Motorcycle fuel tanks and pelvic fractures: A motorcycle fuel tank syndrome

ABSTRACT Objective: Pelvic injuries are a serious and commonly occurring injury to motorcycle riders involved in crashes, yet there has been limited research investigating the mechanisms involved in these injuries. This study aimed to investigate the mechanisms involved in pelvic injuries to crashed motorcyclists. Method: This study involved in-depth crash investigation and 2 convenience-based data sets were used. These data sets investigated motorcycle crashes in the Sydney, Newcastle, and Adelaide regions. Participants included motorcycle riders who had crashed either on a public road or private property within the study areas. The mechanism of injury and the type of injuries were investigated. Results: The most frequent cause of pelvic injuries in crashed motorcyclists was due to contact with the motorcycle fuel tank during the crash (85%). For riders who had come into contact with the fuel tank, the injury types were able to be grouped into 3 categories based on the complexity of the injury. The complexity of the injury appeared to increase with impact speed but this was a nonsignificant trend. The pelvic injuries that did not occur from contact with the fuel tank in this sample differed in asymmetry of loading and did not commonly involve injury to the bladder. They were commonly one-sided injuries but this differed based on the point of loading; however, a larger sample of these injuries needs to be investigated. Conclusion: Overall improvements in road safety have not been replicated in the amelioration of pelvic injuries in motorcyclists and improvements in the design of crashworthy motorcycle fuel tanks appear to be required.


Background
Motorcyclists are significantly overrepresented in crash statistics, with motorcyclists accounting for 23% of road traffic deaths around the world (World Health Organization 2013). Though there has been an overall decrease in road traffic casualties in New South Wales (NSW), Australia, in recent years, the number of serious injury and fatalities for motorcycle riders has been increasing. Since 2006, there has been an overall increase of 6% in motorcycle casualties, and the casualty numbers for other road users have decreased by 4% (Transport for NSW 2012). Increases in casualties among motorcyclists have also been reported in the United States (National Center for Statistics and Analysis 2014) and Europe (Organization for Economic Cooperation and Development/European Union 2014). Currently in Australia, motorcyclists account for 22% of serious casualties on Australian roads. In 2013, 4,668 motorcyclists were killed and 88,000 were injured in the United States, and in Europe, motorcyclists account for over 20% of road transport accident deaths. Motorcyclists therefore present a significant and growing health burden.
Pelvic injuries are known to be a common injury among motorcyclists involved in crashes, with approximately 13% of crashed motorcyclists experiencing a pelvic injury , and motorcycle crashes being the most common cause of traumatic fracture of the pelvis (Rockwood et al. 2010). Pelvic injuries can be defined as any fracture of the bones of the pelvic ring; soft tissue injury to the lower abdomen, perineum, groin, and scrotum/testicular area; or injury to any internal organ within the pelvic cavity. These injuries present a significant burden to the health system and society in general. In addition to the immediate costs associated with the treatment of these injuries, many pelvic injuries are associated with poor long-term outcomes. Pelvic fractures often lead to chronic pain problems and ongoing reductions in quality of life (Gerbershagen et al. 2010;Pohlemann et al. 1996). However, there have been no measures implemented to mitigate pelvic injury in motorcyclists. This is in stark contrast to other regions of a rider's body where helmets provide effective protection of the head (Liu et al. 2008); protective clothing provides effective protection of soft tissue (de Rome et al. 2011); and motorcycle and airbag technology is being developed to provide protection to the thorax (Thollon et al. 2010). This is possibly related to the fact that to date there has only been limited research into pelvic injury mechanisms and potential measures to mitigate pelvic injury and pelvic injury severity.
The aims of this study are to determine the nature and mechanisms of pelvic injury among crashed motorcyclists, and to identify factors influencing their occurrence. This was achieved using prospective review of data collected from a cohort of crashed motorcyclists collected in NSW and South Australia.

Methods
This study used data from 2 sources. The first source (n = 105) involved motorcycle crashes in NSW and was collected prospectively during a 2-year (2012)(2013)(2014) in-depth investigation of nonfatal motorcycle accidents occurring on public roads or on private property within the study area. In summary, eligible participants were motorcyclists aged 14 years and older who had been admitted to 1 of 4 Sydney hospitals or one regional hospital following a crash. All of these hospitals have specialist trauma units, and patients were examined and treated by members of those units. Motorcyclists were recruited by research nurses and data were collected through rider interviews, patient medical records, motorcycle inspections, clothing inspections, crash site inspections, and police reports. This included detailed descriptions of injuries that were then coded using the Abbreviated Injury Scale (AIS), 2008 revision (Association for the Advancement of Automotive Medicine 2005) by certified coders. The Injury Severity Scores (ISS) were calculated using these scores. Injury severity scores were categorized as nonsevere (ISS < 15) and severe (ISS ࣙ 15; Copes et al. 1988). The motorcycle inspection included a detailed examination of the motorcycle to look for evidence of rider contact such as deformation and/or other signs of visible damage ( Figure 1). The fuel tank angle was measured, and this was the angle between the horizontal and the highest point of the fuel tank and is depicted in Figure 2. Protrusions on the fuel tank were also noted. Rider interviews were generally conducted within days of the crash, and vehicle and scene inspections were generally conducted within 2 weeks of the crash. Impact speeds were estimated based on rider  reports and police records. Each injury was assigned a contact description based on the information collected. Each case and data collected were then reviewed by an interdisciplinary panel including motorcycle specialists, engineers, injury specialists, and police officers to confirm the mechanism of each pelvic injury. This was determined through damage to the fuel tank providing evidence of contact, through rider reports, or through the fuel tank being the most likely contact partner during the crash.
The second source was an ongoing (since 2006) prospective on-scene in-depth investigation of both nonfatal and fatal motorcycle crashes occurring within a 100-km radius of metropolitan Adelaide, South Australia. It included any type of road crash that resulted in at least one crash participant being transported to hospital. This source differed from the first in its data collection protocol. The crash investigation team was notified by an automatic paging service every time the South Australian Ambulance Service was called to a crash, so vehicle and scene data were collected within hours of the crash. However, rider interviews were sometimes not obtained and the clothing was not inspected. Cases were reviewed by the crash investigation team consisting of engineers, behavioral scientists, and nurses to determine the causation scenario for the rider and the mechanisms of injury.
Data for all South Australian cases were combined with the NSW data to provide the final data set consisting of 155 cases. All cases where the rider had suffered a pelvic injury were then extracted for further examination of the type and mechanism of pelvic injury.
Ethical approval was given by a lead NSW Department of Health Ethics Committee, the University of New South Wales Human Ethics Committee, and the University of Adelaide Human Research Ethics Committee.

Data analysis
Rider and crash characteristics, including age, injury severity, height, weight, body mass index (BMI), estimated impact speed, and gender, were examined, and significant differences in these characteristics between riders who did and did not suffer pelvic injuries were identified using univariate binary logistic regression. Factors that were found to have significance (P value) of less than .25 in the univariate analysis were entered into a binary logit model as recommended by Hosmer and Lemeshow (2000). All cases where the rider had suffered a pelvic injury were then extracted for further examination. The causation scenario for the rider was investigated and the mechanism of pelvic injury was deduced through the data collected during crash investigation. A consensus opinion on the source and mechanism of the injury was obtained from the expert panel reviews.
Patterns of pelvic injury for each rider were categorized in 2 ways. The first injury groupings categorized pelvic injury by the location of injury; that is, external soft tissue, internal pelvic organ, and bony fractures or dislocations; the second categorized injury according to the complexity of the injuries because complexity is more likely to be associated with loading conditions than simple AIS coding.. Differences in rider and crash characteristics (rider age, rider BMI, impact speed, fuel tank angle, collision partner, and motorcycle type) between cases in each category were then analyzed using analysis of variance.

Results
The characteristics of the crash-involved riders (n = 155) are presented in Table 1. The average age of riders in this sample of crashes was 37 years, with a range of 15-80 years. Most of the riders were male (94.8%). The average BMI fell into the overweight range, with 67.4% of riders having a BMI considered to be overweight (30.4% overweight, classified as a BMI between 25 and 29.9) or obese (37% obese, classified as a BMI of equal to or greater than 30). Both the mean and median impact speeds were 60 km/h and the majority of crashes occurred at speeds of 60 km/h or less (65%). Pelvic injury occurred in 33 of the participants (21.3%).
Univariate logistic regression analysis revealed that the rider's age was significantly higher in riders who sustained a pelvic injury than those without a pelvic injury, with the odds of injury Table . Rider characteristics for the total sample of riders involved in the study (n = ) and differences in rider characteristics between riders who suffered pelvic injuries and those with no pelvic injury (outcome pelvic injury).

Characteristic
Mean (SD) increasing by 3% for every additional year (odds ratio = 1.03, 95% confidence interval, 1.003-1.055, P = .03). Additionally, the overall ISS of riders who had sustained a pelvic injury was significantly higher than those with no pelvic injury, with riders suffering a pelvic injury having a 5 times higher odds of having an ISS in the severe range (odds ratio = 5.72, 95% confidence interval, 2.46-13.30, P < .0001). Because the pelvic injury was commonly the injury with the maximum AIS, ISS was not explored using multivariate analysis. Two variables, age (P = .03) and BMI (P = .17), met the criteria P < .25 for inclusion in multivariate analysis. Controlling for BMI, age was no longer significantly associated with whether or not pelvic injury occurred (odds ratio = 1.02, 95% confidence interval, 0.99-1.045).

Pelvic injury mechanisms
A summary table describing the motorcycle type, fuel tank characteristics, pelvic injury descriptions and severity, overall ISS, and deduced mechanisms of injury and supporting evidence is provided in Appendix A (see online supplement).
Of the 33 pelvic injuries sustained by the riders in this study, 28 (85%) were found to have occurred due to direct contact of the pelvis with the motorcycle fuel tank, and this primarily involved crashes in which the motorcycle impacted with another moving vehicle (85.7%). Of the riders who reported to have contacted the fuel tank, there was definitive evidence of contact from damage and markings observed on the motorcycle in 17 cases. In the remaining 11 cases, there was no definitive evidence on the fuel tank but contact was judged likely to have occurred based on rider reports coupled with results of the expert panel review of the data collected through crash investigation. That is the estimated rider kinematics based on crash orientation, object struck, and likely precrash rider posture suggested that contact with the fuel tank was the most likely source of the pelvic injury. The remaining 5 cases occurred from a variety of contact sources, including impact with the road surface, impact with roadside furniture, force transfer through the feet, and direct contact with another moving vehicle.
The most common types of pelvic injury to the riders who contacted the fuel tank were fractures (67.9%) and external injuries (67.9%). Bladder injuries also occurred in 4 cases (14.3%) and involved haematomas (n = 2) and laceration and herniation (n = 2). Based on the complexity of the injuries, there were 3 distinct groups of pelvic injury. These groups are depicted in Figure 3. Group 1 consisted of riders who had suffered bladder injuries or soft tissue injuries with no fractures present. Group 2 consisted of riders who had suffered a pelvic fracture to the anterior portion of the pelvis. These fracture types included bilateral (n = 2) and unilateral (n = 5) pubic rami fractures as well as subluxation or fracture of the pubic symphysis (n = 3). These fractures resembled anteroposterior type 2 pelvic injuries (Rockwood et al. 2010). Group 3 injuries were again of increasing complexity, and these injuries involved riders who had fractures in the anterior and posterior portions of the pelvis where the force has been transmitted through the whole pelvis. Some examples of the types of injuries seen are shown in Figure 3. These fractures resembled anteroposterior type 3 pelvic injuries (Rockwood et al. 2010). The severity of the pelvic injury arising from contact with fuel tank was frequently AIS 2 or greater (82% of riders).
Differences in the characteristics of the riders and the crashes they were involved in for the 3 injury groups are shown in Table 2. The complexity of injury appeared to increase with age and impact speed, but these associations did not reach significance. The only other characteristic that differed between the injury groups was the type of crash, with 3 of the 4 singlevehicle crashes (impacts with stationary vehicle and loss of control crash) occurring in group 3. Fuel tank angle did not vary significantly between injury groups.

Other pelvic injury sources
From the total number of participants involved in the study, there were 5 cases (3%) in which pelvic fractures occurred through causes other than contact with the fuel tank. One of these involved a scooter rider who sustained a pelvic fracture that was described by the radiologist as a vertical shear injury. Vertical shear injuries are vertically orientated and indicate inferior-superior loading (Young et al. 1986). From the crash investigation it was deduced that the injury occurred via loading of the leg when the rider's feet or knees impacted the ground. This is similar to injuries seen to patients who have a fall from a height when the load is applied as the feet strike the ground. Another involved direct contact between a rider and another vehicle when the rider was t-boned by that vehicle. The remaining 3 cases involved riders who lost control of their motorcycles.
In the first of these, pelvic injury occurred when a rider impacted a guard rail. In the second, a rider was run over by a truck and in the third case pelvic injury occurred when the rider struck the roadway. As shown in Figure 4, the pattern of injury was generally different from that seen among those riders who sustained injury following contact with the petrol tank.

Discussion
The results of this study identified distinct patterns of pelvic injury that are consistent with injuries due to anteroposterior loading of the pelvic region as described in Rockwood and Green's Fractures in Adults (Rockwood et al. 2010). In this case, anteroposterior loading was caused by contact with the fuel tank among motorcyclists. This type and mechanism of injury appears to be the predominant pelvic injury occurring among seriously injured motorcyclists with pelvic injury. Furthermore, the complexity of this injury appears to be associated with an increasing impact speed, yet this result needs to be examined using a larger sample size. This may be the first detailed study of the pattern of pelvic injury associated with motorcycle fuel tank contact. There have been a small number of reports in the literature indicating the fuel tank as a source of injury for motorcyclists since 1980 (de Peretti et al. 1994;Ouellet and Hurt 1981;Ruter et al. 1980). In earlier investigations, the profile of the fuel tank would have differed substantially from current models of motorcycles. For example, in earlier models of motorcycles, the fuel tank tended to gradually rise from the seat, whereas in current motorcycles, the seat is lowered so there is an abrupt vertical rise of the fuel tank from the seat. Our observations that the fuel tank is the primary source of injury to the pelvic region among motorcyclists in the present study aligns with observations reported by Ouellet and Hurt in 1981. These authors examined groin injuries sustained by 117 riders and found that 88% of these injuries occurred following contact with the fuel tank. This mostly occurred in frontal crashes, and the severity of the groin injury increased with impact speed. The similarity of our observations made more than 30 years after those by Ouellet and Hurt (1981) highlight a distinct lack of attention to this problem. Though reporting that injuries occurred from the fuel tank, this study did not examine the types of injuries and fracture patterns occurring from riders who contacted with the fuel tank.
Though there has been discussion in the literature over the last few decades about how motorcycle and fuel tank design might be modified to mitigate injury, contrasting theories on which designs would be most beneficial have been presented (Bothwell et al. 1973;de Peretti et al. 1994;Ouellet and Hurt 1981;Ruter et al. 1980;Wobrock et al. 2006). These include fuel tank shapes that promote ejection (Ruter et al. 1980) and completely minimize contact between the pelvic region and the fuel tank (Bothwell et al. 1973), whereas others have suggested designs to better distribute loads and minimise injurious contact (Bothwell et al. 1973;de Peretti et al. 1994;Ouellet and Hurt 1981). Most recently, in 2006, Wobrock et al. suggested that loads on the pelvis might be managed by controlling the angle of the fuel tank. The studies from which these suggestions have been drawn have a number of limitations, the primary one being a lack of clear understanding of the spectrum of injuries that need to be prevented and/or the definition of the type of loading that needs to be managed. Our definition of the spectrum of pattern of injury likely to be associated with increasing levels of anteroposterior loading to the pelvis provides an important target for more focused attention to the design of the motorcycle and fuel tank to mitigate these injuries. Furthermore, this pattern of injury suggests that the complexity of injury to the pelvis can be minimized if the energy transfer to the pelvis can be better managed.
Further work to characterize the design features of the motorcycle and the fuel tank having the greatest influence of the anteroposterior loads applied to the pelvis is required, and there is clearly potential for passive energy-attenuating technologies to be incorporated into motorcycle design and/or the fuel tank. The complete kinematics of the rider and injury outcome of other body regions also need to be considered when attempting to mitigate pelvic injuries because it may be that the fuel tank can provide some restraint and slowing down of the rider before the rider separates from the motorcycle, reducing the severity of injuries to other body regions. Wobrock et al. (2006) used computer simulations to determine that the force on the pelvis exponentially increased with the fuel tank angle, suggesting that as the fuel tank angle increases, the risk of pelvic injury may also increase. There was no difference observed in the tank angle between the injury groups here. However, our results suggest that there is a need to study this effect while controlling for impact speed. Furthermore, the simulations conducted by Wobrock et al. (2006) used a 50th percentile male riding a sports motorcycle based on a "normal" riding position. From this current sample of motorcycle riders involved in crashes, with riders on average being heavier than the 50th percentile male, it seems unlikely that the 50th percentile adequately represents the population at risk, there is therefore also a need to better understand how rider anthropometry and posture might also interact with features of the fuel tank design.
Using the very crude figures available, in Australia alone, 20% of the 6,270 people injured in motorcycle crashes every year are likely to suffer significant bony and/or soft tissue pelvic injury, and most of this appears likely to be attributable to contact with the fuel tank. This equates to an expected in excess of 1,200 cases of pelvic injury attributed to motorcycle crashes per annum in Australia. In countries like the United States this translates to more than 17,000 riders sustaining pelvic injuries every year (based on 88,000 casualties in 2013; National Center for Statistics and Analysis 2015). Furthermore, pelvic injuries can carry a high threat to life, as indicated by the AIS (Association for the Advancement of Automotive Medicine 2005). Most pelvic injuries observed in the series were AIS 2+ or greater, and riders with pelvic injury had significantly higher ISS than those with other injuries. Remembering that motorcycling is growing in popularity in most developed countries (Transport for NSW 2012), these numbers clearly highlight the potential benefit and urgent need for identifying effective ways to mitigate pelvic injury attributable to the fuel tank.
As with any study, there are a number of limitations that should be kept in mind when assessing the data obtained. The first of these is that in analysis of in-depth crash data it is impossible to be completely certain of the assigned contact source, and in motorcycle crashes, the potential for multiple contacts cannot be completed ruled out. Another possible contact point might be protrusions on the fuel tank such as the fuel tank cap, handlebars, or instrument panel. The handlebars in particular has been shown to be a frequent contact source for pelvic injuries in motorcycle crashes with Hurt et al. (1981) reporting that from a sample of 278 crashed motorcyclists, 46 pelvic injuries were caused by the handlebars as opposed to 20 from the fuel tank. However, only 4 of the motorcycles in this investigation had protrusions and 2 of these had definite signs of denting on the fuel tank, and contact with the handlebars was not identified in any cases in this study. It is possible that in the remaining 2 cases, the rider contacted the protrusion. Due to uncertainty regarding whether the rider had contacted the protrusion, this was just considered a fuel tank contact. Additionally, the injuries seen to these 2 riders were fractures to the pelvic ring. The injuries seen to riders who contacted protrusions on the tank top in the investigation by Ouellet and Hurt (1981) were lacerations of the thighs, scrotum, and penis. We are as confident as we can be that the contact sources identified in this study are accurate because they are based on evidence collected during the investigation and reviewed by multidisciplinary experts. A further limitation is that the only characteristic of the fuel tank examined in detail was the fuel tank angle. It is possible that other aspects of the tank geometry or construction might be important, and this should be explored in future studies.
As part of the study design, only injuries that were reported in the hospital medical records were taken into account. Therefore, there is some possibility that some injuries, particularly minor external injuries, were omitted in this analysis. The number of pelvic injuries reported is therefore likely a conservative estimate. The estimations of impact speed are also limited in that they are estimations with unknown accuracy. Though these estimations were made on the best available evidence, there is no easy mechanism for calculating likely speeds and in most crashes included in this analysis, investigations were conducted after the crash had occurred, with scenes investigated after the day of the crash. This means that calculating speeds based on the final positions of the rider and vehicles and/or crash reconstruction procedures cannot be performed to estimate the impact speed. Additionally, this sample is biased toward seriously injured riders and represents a convenience sample of crashes. The mean and median impact speeds in this sample appeared to be higher than those reported in the MAIDs (Association of European Motorcycle Manufacturers 2004) and Hurt, Oullet, and Wagar (1981) studies, which found median speeds of 49 and 47.9 km/h, respectively. This difference is most likely a reflection of the bias in the data collection methods, with the cases in this study being recruited from hospital, whereas the other studies involved on-scene recruitment, where even riders who did not present to hospital were involved. As a result, the cases presented in this study reflect high-severity crashes and the frequency of pelvic injuries presented here may not necessarily reflect the numbers of pelvic injuries seen among all motorcyclists involved in crashes, because most are not admitted to hospital after a crash or minor spill (Hurt, Ouellet, and Thom 1981). However, it is worthwhile noting that the cases collected in Adelaide required only one crash participant to be transported to hospital following the crash, and this did not necessarily have to be the motorcycle rider. Therefore, there was a sample of less seriously injured riders included in this investigation. Because this investigation was performed after the crash, there was not enough information on the final rider and motorcycle positions to be able to conduct crash reconstructions to obtain impact speed estimates, which contrasts with the MAIDs investigation (Association of European Motorcycle Manufacturers 2004). This means that impact speeds were based on rider reports and are likely only estimates. Finally, this study is based on a small sample size and should only be considered observational.
The study's strength is the detailed review of injury types coupled with the inspection of the motorcycle. This has allowed, for the first time, the identification of a motorcycle fuel tank syndrome involving injury to the pelvic region of increasing complexity with increasing impact speed. This level of injury detail is critical to the development of effective countermeasures to pelvic injury among motorcyclists.