The Characteristics of Injuries in Motorcycle to Barrier Collisions in Maryland

Motorcycle to barrier collisions are more serious than many other motorcycle crash modes, such as collisions with only the ground or passenger cars. In order to identify the potential need for design improvements to traffic barriers to reduce the severity of these crashes, the injuries incurred during these collisions must first be better understood. The objective of this study is to determine the type, relative frequency, and severity of injuries incurred in motorcycle to barrier crashes in Maryland. The Crash Outcome Data Evaluation System (CODES) was used to analyze motorcycle crashes in Maryland from 2006-2008. CODES links police-reported crashes to hospital data, providing detailed information about the injuries incurred during the collision. This study focused on four different crash modes for motorcyclists: single-vehicle barrier collisions, single-vehicle fixed object collisions, multi-vehicle collisions, and single-vehicle overturn collisions. The most commonly injured regions for all motorcycle crashes were the upper and lower extremities – over 70% of motorcyclists involved in the crashes analyzed suffered an injury to the upper and/or lower extremities. Motorcyclists involved in barrier collisions were 2.15 (95%CI: 1.17-3.92) times more likely to suffer a serious injury to the thorax than motorcyclists involved in overturn-only collisions. Additionally, severe lacerations were 2.26 (95% CI: 0.75-6.86) times more likely in motorcycle barrier collisions than overturn only collisions, though this was not found to be statistically significant. Daniello and Gabler 3 INTRODUCTION Motorcycle to barrier collisions are more serious than many other motorcycle crash modes, such as collisions with only the ground and collisions with passenger cars (1-9). Quddus et al. (2002) demonstrated that colliding with stationary objects increased the risk of severe injury in motorcycle crashes in Singapore (10). Likewise, Tung et al. (2009) demonstrated that the odds of serious or fatal 5 injury in motorcycle-guardrail collisions were 1.7 times higher than those in motorcycle collisions that did not involve other objects (4). Rigid, sharp surfaces have been demonstrated to cause more severe injuries in motorcycle collisions (5). Motorcyclists have a much higher fatality risk in collisions with traffic barriers than do other road users. Head injuries have been found to be the most common cause of fatality in all motorcycle crashes 10 (11-13). Bambach et al. (2011) found that the most frequently injured region was the thorax, and the head was the second most commonly injured region (13). There are anecdotal reports that motorcycle to barrier crashes may result in a very different pattern of injuries, such as amputations or severe lacerations that are rarely observed in collisions with other objects. It is important to understand these injury patterns in order to identify the potential need for design improvements to traffic barriers. 15 Unlike passenger car crashes, there is currently no in-depth investigation database for motorcycle crashes in the United States. State crash databases do not include detailed injury information, making it difficult to determine how injury patterns differ across crash types. However, the Crash Outcome Data Evaluation System links police-reported crashes to hospital records. For this study, crashes in Maryland will be analyzed to determine differences in injury patterns across motorcycle collision types. 20 OBJECTIVE The objective of this study was to determine the type, relative frequency, and severity of injuries incurred in motorcycle to barrier crashes. These injury distributions were compared to motorcyclist injury distributions in other crash modes to identify how barrier collisions differ from other collision modes. METHODS 25 The Maryland Crash Outcome Data Evaluation System (CODES) was used to analyze three years of motorcycle collisions, from 2006-2008. Data sources for the Maryland CODES include, but are not limited to, police records, EMS, emergency department, and toxicology reports (14). The CODES data is the result of linking these datasets using a probabilistic method (14). Injury data is reported in CODES using the International Classification of Disease 9 Revision 30 Clinical Modification (ICD-9-CM). The ICD-9-CM codes provide detailed injury information, but do not give a measure of injury severity, such as threat to life. The Abbreviated Injury Scale (AIS) is another coding metric used to describe injuries. AIS also reports injury severity in terms of threat to life (15), and is widely used in in-depth crash investigation databases. AIS codes rank injury severity from AIS=1 (minor) to AIS=6 (not survivable). In this study, the ICDMap-90 Program (Johns Hopkins and Tri35 Analytics, 1998) was used to map the ICD-9-CM codes to the AIS-90 codes. In a small number of cases, ICD-9-CM codes did not map directly to AIS codes. When not enough information was provided in the ICD-9-CM code to identify a unique AIS code, the AIS code with the lowest potential severity was used (16). Four categories of motorcycle crashes were analyzed in this study: crashes with traffic barriers, 40 crashes with fixed objects, multi-vehicle crashes and overturn crashes. Traffic barrier crashes involved a collision with a guardrail, construction barrier, or crash attenuator. Fixed object crashes included collisions with bridges, buildings, culverts, embankments, fences, poles, and trees. Both the barrier and fixed object crashes included in this study were limited to single-vehicle crashes. If a motorcycle struck multiple objects, e.g., a guardrail followed by a tree, the object which caused the rider injury could not be 45 determined. Multi-event collisions were therefore excluded from the barrier and fixed object analysis. The multi-vehicle crash category would include crashes between motorcycles and cars, but would exclude crashes where there was also a collision with a barrier or fixed object. Overturn crashes analyzed were Daniello and Gabler 4 likewise restricted to single-vehicle crashes. All motorcyclists included in this study were operators of the vehicle. 50 Severity of all crashes was analyzed using the maximum AIS severity score (MAIS). Serious injuries were defined as those with an AIS greater than or equal to 3. In addition, injuries were analyzed by body region to determine whether injury patterns of motorcyclists involved in barrier collisions differed from other collision types. Serious lacerations and amputations were tabulated separately to investigate concerns that the sharp edges of metal barrier posts and rail edges may lead to these types of 55 cutting injuries. Lastly, the number of fatally injured riders in Maryland CODES was compared with the number of riders fatally injured in Maryland using the FARS database. RESULTS There were 5,586 motorcycle crashes of all severity in Maryland from 2006 – 2008. The CODES data linked 2,357 of these crashes with hospital inpatient or emergency department data. The injury data 60 associated with all of these crashes was for the motorcycle operator. No motorcycle passengers were included in this study. Seven of the linked cases did not have any injury codes associated with them. There were 1,707 motorcyclists included in this study, which were divided into 4 crash categories: single vehicle barrier crashes, single-vehicle fixed objects crashes (excluding collisions with barriers), multivehicle crashes (excluding multi-vehicle collisions with barriers and fixed objects), and overturn only 65 crashes. The number of crashes of each collision type is shown in TABLE 1. The ‘Other’ category includes all crashes not falling into the 4 analysis categories, such as multi-event collisions into barriers and fixed objects. TABLE 1. Distribution of Crashes in Maryland (2006-2008) 70 Crash Type MD CODES Composition % Successfully Linked Crashes Linked Crashes All Crashes Single Vehicle Barrier 107 242 44.2% Single Vehicle Fixed Object 260 654 39.8% Multi-Vehicle 1,103 2,601 42.4% Single Vehicle Overturn Only 242 452 53.5% Other 645 1,637 39.4% Total Crashes 2,357 5,586 42.2% Not including barrier collisions Data linkage between two dissimilar datasets, e.g. police-reported crashes and hospital data, is seldom perfect. When using linked datasets, one question is how representative is the linked dataset of the overall dataset. TABLE 2 presents the distribution of police reported injury severity for all cases and for the linked subset of these cases. Only 42% (2,357 of 5,586) of police-reported crashes could be linked 75 with hospital data. However, as the linked cases required hospital admission, we expected that the linked crashes would not include property damage only cases, most minor injury cases, and many fatal cases. TABLE 2 confirms that the linked cases are biased towards injury and disabled cases, and almost entirely exclude property damage only cases. Only 27.7% of the fatal cases were linked to hospital records. Indeed, a χ test showed that there is a significant difference in the injury distributions of the linked and 80 unlinked datasets (p < 0.0001). TABLE 2. Police Reported Injury Severity in MD CODES Data for the Entire Dataset KABCO Police Reported Injury Severity % Linked Cases % Un-Linked Cases O Not Injured 5.94 33.01 C Possible Injury 18.16 16.01 B Injured 48.88 30.54 A Disabled 24.18 15.02 K Fatal 2.84 5.42 Daniello and Gabler 5 However, when the seriously injured riders likely to have been hospitalized (‘Disabled’ and ‘Injured’) are compared as shown in TABLE 3, the linked and unlinked datasets are remarkably similar. 85 A χ test showed there was no significant difference in the injury distributions of the linked and unlinked datasets (p = 0.908) in the “Injured” and “Disabled” groups. We conclude that using the linked CODES data to analyze the injury distributions of the A+B crashes is representative of the serious injuries in the entire dataset. TABLE 3. Seriously Injured Riders in MD CODES Data 90 KABCO Police Reported Injury Severity Number of Linked Cases Number of UnLinked Cases % Linked Cases % Un-Linked Cases B Injured 1,152 986 66.90 67.03 A Disabled 570 485 33.10 32.97 A + B Injured + Disabled 1,722 1,471 100 100 General characteristics of the crashes included in this analysis are given in TABLE 4. All injury severities were included for this analysis. The gender distributions were approximately the same for all collision types. Overall, 93% of motorcyclists included in this analysis were male. Maryland has a full helmet law which requires riders to wear a helmet at all times. Police reported that 81% of all 95 motorcyclists were helmeted at the time of the crash. The distribution of helmet usage was approximately the same across all collision types. TABLE 4. Composition of the Data Set Barrier Crashes Fixed Object Crashes Multi-Vehicle Crashes Overturn Only Crashes Total Total Crashes 106 260 1,101 240 1,707 Gender Male 98 234 1,041 215 1,588 Female 8 26 58 25 117 Unknown 0 0 2 0 2 Helmet Usage Helmet Used 86 225 870 202 1,383 Eye Shield Used 1 1 6 2 10 None Used 7 16 71 15 109 Unknown 12 18 154 21 205 100 The vast majority of ICD-9-CM codes were successfully mapped onto AIS codes. The maximum injury severity could not be determined in fewer than 2% of cases (27 of 1,707). When mapping the ICD9-CM scores to AIS scores, these 27 cases had at least one injury for which the severity could not be determined. The most common body regions to be injured regardless of severity were the upper and lower 105 extremities. Approximately 70% of all motorcyclists analyzed in this study suffered at least one injury to the upper and/or lower extremities. One in five riders (19.5%) suffered injuries to both the upper and lower extremities. For all collision modes analyzed, with the exception of overturn crashes, the lower extremities were most often the region of principal diagnosis (FIGURE 1). The region of principal diagnosis corresponds to the first ICD-9 code (16), but does not provide a measure of severity. The upper 110 extremities were the second most frequent body region for the principal diagnosis for all collision modes analyzed except overturn crashes. Daniello and Gabler 6 FIGURE 1. Region of Principal Diagnosis 115 FIGURE 2 presents the distribution of MAIS 3+ injuries by body region. For all crash modes analyzed except multi-vehicle crashes, the thorax was the most common region for an AIS 3+ injury. For multi-vehicle crashes, the lower extremities suffered AIS 3+ injuries most often. 120 FIGURE 2. Distribution of AIS 3+ Injuries by Body Region Extremity Injuries and Amputations There were 1,206 motorcyclists who suffered an upper or lower extremity injury from the crashes analyzed for this study. As noted above, the extremities were the most frequently injured body regions. To investigate reports of amputations in barrier crashes, the CODES dataset was searched for this type of 125 injury. In our dataset, only 4 motorcyclists suffered an amputation. None of these motorcyclists collided 0% 5% 10% 15% 20% 25% 30% 35% 40% Lower Extremity