Injury patterns to other body regions and load vectors in nearside impact occupants with and without shoulder injuries

Abstract

CIREN and NASS-CDS databases were used to analyze nearside impact injuries. Front seat occupants with and without shoulder injuries were examined on an individual basis in both databases. All vehicles were from model year 2000 or newer. Variables such as the type of collision, change in velocity, principal direction force, demographics, injuries scored by the MAIS and ISS metrics, and injuries to the head, thorax, abdomen and pelvis were included. Shoulder injuries included fractures to the humerus, scapula and clavicle, and associated joint traumas. The median changes in velocities for occupants with and without shoulder injuries were 36 and 32 km/h in CIREN and 29 and 32 km/h in NASS databases. Approximately two-thirds of all cases occurred below 40 km/h. In both databases, the clavicle, scapula and humerus fractures, and AC joint dislocations were found, and the scapula fracture was associated with the clavicle, AC joint, acromion and humerus injuries in few occupants. The clavicle fracture was associated with AC joint and humerus injuries only in the NASS database. Thorax, abdomen and pelvic injuries and skull fractures increased with the presence of shoulder injuries in both databases, albeit not at the same rate. Anterior oblique loading was more frequent than pure lateral loading in both databases suggesting the importance of the oblique vector in side impact trauma. These findings underscore a need for detailed examinations of shoulder load-sharing using biomechanical studies to better understand its role in side impact traumas, shoulder biofidelity and injury assessments in dummies.

INTRODUCTION

The United States National Automotive Sampling System-Crashworthiness Data System (NASS-CDS) and Fatality Analysis Reporting System databases have been routinely used to analyze motor vehicle crashes for continued understanding of injuries for over three decades. Likewise, the Crash Injury Research Engineering Network (CIREN) database has been used since 1996 for obtaining and analyzing more detailed medical aspects of injuries along with crash characteristics. Together, these databases have contributed to recent findings, including reasons for fatalities despite the increased use of manual seatbelts and airbags in frontal crashes (Bean et al., 2009). Data were obtained in the cited study from an analysis of the NASS-CDS databases for vehicle model and calendar years from 2000 to 2007. The structural engagement between the vehicle and its collision partner, corner impacts, oblique crashes and impacts with narrow objects were identified to be the causal agents, and these types of impacts are not in the full frontal impact regulatory tests with the rigid barrier (FMVSS-208, 2001).

Recent analysis of the NASS-CDS and CIREN databases have shown an increase in the incidence rate of thoracolumbar spine fractures in frontal crashes as a function of the vehicle model year from 1986 to 2008 (Pintar et al., 2012). Another study by the same group of authors indicated that oblique side impact crashes are more prevalent than pure lateral crashes, and an obliquely directed vector may be more detrimental to occupant safety than a pure lateral vector (Pintar et al., 2007; Yoganandan and Pintar, 2008). It should be noted that the currently regulated side impact dummy (ES-2re) is based on data from pure lateral tests (Kuppa et al., 2003). Although these types of analyses are available, injury patterns in surviving occupants of nearside impacts with and without shoulder injuries have not been explored in modern vehicles from field data. This is important as the shoulder has been postulated as a secondary load path to influence the kinematics and load-sharing of other components in lateral impacts (Melvin et al., 1998). The purpose of the present study was to determine variations/associations in injury patterns of nearside occupants with and without shoulder injuries in motor vehicle crashes using the two national databases in the United States and determine the role of the shoulder in influencing injuries to other body regions, i.e., head, thorax, abdomen and pelvis.

METHODS

The NASS-CDS and CIREN databases with vehicle model years 2000 or newer were used in the study. The CIREN database is populated with a sample of real-world crashes from several hospitals within the United States. Currently, approximately 4,000 cases are available with detailed crash and occupant medical information. To enroll a case occupant in the database, injuries sustained by the occupant must be at least AIS=3, or moderate to severe trauma with some exceptions. The abbreviated injury scale (AIS) incorporated by the National Highway Traffic Safety Administration (NHTSA) into their standard crash analysis routines is used to document all injuries sustained by the case occupant. The current enrollment criterion is such that the case vehicle should be within six model years from the crash date.

The CIREN database for the years 2000–2011 was analyzed on a case-by-case basis. Left or right side impacts with the highest ranked collision deformation code were used to identify nearside crashes. All occupants were in the front outboard seat, restrained and not ejected. Rollovers were excluded. Only occupants in passenger cars, light trucks and vans seated on the side of the impact, i.e., left side for the driver and right side for the front seat passenger, were considered in the analysis. Medical records, including x-rays, computed tomography (CT) scans, nurses’ notes, emergency medical services and emergency department records were examined, as well as crash information, including potential contacts with the interior or the other front seat occupant. The damage class was determined based on the crush profiles and the assigned vehicle collision deformation code. Data extraction included the principal direction of force (PDOF, Figure 1), categorized into the anterior oblique impacts (1, 2, 10 and 11 o’clock), pure lateral impacts (3 and 9 o’clock) and posterior lateral impacts (4, 5, 7 and 8 o’clock). In addition, the change in velocity data, based on the barrier equivalent speed computed by the NHTSA-developed software and vehicle type and collision partner variables, were included.

Figure 1:

Schematic showing principal direction of forces, loading modes and clock directions.

Data were obtained for the following four categories from vehicle perspectives: matched-vehicle crashes representing two vehicles of the same mass/type; mismatched vehicle crashes wherein the impacted vehicle was different from the other in terms of mass/type; vehicle-to-fixed narrow object (pole, tree) crashes; and other fixed object crashes. A matched crash was defined as a passenger car to passenger car impact, or a van/SUV/pickup truck to another van/SUV/pickup. A mismatched crash was defined as a passenger car versus either a van/SUV or pickup. Based on occupants, demographic factors (age, stature, total body mass, sex and body mass index), data were gathered. The change in velocity, abbreviated injury score (AIS), injury severity score (ISS), maximum AIS (MAIS) data were obtained along with the MAIS of different body regions: head, thorax, abdomen and pelvis.

Injuries to the shoulder consisted of the identification of traumas to the humerus (location issues described later), scapula, clavicle, acromio-clavicular (AC) joint and glenohumeral (GH) joint. Injuries to the thorax were classified according to organ and skeletal traumas. Injuries to the abdominal organs and pelvis were evaluated along with skeletal and organ tissue injuries to the head. All these AIS 2+ injuries were analyzed based on the presence or absence of shoulder injuries.

Procedures incorporated in the analysis of the NASS-CDS database were similar to those used in the CIREN database, with the exception that data were used for the years 2009 and 2010, and the unknown categories were included for the change in velocity. All occupants in the NASS-CDS database had at least one AIS 2 injury to meet the inclusion criteria. Raw data were used in the analysis and the reasons are described later. The NASS-CDS is abbreviated as the NASS in the following sections. Further, the present study was focused primarily on examining associations between different types of shoulder injuries rather than the overall incidence or risk. In addition, only raw data were used from the NASS-CDS files for comparing outcomes from the CIREN database. In other words, recommended weighting factors from the NASS database were not applied.

In addition, MAIS 0 and 1 level injuries were not considered from the NASS-CDS dataset to ensure compatibility of the data collection process with the CIREN database

Data were analyzed based on the impacted vehicle match/mismatch/pole, change in velocity (less than 23 km/h, 24–39 km/h, and 40 km/h+) and ISS (4–13, 14–24, and greater than 25), and these analyses were performed by combining left and right nearside occupants into two groups: with and without shoulder injuries. The percentage of injuries in the two groups and the patterns of injuries were analyzed to different body regions: head (skull bone and brain), thorax (skeletal and organ), abdomen, and pelvis.

RESULTS

The analysis of the CIREN database resulted in 337 nearside impacts to outboard front seat occupants. There were 61 and 276 occupants with and without shoulder injuries, and these two groups were analyzed on an individual basis. The analysis of the NASS database resulted in 211 cases of nearside occupants: 49 and 162 with and without shoulder injuries. Occupants with shoulder injuries had similar distributions in the two databases. Isolated fractures of the clavicle, scapula and humerus, and AC joint dislocations were found in 59, 15, 15, and 2% of the occupants in the CIREN and 45, 16, 6, and 10% in the NASS databases (Figure 2). The scapula was associated with clavicle, AC joint, acromion and humerus injuries in 10% in both databases. In contrast, clavicle was associated with AC joint and humerus in ten percent of NASS occupants only. Likewise, the GH joint injuries occurred in only one occupant in the NASS database.

Figure 2:

Percent of occupants with shoulder injuries in the two databases. Other injuries accounted for small percentages. They are not shown in the bar chart.

Occupants without and with shoulder injuries accounted for 27 and 28% in matched vehicle, 52 and 59% in mismatched vehicle, and 17 and 11% in vehicle-to- narrow object crashes in the CIREN database. However, in the NASS database, occupants without and with shoulder injuries were 32 and 29% for matched, 43 and 53% for mismatched, and 18% each for vehicle-to-narrow object crashes, respectively. Figure 3 shows the distribution of these impacts.

Figure 3:

Percent of occupants with and without shoulder injuries matched, mismatched, and vehicle-to-narrow object crashes from the CIREN (left) and NASS (right) databases. The “other fixed object” category is not included as it constituted a small percentage.

The nearside occupant was in the right front passenger seat in 30% of crashes with shoulder injuries in the CIREN and 35% in the NASS databases, while the occupant was in the right front seat in 24% of non-shoulder injury crashes in the CIREN database se and 29% in the NASS database. The higher percentage of right front seat occupants with shoulder injuries may be a function of the differences in seating position between the drivers and passengers. Males represented contributed 48% of the shoulder injured occupants in both databases, compared to 42 and 43% of non- shoulder injured occupants in the CIREN and NASS databases.

The case vehicle type was a passenger car for the majority of occupants in both databases. Passenger cars accounted for 87 and 71% of the case vehicles with shoulder injuries in the CIREN and NASS databases. For the non-shoulder injured occupants, passenger cars accounted for 79 and 70% in the CIREN and NASS databases.

The highest change in velocity (40 km/h+) category accounted for a large proportion of the shoulder injured occupants: 44 and 18% in the CIREN and NASS databases. In contrast, these data for non-shoulder injured occupants were 31 and 23%, respectively. While all crashes in the CIREN database were coded with known change in velocities, the unknown change in velocity group accounted for 43% of the NASS crashes with and 28% without shoulder injuries. Figure 4 shows these data.

Figure 4:

Percent of occupants with and without shoulder injuries based on changes in velocities from the CIREN (left) and NASS (right) databases

The cumulative frequency distributions of the change in velocities for occupants with and without shoulder injuries from both databases are shown (Figure 5). Approximately two-thirds of all crashes occurred at < 40 km/h. At the 50% level, the changes in velocities for occupants with and without shoulder injuries were: 36 and 32 km/h in the CIREN database, and 29 and 32 km/h in the NASS database. These data are also shown (Figure 6). The principal directions of force are illustrated in figure 7. Data expressed as stacked bar charts are shown in figure 8.

Figure 5:

Changes in velocities for occupants with and without shoulder injuries from both databases.

Figure 6:

Changes in velocity magnitudes at the 50% level occupants with and without shoulder injuries from both databases.

Figure 7:

Percent of occupants with and without shoulder injuries in pure lateral and anterior and posterior oblique loadings from both databases based on the principal direction of force.

Figure 8:

Percent of occupants sustaining MAIS 3+ injuries in the CIREN (left) and NASS (right) among occupants with and without shoulder injuries.

The majority of crashes in both databases were in the anterior oblique category, accounting for 59 and 49% of occupants with and without shoulder injuries in the CIREN, and 59% each in the both groups in the NASS databases. Pure lateral impacts consisted of 38 and 41% of occupants with and without shoulder injuries in the CIREN and 26 and 20% in the NASS databases, respectively. The posterior oblique loadings in both databases for both groups were below ten percent.

The majority of crashes in both databases were in the anterior oblique category, accounting for 59 and 49% of occupants with and without shoulder injuries in the CIREN, and 59% each in both groups in the NASS databases. Pure lateral impacts consisted of 38 and 41% of occupants with and without shoulder injuries in the CIREN and 26 and 20% in the NASS databases, respectively. The posterior oblique loadings in both databases for both groups were below ten percent.

MAIS 3 accounted for 47 and 39% of occupants without and with shoulder injuries in the CIREN and 35 and 20% in the NASS databases (Figure 8). The trend was opposite for MAIS 4 level injuries: occupants without and with shoulder injuries represented 33 and 48% in the CIREN, and 10 and 22% in the NASS databases. MAIS 5 occupants were below 11% each group in the CIREN database and 9 and 8% for occupants without and with shoulder injuries in the NASS database.

Occupants with and without shoulder injuries in the ISS 4–13 group accounted for 20 and 10% in the CIREN and 58 and 47% in the NASS databases, respectively. These data at the ISS 14–24 level in the CIREN database were 43 and 33%, and in the NASS databases were 22 and 24%, respectively. Occupants with and without shoulder injuries in the ISS 25+ group were 37 and 57% in the CIREN database and 20 and 29% in the NASS database. All these data are shown (Figure 9).

Figure 9:

Percent of occupants sustaining ISS 4+ injuries in the CIREN (left) and NASS (right) among occupants with and without shoulder injuries.

Head injuries in the occupants with and without shoulder injuries were found in 50 and 51% of the occupants in the CIREN database, and 46 and 41% in the NASS databases. In contrast, skull fractures (with or without associated brain trauma) in the occupants without and with shoulder injuries occurred in 11 and 13% in the CIREN database and 10 and 22% the NASS database (Figure 10).

Figure 10:

Percent of occupants sustaining head injuries and skull fractures in the CIREN (left) and NASS (right) among occupants with and without shoulder injuries.

Regarding thoracic fractures to the ribs and sternum, both databases showed higher rates in the occupants with shoulder injuries: 69 and 41% in the CIREN and NASS databases than the 49 and 30% without shoulder injuries. Regarding abdominal injuries, in the shoulder injury group, the CIREN and NASS databases showed 49 and 20%, and these data for the non-shoulder injury group were 39 and 20%.

Regarding pelvic injuries, the CIREN database showed similar rates between the shoulder injured and non-injured occupants (62 and 61%). The NASS database showed an increase with shoulder injuries at 31% compared to 23% without shoulder injuries. In general, torso injuries appeared to increase with the presence of shoulder injury in both databases (Figure 11).

Figure 11:

Percent of occupants sustaining thorax, abdomen and pelvis injuries in the CIREN (left) and NASS (right) among occupants with and without shoulder injuries. Torso injuries increased with shoulder injuries in both databases, albeit not at the same rate.

DISCUSSION

As stated in the Introduction, the objective of the study was to determine variations/associations in injury patterns to different body regions for nearside front seat occupants with and without shoulder injuries. This was accomplished using the CIREN and NASS databases. The analysis was performed using raw data because of two primary reasons. The former database is not population-based and traditional weighting factors are not available for converting the unweighted to weighted data. It should be noted that the NASS database has been used with the weighted data and results are reported in AAAM and other venues (Pintar et al., 2000). A comparison with the NASS database is more appropriate with the use of unweighted data from both databases, although NASS allows the flexibility to convert from raw to weighted data. In other words, comparing unweighted data from the CIREN database to weighted results from the NASS databases leads to incompatibility issues. Another reason: the use of unweighted data allowed the authors to examine individual occupants for injuries to different body regions and to differentiate between the bone and organ components in both databases. It should be noted that the methodology used in the present study removed this attribute by discarding non-injury and minor injury data and by not applying the weighting factors. This above approach to better understand injury patterns and associations has been followed in the automotive and clinical literatures (Brasel et al., 2002; Tagliaferri et al., 2009; Yoganandan et al., 2010; Yoganandan et al., 2009; Yoganandan et al., 2004; Yoganandan and Pintar, 2005). However, it should be noted that, while methods have been recently reported to ‘weight’ CIREN data, general consensus is yet to be reached for wider use (Yu et al., 2008). Another factor in the use of weighted data is that the use of such techniques may provide additional information based on ‘weighted analysis.’ However, limitation in the current NASS sample size is a constraining factor. As more information is gathered in due course using AIS 2005 in the NASS database, it would be possible to undertake the weighted data analysis approach.

Generally, samples sizes are greater in the NASS than the CIREN database for a given analysis period. This is because the former database gathers more crashes and crashes of all severities are included. In contrast, the latter database limits the enrollment to only occupants meeting a specific injury severity criteria, specified by the NHTSA (Pintar et al., 2012). The current specification is such that the case vehicle must be within six model years of the crash date, whereas such restriction did not exist in the NASS database until 2009 (Yoganandan et al., 2009).

The analysis of data in the present study from the two databases was not confined to the same years, although identical vehicle model years were used Two years of NASS data were chosen to take advantage of the switch to AIS 2005 codes, which NASS instituted in the year 2009. AIS 2005 codes provide more details on fracture locations than the AIS 1990–1998 update. For example, it is not possible to differentiate the location of the humerus fracture in AIS1990–1998 update, while in AIS 2005, different codes are used for proximal, mid-shaft and distal locations. As indicated in the methods, only proximal humerus fractures were considered to be a shoulder injury in the study. Mid-shaft and distal fractures were not considered to be a consequence of shoulder loading, as arm interaction with the thorax can be a load path. Because the NASS database does not contain radiological images, it is not possible to retrospectively extract fracture locations from cases coded in the AIS1990–1998 scheme. This is in contrast to the CIREN database wherein images and other medical records were available. In addition, from the year 2009, NASS investigators did/do not perform full vehicle inspections on vehicles over ten-years-old from the year of the crash. Because of these factors, the sample size from this database is a limitation.

The human shoulder region consists of the clavicle, acromion, humerus and scapula and their associated joints and soft tissues. This includes the AC, GH and SC joints. Fractures and joint injuries were examined to these components in isolation and combination, in the two databases, as described in the methods and results sections. From anatomical and structural perspectives, the manubrium, which articulates with the sternal end of the clavicle, is the superior most medial connection to the clavicle bone. Therefore, it would be prudent to include manubrium injuries in any shoulder trauma analysis. This was not possible as the coding schema used did not differentiate between the manubrium and sternum components.

Likewise, the medial connection of the clavicles with the manubrium component articulates with the first pair of ribs. These components might also play a role in load-sharing during shoulder loading in side impact trauma. Therefore, it would be prudent to include first rib injuries. While this rib is known to be the strongest structural component in the thoracic skeleton, injuries may ensue from side impacts to its relatively softer articulations: with the manubrium at the rostral end and sternum at the caudal end, which is also the combined connection to the clavicle. This was also not investigated in the present study as rib fracture codes do not accommodate level specificity. Because anthropometric devices continue to be designed with capabilities to measure kinematics and forces in the shoulder (e.g., a sensor is housed in the ipsilateral shoulder rib of the mid-size WorldSID dummy for recording deflections), it may be also necessary to gather such field data to assist in developing biofidelic dummies (Yoganandan et al., 2012a; Yoganandan et al., 2011; Yoganandan et al., 2013). As recent post mortem human surrogates (PMHS) sled tests have been conducted with a modular and scalable load-wall to accommodate subject-specific anthropometry and determine region-specific (example, shoulder) injury metrics, more inclusive field data will be of value for dummy and PMHS injury assessments (Yoganandan et al., 2012b). From these view points, the present study has made a case for shoulder injury-related research.

The SC joint was not traumatized in both databases, indicating that shoulder injuries in nearside crashes do not involve the ribcage-sternum-clavicle complex. As discussed, the AIS coding does not separate the manubrium from the sternum. Although the sternum complex is considered a part of the thorax body region in the AIS coding manual, this junction is an integral part of the shoulder complex. Separation of the anatomical structures, like the manubrium, in the coding process contributes to a greater understanding of injuries to this complex.

The scapula and humerus bone fractures were the most frequently injured shoulder components in both databases. This finding parallels other studies wherein both databases have identified similar results to other body regions (Yoganandan and Pintar, 2005). Further, the scapula and clavicle fractures were often associated with injuries to other parts of the shoulder complex, particularly the humerus and AC joint. These results suggest that these bones act as a medium to transfer the side impact energy to the other components of the shoulder complex (and its inferior anatomical structures). In other words, they are the most frequent load paths for nearside occupants.

An important data for shoulder loading is the seated height of the occupant with respect to the beltline. This was not analyzed in the present study because the precise seated posture of the occupant was not known and furthermore, it was difficult to determine the beltline location due to a paucity of information in the literature. However, an initial attempt was made to extract data from the NHTSA crash tests.

Data from more than 160 crash tests were analyzed on a case-by-case basis to determine the clear distance between the ‘H-point’ (seated dummy’s hip point in a crash test) and vehicle beltline (the bottom of the front door’s window sill). The distance was available for three popular models and three platforms of the same vehicle: years 1998, 2011 and 2012 (Figure 12). It appears that no industry-wide trends exist as the distance in one model increased over all three platforms and the other two models showed an increase from the year 1998 to the year 2011 and then a decrease in the year 2012. The changing beltline across vehicle platforms presents a challenge in retrospective analyses similar to those conducted in the present study. Consequently, field measurements such as beltline location should be recorded along with seated heights of occupants, where possible.

Figure 12:

‘H-point’-to-beltline measurements in popular passenger cars in the United States across three vehicle platforms for the same make and model.

In general, similar trends were observed in the types of collisions between the two databases for occupants with and without shoulder injuries (Figure 3). The median changes in velocities were also confined to a small range (Figure 6). In addition, both databases showed a decrease in MAIS 3 injuries for occupants with shoulder injuries, and an increase in MAIS 4 injuries (Figure 8 Figur). MAIS 2 injuries accounted for no occupants with shoulder injury in the CIREN and only 9% of non- shoulder injured occupants, while they were 43% of shoulder injured occupants and 40% of non-shoulder injured occupants in the NASS database. The difference stems from the fact that the CIREN cases are generally from higher severity crashes due to more specific inclusion criteria than the NASS cases.

Regarding head injuries (Figure 10), the percentage of occupants with skull fractures increased with shoulder injuries in both databases, although the rate was considerably greater in the NASS (2.2 times with than without shoulder injuries) than CIREN (1.2 times) database. Acknowledging that skull fractures occur from contact loading, these results suggest that shoulder injuries resulting from shoulder contact loading exposes the cranium for trauma (Yoganandan and Pintar, 2004; Yoganandan et al., 1995). As side airbag exposures were small in both populations, and the types and technology of side airbags and their deployments have changed over the years (curtain bags, combo bags, etc.), additional data are needed to address field efficacy for shoulder injured occupants (Yoganandan et al., 2007). This is a future research topic.

In addition, thorax bone injuries increased with shoulder injuries in both databases (Figure 11), indicating that nearside occupants sustaining AIS2+ shoulder injuries also sustain fractures to the thoracic skeletal complex. This result suggests a related load-sharing/load path between the two regions: shoulder and thorax. In an analysis of Indy car crashes, it was postulated that shoulder loading may be protective of “internal organ damage in the chest” (Melvin et al., 1998). Differences exist in the occupant, vehicle and crash environments between this and current studies. A study reported PMHS sled tests conducted at 15 km/h in the pure lateral mode (Lessley et al., 2010). One PMHS sustained 16 rib fractures and shoulder injuries, while the other two had no injury, and the authors stated: “the collected response data suggest that the shoulder injury may have contributed to rib fractures in the injured subject.” Acknowledging the lower change in velocity used in sled tests compared to the present field data (Figure 5), and as indicated in their biomechanical results, the shared load path between the shoulder and thorax may render additional forces to the inferior regions of the shoulder in the presence of shoulder injury. The present findings of increases in torso injuries with shoulder injuries (Figure 11) appear to support this biomechanical postulate as applied to motor vehicle environments, although additional efforts are needed to quantify the role of the shoulder to influence traumas to the other body regions, including the head.

Anterior oblique impacts (Figure 7) accounted for a larger share than pure lateral and posterior oblique crashes in both databases, with and without shoulder injuries. The lowest percentage for the posterior oblique loading indicates that these impacts are relatively infrequent, but this observation is tempered by the inclusion/exclusion criteria: only front seat outboard occupants were considered in the present study. However, the recognition of the anterior oblique vector and its relatively greater importance than the pure lateral condition, underscores the importance of oblique loading in side impact crashworthiness. This finding is further reinforced when one considers that current side impact crash test devices are based only on pure lateral impacts (Kuppa et al., 2003; Maltese et al., 2002; Yoganandan et al., 2007). Anterior oblique crashes might be more severe than pure lateral crashes because both of the kinematics of both vehicles. An examination of the changes in velocities as a function of PDOF would be necessary and this is considered as future research.

Along the same vein, it should be noted that the two current mid-size devices, the ES2-re and WorldSID, have distinctly different shoulder designs. Shoulder deflection metrics can be measured with the latter dummy. The seated heights of the two test devices are also different (Yoganandan et al., 2011). The biofidelity of the WorldSID shoulder rib has not been fully assessed, and one of the factors is the role of the shoulder in side impact trauma. Recognizing the importance of shoulder injuries and implications of load paths as discussed above, controlled matched-pair tests should be conducted with PMHS to investigate injury biomechanics and assess the dummy biofidelity. This includes anterior oblique full-scale vehicle and sled tests with anthropometry-specific modular and scalable load-walls to isolate region-specific deflections and forces for deriving injury criteria for these conditions. As injury metrics can be measured in the WorldSID shoulder, it would be necessary to clearly define the role of this body region in side impacts and develop an injury criterion, perhaps by measuring shoulder-specific force and deflections from matched pair tests with PMHS. These discussions are of priority as this dummy is being evaluated for its biofidelity, and injury assessment capabilities by the Working Group (WP29) of the International Standards Organization, the partners of which include the industry, original equipment manufacturers, academic researchers and regulators from around the world. The possibility of the shoulder acting as a load path and affecting the kinematics, loads, energy transfer and injury assessments to the other body regions should be more thoroughly examined in light of these field data from the two databases.

CONCLUSIONS

A retrospective analysis of the CIREN and NASS databases was conducted to determine injuries to front seat occupants in nearside impacts. Occupants with and without shoulder injuries were examined on an individual basis. Variables such as crash type, velocity change, load vector, injuries as scored by the MAIS and ISS, and injuries to the head, thorax, abdomen and pelvis were included. Isolated fractures of the clavicle, scapula and humerus, and AC joint dislocations were found in both databases. The scapula was associated with clavicle, AC joint, acromion and humerus injuries in both databases. Torso injuries and skull fractures increased with the presence of shoulder injuries in both databases, albeit not at the same rate. In addition, anterior oblique loading was more frequent in both databases suggesting the importance of the oblique vector in side impact-induced trauma. These findings underscore a need for detailed examinations of shoulder load-sharing using biomechanical studies to better understand its role in side impact traumas and shoulder biofidelity and injury assessments in modern anthropomorphic test devices.