States increased from 1.8 million
to 10.6 million from 1990 to 2010. Similar to other forms
All-terrain vehicles (ATVs) continue to be popular for both of transportation, as the prevalence of ATVs has increased,
work and recreational applications. An ATV is defined by associated injuries and fatalities have become an increasing
the Consumer Product Safety Commission (CPSC) as an off- concern both within the United States and internationally. In
road, motorized vehicle having 3 or 4 low-pressure tires, a the United States during 2011, the CPSC estimated that there
straddle seat for the rider, and handlebars for steering con- were 107,500 ATV-related emergency department–treated in-
trol. The CPSC has estimated that the number of 4-wheel juries (27% were children younger than 16 years of age) and
327 reported ATV-related fatalities (57 were children younger
than 16 years of age). Coinciding with recent efforts by gov-
Associate Editor Joel Stitzel oversaw the review of this article ernment, industry, and public interest groups to address ATV
Address correspondence to C. A. Van Ee, Design Research safety through rider education and training, the CPSC also
Engineering, 46475 DeSoto Ct., Novi, MI 48377. E-mail: reported significant decreases in both injury and fatality risk
[email protected] since 2007 (CPSC 2013).
Color versions of one or more of the figures in the article can be One area of continued focus has been injuries and fatali-
found online at www.tandfonline.com/gcpi. ties occurring as a result of ATV rollovers (Helmkamp et al.
ATV Rollover Events 191

2011; Lower et al. 2012; Rechnitzer et al. 2013; Williams et al. rollover scenarios where ATV dynamics and rider active and
2014). Though injuries in ATV rollover scenarios may oc- passive responses are video documented. This allows for anal-
cur through a wide variety of mechanisms, crush protection ysis of the complex interaction between ATVs and active riders
devices (CPDs) are proposed to mitigate injuries that occur in rollover events.
specifically due to static or dynamic ATV–rider contact when
the rider is between the inverted ATV and the ground. CPDs
are roll bar–like structures that typically attach to the rear of Methods
the vehicle but, unlikeroll over protective structures, or ROPS,
CPDs do not incorporate occupant restraints such as a lap belt YouTube (www.YouTube.com), an open video-sharing service
or 3-point belt. The effectiveness of a CPD is predicated upon with more than 100 million videos, was searched for ATV
2 factors: First, the CPD must mitigate injury in the identified rollover events. For this study, an ATV rollover event was
inverted ATV injury mode, and second, the CPD must not in- defined as an incident where a moving 4-wheel ATV rotates
troduce new injury modes or exacerbate current injury modes at least one quarter turn rearward, forward, or to the side.
across the full range of ATV use and accident scenarios (unin- This study focused on live action videos that appeared to
tended consequences). If the unintended consequences are not be recordings of real-life events. The search terms used were
adequately evaluated prior to implementation of a CPD, the a combination of “ATV” or “quad” combined with one of
addition of the device may have an overall deleterious effect the following: “roll,” “rolls,” “rollover,” “accident,” “crash,”
on the ATV user community, resulting in increased frequency “end over end,” “hill wrecks,” “hill crash,” “flip,” “forward
and/or severity of injury as opposed to overall mitigation of flip,” or “wreck.” Videos containing rollover events were
harm. Accurate evaluation of a CPD must be grounded in real- downloaded to a local storage device using Real Player
world ATV and rider dynamics across the range of accident (http://www.real.com/). For each set of search terms, con-
scenarios that commonly occur. Injury and fatality databases secutive hits were examined until 75 consecutive hits yielded
of ATV accidents contain little information on the ATV and no new videos, and then a new search using a different set of
rider dynamics necessary to evaluate CPD performance. Fur- search terms was performed. If the video coverage was not
ther, by their design, injury and fatality databases, such as the sufficient to determine the primary rider and vehicle dynamics
CPSC’s in-depth investigations of ATV events, contain little, in the preroll and roll phases it was excluded from further anal-
if any, information on minor injury and noninjury rollover ysis and not included in the final categorized video database.
scenarios necessary to evaluate the possible unintended con- Two hundred and twelve ATV rollover videos were down-
sequences of the introduction of a CPD. loaded between May and July 2012. Of these, 129 videos con-
One of the largest challenges in the evaluation of CPD tained sufficient viewable information to be systematically an-
effectiveness is defining the position and movement of the alyzed and coded to form the rollover event database. Each
rider(s) during an accident sequence. ATVs, which incorpo- video was analyzed numerous times, in standard playback,
rate a saddle seat design and no occupant restraints, allow slow motion, and frame-by-frame mode to code the environ-
for a wide range of actively changing rider positions, a unique mental conditions and the rider and vehicle movements before,
and characteristic feature that adds to the utility, function, and during, and after the rollover event. The video researchers sys-
performance of ATVs. The relative freedom of an ATV rider tematically coded the videos into 50 data fields across 5 gen-
also allows the rider to perform a voluntary dismount of the eral topics of interest: (1) rider and vehicle characteristics, (2)
ATV at the onset of a rollover scenario. This greatly increases pre-rollover operating environment and driving behavior, (3)
the challenge in identifying the injury risk to an ATV rider in kinematics of the dynamic rollover event, (4) rider response
a given rollover scenario and in determining whether a CPD to rollover event and outcome positions, (5) and rider injury
would be a net benefit. For example, in a simple rearward outcome (Table 1).
rollover, the rider could successfully dismount and separate Injury outcome was categorized, broadly, as minimal or no
from the vehicle at the loss of control or could maintain the injury or possible or confirmed injury based on the observed
riding position and could be injured when the ATV becomes rider response during and after the rollover event, words spo-
inverted and the rider is between the ATV and ground. CPDs ken by persons in the video, and review of the video captions
and other rollover protective structures have been evaluated and participant comments which in some cases also indicated
using both physical testing and computer simulations (Pizialli the presence or absence of injury. Table 2 lists the injury out-
et al. 1993; Richardson et al. 2013; Snook 2009; Zellner et al. come coding descriptions. The estimated maximum vehicle
2008, 2013). These analyses either do not incorporate any rider travel speed in the seconds preceding the rollover and at the
or rely upon a passive crash dummy or a computer-simulated onset of roll were categorized as less than 3 mph, 3–10 mph,
passive crash dummy to evaluate potentially injurious con- or greater than 10 mph, based on the vehicle movement rel-
tact between the ATV and the rider. Clearly, a more complete ative to the operating environment. For reference, the coder
understanding of the active rider’s positioning and range of considered that 3 mph is an average person’s walking speed
movement is important in predicting the mechanisms and fre- and 10 mph is a moderate to fast running speed.
quency of injury in ATV rollover scenarios and the possible Past studies have shown that the vast majority of accidents
effectiveness of CPD designs. involve unsafe warned-against driving conditions and rider be-
In an effort to address the current lack of definition of ATV haviors. Unsafe ATV operating environments and operator be-
and associated rider dynamics, we have collected, systemati- haviors are defined as behaviors that the ATV Safety Institute
cally categorized, and analyzed a diverse group of real-world warn against (Specialty Vehicle Institute of America 2007).
192 Van Ee et al.

Table 1. ATV rollover video data categories by topic

Rider and vehicle characteristics—Gender, age, safety gear, number of occupants, ATV type, ATV payload
Pre-rollover operating environment and driving behavior—Terrain, environment/setting, vehicle speed, vehicle travel orientation, extreme risk-taking behaviors (e.g.,
jumping, stunts/wheelies, excessive speed, traversing too close to obvious hazard, driving on steep terrain, rolling backwards)
Rollover event—First event, terrain type, terrain composition, loss of control factors, direction roll onset, primary roll factors, rider roll factors, environment roll
factors, speed at onset of roll, travel direction at onset of roll, first vehicle ground contact, number of rolls, roll trajectory, increasing kinetic energy during roll,
vehicle trajectory into hazard area
Rider response to rollover event and outcome positions—Response at loss of control, response at onset of roll, separation outcome, separation trajectory, separation
direction relative to ATV, separation direction relative to slope, significant contact event, duration of contact, nature of contact, extrication, multiple contact events,
ATV surface of contact(s)
Rider injury outcome—Rider injury status, injured body area

This includes driving an ATV on a steep hill, driving on a videos. While collecting observational data on a single video,
public road, driving too fast on rough slippery terrain, turning researchers had instant access to previously coded videos and
an ATV at high speeds, rolling backwards on a hill, or a rider data coding. This helped insure uniform and comprehensive
or passenger not wearing a helmet and other protective safety coding within and between the selected videos. In addition,
gear (e.g., gloves, goggles, boots). Observed warned-against the research team collaborated regularly for multiple reviews
behaviors were coded into the database. Operating an ATV of the videos, as needed, for iterative adjustments to the coded
while under the influence of alcohol or other drugs is another fields. These quality controls, built into the data collection
significant warned-against behavior. However, this research, process, were supplemented with rigorous post–data collec-
reliant primarily on video, could not determine whether the tion validation of the logic and consistency between the ATV
rider was operating the ATV while under the influence of al- rollover video data fields. Using this process, most videos were
cohol or other drugs. viewed multiple times, often by more than one researcher, to
One of the primary objectives of this research was analyzing refine and edit each video’s coded values. Finally, application
the ATV rider’s response and positioning prior to, at the onset, of the database is demonstrated by evaluation of the possible
and throughout the rollover sequence. The ATV rider’s physi- influence of a CPD on injury and noninjury scenarios.
cal behaviors were observed and coded into 2 active response
categories: active dismount attempt or active stay-in-position
attempt. An active attempt to dismount was coded if, for ex- Results
ample, the rider tried to dismount from ATV by releasing
handlebars, swinging leg over the seat, or pushing off of the All 129 rollover videos involved a 4-wheel ATV, with most
ATV. An active attempt to stay in position was coded if, for designed as utility or sport ATVs; 2 ATVs were small, designed
example, the rider was intentionally gripping the handlebars for youth. Thirty-one ATVs carried a payload, most on a
tightly, hunching onto the ATV, or actively keeping feet in po- rear rack. None of the ATVs were equipped with a CPD. A
sition. Other riders were coded as having a passive response majority of riders wore a helmet (73%), with about half of
when, for example, the onset of the rollover occurred quickly those 94 riders also wearing other safety gear (e.g., boots,
and the operator did not have time to react or there were none gloves, goggles). Almost all of the riders appeared to be males
of the observed response characteristics of the active rider (97%) riding without a passenger (98%). Four riders (3%) were
responses. youth, 15 years old or younger, and 2 of these riders were on
ATV rollover video analyses and data collection were con- a youth-sized ATV, one on a utility ATV and one on a sport-
ducted using a structured set of observational and coding pro- styled ATV.
tocols. Four researchers were involved in the video data col- Based on the initial roll direction at the onset of vehicle roll,
lection: a supervisory coder, 2 primary coders, and a database 47% of the roll events were classified as side rolls, 44% were
design and check coder. After the videos were viewed and se- classified as rear, and 9% were classified as forward rolls. Over
lected, the initial data collection protocol was tested on about 80% of all rolls occurred on sloped terrain. In rear rolls, the
20 videos. These videos were analyzed by 2 coders, individually ATV rider was traveling uphill at the onset of roll for 95% of the
and collaboratively. Based on this initial work, the data col- events. In forward rolls, the direction of travel at the onset of
lection instrument and protocols, including a data dictionary, roll was equally split between downhill (50%) or level ground
were refined and implemented on the analysis of the remaining (50%). Side rolls occurred in many travel directions—uphill

Table 2. Injury coding

Minimal or no injury
No injury beyond abrasions, scratches, and/or bruises based on images/text, no or minimal injury based on postevent rider response
Possible or confirmed injury with ATV contact
Possible—Ambiguous/no rider response recorded, dynamics indicate a possible injury scenario, or both postevent rider response and video indicate a possible injury
Confirmed—Rider response indicates injury or injury indicated by video and/or text
Possible or confirmed injury unrelated to ATV contact
Possible—Rider response or other images/text indicate possible injury unrelated to ATV contact
Confirmed—Injury indicated by images/text unrelated to ATV contact
ATV Rollover Events 193

Fig. 1. Number of rolls per event (129 events).

(38%), level ground (27%), traversing a side slope (22%), and

Fig. 2. a. Injury outcome (129 events). b. Roll direction for riders
downhill (13%).
injured by ATV contact. Of the 129 events, 16% or 20 riders were
The speed at the onset of roll was low, with 86% of the
injured by ATV contact.
rolls occurring at speeds of 10 mph or less and 53% occurring
at travel speeds of less than 3 mph. Though the speed at the
onset of roll was low, speeds just prior to roll were somewhat occurred on sloped terrain, the primary contributing factor
higher, particularly in hill-climbing scenarios, where the ATV’s or trigger for many of the rollovers was rider input (46%; e.g.,
speed gradually decreased during the climb until the onset of braking, throttle, steering), closely followed by terrain effects
rollover. Over 60% of the events resulted in one roll or more (43%), including local features such as a bump or ridge, with
of the ATV and 33% of the events resulted in greater than 2 many events occurring while driving up a steep hill.
rolls (Figure 1). The relatively high number of rolls despite the There was no apparent rider injury for 79% of the 129
relatively low speeds at the onset of roll are reflective of the events; 16% resulted in injury due to ATV contact, and 5%
potential energy of the relatively steep terrain under which a resulted in injury unrelated to ATV contact (Figure 2a). A
majority of the roll events took place. Though 80% of the rolls majority of riders (55%) who were injured by ATV contact

Table 3. Warned-against unsafe driving conditions and rider behaviors (n = 129 events)a
Events Percentage Observations Percentage

Unsafe driving conditions 110 85

Riding on hill with more than 25◦ slope 95 74
Riding on an angle on steep hill (oblique to grade) 31 24
Riding over large obstacles (rocks, fallen trees) 8 6
Riding in water more than footrest depth 7 5
Riding on public road 3 2
Unsafe driving behaviors 71 55
Rolling backwards on hill 31 24
Driving too fast on rough, loose or slippery terrain 27 21
Exhibition driving (jumps, wheelies, or other stunts) 19 15
Using throttle or brake while rolling backwards 7 5
Turning at high speed 6 5
Accelerating over a hilltop 1 1
Unsafe rider behaviors 76 59
Operating ATV while not wearing safety gearb 73 57
Operating without feet on footrest and/or hands on handlebars 18 14
Operating ATV while carrying passengers (single-seat ATV) 2 2
Total (one or more warned-against riding conditions or behaviors) 129 328
Average number of warned against behaviors per event 2.53
aWarnings indicate a hazardous situation that, if not avoided, may result in death or serious injury. All of the 129 events had at least one warned-against

operation or behavior, with an average of 2 to 3 per roll event. As such, numbers and percentages within each category do not sum to 129 total events or 100%.
bSafety gear: helmet or gloves, protective clothing, boots, and goggles.
194 Van Ee et al.

Table 4. Prevalence of warned-against unsafe driving conditions

and rider behaviors (n = 129 events)
Most events had multiple warned-against behaviorsa Number Percentage

Observed none of the warned-against categories 0 0

Observed one of the 3 warned-against categories 31 24
Observed 2 of the 3 warned-against categories 67 52
Observed all 3 of the warned-against categories 31 24
Total 129 100
aThree categories of warned-against behaviors: (descriptions of the general
categories and specific behavior types listed in Table 3): Warned-against op-
erating environments (5); warned-against driving behaviors (6); and warned-
against rider behaviors (3).

were injured in rear roll scenarios (Figure 2b). Though rear

rolls resulted in the largest number of riders injured from ATV
contact, Figure 3 shows that forward rolls had the highest rate
of ATV contact injuries (25%), followed by rear rolls (19%)
and side rolls (10%).
Active dismount of the ATV was a commonly employed
strategy, with 63% of the riders attempting an active dismount
resulting in successful separation from the ATV in 72% of the
cases where a dismount was attempted. The overall injury rate
for riders attempting an active dismount was 15% and the in-
jury rate for riders not attempting an active dismount was 32%.
At least one warned-against behavior was observed in each
of the 129 ATV rollover events, with an average of 2 to 3
warned-against operations or behaviors per rollover event
(Tables 3 and 4). Most of the ATVs were operating in un-
safe conditions (85%), including riding on a hill with more
than a 25◦ slope or riding at an oblique angle on a steep hill.
Many riders were driving in an unsafe manner (55%), includ-
ing rolling backwards down a hill or driving too fast on rough,
loose, or slippery terrain. Fig. 4. Forward roll case study (https://www.youtube.com/
As depicted in Figure 2, 20 injurious scenarios were identi- watch?NR=1&v=ogjbfsMbLl4). The left-side frame-grab video
fied as a result of ATV–rider contact. In one of these scenarios, sequence shows a forward roll scenario where the rider avoided
the ATV clearly came to rest in the 180◦ inverted position upon injury by an active dismount. Using an oriented 3D model with
the driver’s upper thigh and hip. It is unclear whether a CPD a rear-mounted CPD and superimposing images, the second se-
would have substantively changed that minor injury outcome. quence on the right demonstrates that the CPD would likely have
directly contacted the rider in the lower middle back during his ac-
tive separation if the rider’s dismount in the presence of the CPD
would have followed the same trajectory as when no CPD was
present. The presence of a CPD in this scenario could potentially
result in significant injury to an otherwise uninjured rider.

There were no injury cases that demonstrated a risk for com-

pressional asphyxia with the ATV oriented in the 180◦ inverted
position resting upon the rider’s chest. The 180◦ inverted as-
phyxia scenario is a specific risk mode that CPDs are aimed to
reduce (Snook 2009). Further analysis using techniques such
as physical testing or computer simulation may be instruc-
tive for determining the CPD’s influence on the chaotic ATV
roll dynamics and resulting injury outcome for cases where
injurious ATV–rider contact occurs following CPD–ground
interaction.
Figures 4 and 5 depict 2 case studies identified in the video
database that did not result in rider injury but may result
in unintended consequences if a CPD is implemented. The
Fig. 3. Injury outcome by roll direction (129 events). time of interest in these case studies focuses on the ATV–rider
ATV Rollover Events 195

Fig. 5a. Rearward roll case study (https://www.youtube.com/

Fig. 5b. Rearward roll case study. These demonstrations show the
watch?v=qLddW8Iy2RY). This freeze-grab video sequence
potential of a CPD to obstruct and entangle a rider during a rear
shows a rearward roll scenario where the rider avoided injury
dismount attempt. The presence of a rear-mounted CPD in the
by an active dismount over the middle rear of the ATV. The low-
scenario shown in Figure 5a may have obstructed and possibly
est and enlarged photo shows the mid-rear dismount path used
entangled the rider in a manner similar to that demonstrated, po-
by the rider to separate from the ATV.
tentially resulting in an unsuccessful separation and subsequent
injury.

interactions prior to the time of CPD–ground interaction;

thus, the influence of the CPD on ATV roll dynamics in this Discussion
time frame would be insignificant. Figure 4 demonstrates a
forward roll scenario where the rider dismounts over the front This is the first data set we are aware of containing detailed
of the ATV and narrowly avoids postdismount contact of the ATV and rider kinematic data of real-world ATV rollover sce-
tumbling ATV. Introduction of a typical rear-mounted CPD narios. Use of Internet video to quantify and evaluate pro-
demonstrates that the CPD trajectory would have intersected posed injury mechanisms has been previously reported by
with the separation trajectory taken by the rider. As demon- Kwon et al. (2010), who used YouTube videos of actual an-
strated in the figure, if the rider would have employed the kle fracture events to correlate ankle injury mechanisms with
same dismount technique in the presence of the CPD, the fracture patterns. In the application of ATV rollover, these
rider would have been directly impacted in the lumbar region videos offer a unique and previously unavailable way to quan-
by the rear-mounted CPD as the ATV rotated forward. This tify the range of rider dynamics in rollover events. Though
direct contact could have potentially resulted in injury to a some of these videos demonstrate injurious outcomes, a ma-
previously uninjured rider. Figure 5a demonstrates a second jority of the scenarios demonstrate little or no injury to the
scenario where the addition of a CPD may result in an ad- rider despite being involved in an ATV rollover event. Though
verse outcome. In this case the rider dismounts the ATV to there may be some selection bias inherent in which videos are
the rear as the ATV pitches up near the apex of a hill. Figure posted on YouTube (example: ATV users who are frequent
5b demonstrates how a particular CPD design may obstruct, users of YouTube, ATV events that are recorded and may in-
hinder, or entangle a rider when attempting a rear dismount clude stunts or “showing off” but likely include little if any
on a pitched up ATV. Integration of a CPD in these scenarios of day-to-day routine farming-style tasks, events that are not
results in the potential for rider injury from either direct con- emotionally sensitive and less likely to include serious injury
tact with the CPD (forward roll scenario) or by obstructing the and fatal scenarios), the identified rollover events represent a
rider’s ability to actively dismount from the vehicle (rearward diverse set of conditions, with the vast majority of the rollover
roll scenario). This analysis was performed using one type of events occurring on steep terrain including local terrain fea-
commercially available CPD (http://www.quadbar.com.au/), tures with significant grade. Though video-documented events
but similar methods could be used to analyze the influence of have inherent limitations, they also offer distinct advan-
other historical and current CPD designs. tages over traditional ATV accident data composed of many
196 Van Ee et al.

unwitnessed events, collected after the fact, and often with self- investigate potential unintended consequences. Relying only
reported information. The videos provide information that is on injury and fatality databases severely limits the reliability of
not compromised by, for example, ATV rider perception and any analysis of a device’s overall effectiveness. These adverse
recall accuracy or the tendency to not disclose high-risk behav- outcome databases generally do not have sufficient rider and
iors. Video events offer an objective and accurate time history ATV kinematic data to determine whether a particular device
of rider and vehicle dynamics. Generalization of this study’s may have made a difference in any given case. Additionally,
results and the application of individual case studies reported even if there were sufficient data, consideration of only serious
herein would be further strengthened by quantitative data on injury and fatality data may lead to an overestimate of the
the total number of ATV rolls that occur worldwide, the rel- benefits of a strategy and an underestimate of the unintended
ative frequency of roll types, and the frequency of injury and consequences because almost any significant change will
relative severity of injury for roll events. Unlike for automo- either make no difference or likely improve outcomes for
bile accidents for which the National Automotive Sampling the seriously injured or fatal groups. Prior to the responsible
System is available (http://www.nhtsa.gov/NAS), there is not implementation of a safety device on an existing product, a
currently a statistically based representative sample of rollover comprehensive approach should demonstrate overall benefits,
events to refer to when trying to determine how representative documenting unintended consequences across the broad
the current sample is with reference to accident conditions and spectrum of rollover events, variations in vehicle design, rider
injury outcome. The rollover events included in this database responses, and terrain conditions. These new and relevant
do represent a wide range of real-world conditions with rolls data and analysis techniques should be used to develop and
occurring on varied slope, terrain, and speed with a variety guide effective implementation of injury mitigation strategies.
of ATV models and riders of varied skill resulting in a broad
range of roll dynamics that have never previously been sys-
tematically collected and reported. References
Previous studies have evaluated the performance of CPDs Consumer Product Safety Commission. 2011 Annual Report of ATV-
(Richardson 2013; Snook 2009; Zellner et al. 2008, 2013). Related Deaths and Injuries. 2013. Available at: http://www.cpsc.
Snook (2009) reported the effect of a CPD on ATV roll kine- gov//Global/Research-and-Statistics/Injury-Statistics/atv2011.pdf.
matics under physical testing conditions where the ATV roll Helmkamp JC, Marsh SM, Aitken ME. Occupational all-terrain vehi-
was initiated using a gravity-driven sled incorporating a cat- cle deaths among workers 18 years and older in the United States,
apult release to induce rotation. Because no rider model was 1992–2007. J Agric Saf Health. 2011;17(2):147–155.
Kwon JY, Chacko AT, Kadzielski JJ, Appleton PT, Rodriguez EK. A
incorporated in the testing (such as a crash dummy), only in-
novel methodology for the study of injury mechanism: ankle fracture
formation on vehicle kinematics could be ascertained. More analysis using injury videos posted on YouTube.com. Orthop Trauma.
recently, Richardson et al. (2013) reported the results of a 2010;24:477–482.
computer simulation using a PC-crash model of an ATV that Lower T, Herde E, Fragar L. Quad bike deaths in Australia 2001 to 2010.
was developed in part using Snook’s (2009) vehicle kinematic J Health Safety Environ. 2012;28:7–24.
data. The simulated rider was positioned and attached to the Piziali RL, Ayres TJ, Paver JG, Fowler G, McCarthy RL. Investigation
ATV seat, foot platform, and handlebars using tension only of the Net Safety Impact of an Occupant Protection System From All-
terrain Vehicles. Warrendale, PA: Society of Automotive Engineers;
tethers. No active action of the rider was incorporated and the 1993. Paper SAE 930208.
model did not evaluate protective benefits of a helmet. Based Rechnitzer G, Grzebieta RH, McIntosh AS, Simmons K. Reducing all
on their simulations, the authors concluded that “considera- terrain vehicle injuries (ATVs) and deaths—a way ahead. In: Proceed-
tion should be given to fitting either: CPD, ROPS or ROPS ings of the 23rd International Technical Conference on the Enhanced
with rider restraint” (p. 1). An extensive computer simulation Safety of Vehicles (ESV), Seoul, Republic of Korea, May 27–30,
study reported by Zellner et al. (2013) incorporating a simu- 2013.
lated rider found that the addition of a CPD created injury Richardson S, Sandvik A, Jones C, et al. Simulation and analysis of quad
bike rollovers using PC-CRASH to evaluate alternative safety systems.
and fatality risks that were greater than the possible benefits In: Proceedings of the 23rd International Technical Conference on the
and recommended against the use of a CPD. To date, however, Enhanced Safety of Vehicles (ESV), Seoul, Republic of Korea, May
these studies and methods have not accounted for the move- 27–30, 2013.
ments of an active rider, including active separation, which Snook C. An Assessment of Passive Roll over Protection for Quad Bikes.
this study shows is an important element in understanding the Brisbane, Australia: University of Southern Queensland; 2009.
injury dynamics of ATV rollover events. Specialty Vehicle Institute of America. Special Report Summer
2007. 2007. Available at: http://www.atvsafety.org/sviapressreleases/
ATV rollover safety intervention strategies such as CPDs
SVIA Special Report.pdf.
that change ATV exterior geometry, structure, and inertial Williams AF, Oesch SL, McCartt AT, Teoh ER, Sims LB. On-road all-
properties and potentially limit or inhibit rider separation terrain vehicle (ATV) fatalities in the United States. Journal of Safety
should be thoroughly evaluated for possible beneficial Research. 2014;50:117–123.
outcomes as well as for potential unintended consequences Zellner JW, Kebschull SA, Van Auken RM. Evaluation of Injury Risks and
across the full range of reasonable use environments, includ- Benefits of a Crush Protection Device (CPD) for All-terrain Vehicles
ing real-world rollover scenarios, before being implemented. (ATVs). Warrendale, PA: SAE; 2013. SAE Technical Paper 2013-32-
9173.
The evaluation of the unintended consequences of CPDs for Zellner JW, Van Auken RM, Kebschull SA, Munoz S. Injury risk–benefit
real-world rollover scenarios has been extremely limited due analysis of rollover protection systems (ROPS) for all-terrain vehicles
to the dearth of information on rider responses, ejection paths, (ATVS) using computer simulation, full-scale testing and ISO 13232.
and determinants of injury during ATV roll events. The 2 case In: FISITA World Automotive Congress Proceedings, Munich, Ger-
studies are an example of how video analysis can be used to many, September 14–19, 2008.

View publication stats