Structural and Material Changes in the Aging Thorax and Their Role in Crash Protection for Older Occupants

05S-03

Richard Kent, Sang-Hyun Lee, Kurosh Darvish Center for Applied Biomechanics, University of Virginia

Stewart Wang, Craig S. Poster, Aaron W. Lange, Chris Brede, David Lange University of Michigan

Fumio Matsuoka Toyota Motor Corporation

______

ABSTRACT – The human body undergoes a variety of changes as it ages through adulthood. These include both morphological (structural) changes (e.g., increased thoracic kyphosis) and material changes (e.g., osteoporosis). The purpose of this study is to evaluate structural changes that occur in the aging bony thorax and to assess the importance of these changes relative to the well- established material changes. The study involved two primary components. First, full-thorax computed tomography (CT) scans of 161 patients, age 18 to 89 years, were analyzed to quantify the angle of the ribs in the sagittal plane. A significant association between the angle of the ribs and age was identified, with the ribs becoming more perpendicular to the spine as age increased (0.08 degrees/year, p=0.012). Next, a finite element model of the thorax was used to evaluate the importance of this rib angle change relative to other factors associated with aging. A three-factor, two-level factorial design was used to assess the relative importance of rib cage morphology (“young” and “old” rib angle), thickness of the cortical shell (thick = “young” and thin = “old”), and the bone material properties (“young” and “old”) on the force-deflection response and injury tolerance of the thorax. The simulations showed that the structural and material changes played approximately equal roles in modulating the force- deflection response of the thorax. Changing the rib angle to be more perpendicular to the spine increased the effective thoracic stiffness, while the “old” material properties and the thin cortical shell decreased the effective stiffness. The offsetting effects of these traits resulted in similar effective thoracic stiffness for the “elderly” and baseline thoracic models, which is consistent with cadaver data available in the literature. All three effects tended to decrease chest deflection tolerance for rib fractures, though the material changes dominated (a four- to six-fold increase in elements eliminated using a maximum strain criterion). The primary conclusion, therefore, is that an older person’s thorax, relative to a younger, does not necessarily deform more in response to an applied force. The tolerable sternal deflection level is, however, much less.

KEYWORDS – Aging, Thoracic Response, Restraints, Finite Element Model, Anatomy, Thoracic Injury ______

INTRODUCTION the total population by 2010. Recognizing this changing demographic, the U.S. National Highway Importance of an Aging Population Traffic Safety Administration (NHTSA) (1993) Life expectancy in the U.S. has doubled since the identified eight Problem Identification Projects – beginning of the 20th century (Oskvig 1999) and by deficient research areas that will need consideration 2030 25% of the population will be age 65 or older in a proposed national traffic safety plan for older (OECD 2001). Due to longer life expectancy and drivers. Two of the areas identified as needing decreasing birth rates, the population growth rate of additional research are 1) knowledge about crash risk older Americans is expected to be 3.5 times that of for specific medical/functional conditions (which are strongly correlated with age) and 2) analysis of vehicle crashworthiness for older occupants. Address correspondence to Richard Kent, PhD, 1011 Linden Avenue, Charlottesville, VA 22902. Electronic mail: Aging is not just a U.S. domestic public health issue. [email protected] China will have 285 million people over the age of

1 2 Kent et al. / Stapp Car Crash Journal 49 (November 2005)

60 by 2025. The international Organization for Marcus et al. (1983) using laterally struck cadavers. Economic Co-operation and Development (OECD) Schmidt et al. (1975) found a similar age trend for concurred with NHTSA’s findings and identified a cadavers restrained by 3-point seat belts in frontal “pressing need” (OECD 2001) for research to sled tests. At 40 km/h, the average number of rib improve older people’s ability to survive crashes. In fractures increased from one fracture at age 20 to fact, they recommended that governments mandate approximately 10 at age 60. At 50 km/h, the 20-year- improved vehicle safety for older occupants. old average was 2 rib fractures, while the 60-year-old average was approximately 15 fractures. Similar Older drivers are unique in both the circumstances albeit less pronounced trends were also shown with and the outcomes of their collisions. For example, 2-point belt loading. Bending tests of rib segments NHTSA (Cerelli 1998) found that older drivers had from these cadavers indicated a substantial decrease higher belt usage than younger (Figure 1), and Morris in failure load with aging, but the study does not et al. (2002, 2003) identified several characteristics of indicate how the material properties and other factors older-driver crashes, including the disproportionate (e.g., the cross-sectional size and shape of the tested importance of chest injuries for older drivers. bone segment) contributed to this finding. Foret- Thoracic injury from restraint loading was shown to Bruno et al. (1978) studied injury severity as a be particularly significant for older drivers. Kent et function of shoulder belt force using cadavers in al. (2005a) corroborated these findings and showed staged collisions and living humans in field crashes. that 47% of drivers over 64 years of age who died in In both populations, the number of rib fractures for a a frontal crash sustained a fatal chest injury. In given belt load increased with aging. contrast, this number was only 22% for drivers age 16-33. There was a clear shift from fatal head The age-related decrease in thoracic injury tolerance injuries in the younger population to fatal chest has been observed in lateral impacts, as well. In the injuries in the older (Figure 2). Analysis of side development of the Thoracic Trauma Index (TTI), impacts revealed the same over-representation of Eppinger et al. (1984) showed that age was a chest injuries in the older population. Furthermore, significant modifier of the TTI threshold. To account many of these fatal thoracic injuries are not for this sensitivity, they included an age term with the particularly severe. A recent study conducted using kernel TTI function in an injury prediction algorithm. data from the Crash Injury Research and Engineering Kallieris et al. (1992) confirmed this sensitivity in a Network (CIREN) found that rib fractures were the series of lateral cadaver sled tests and included age as most serious injury sustained by 40% of patients over a covariate in a logistic regression model of injury 60 who died of chest injuries from automobile risk, as did Cavanaugh et al. (1993) in their collisions (Wang 2001). As the population continues development of the average spine acceleration (ASA) to age, it is likely that rib fracture frequency and the injury criterion for lateral thoracic loading. resulting morbidity and mortality will increase, and that chest injuries from restraint loading will become More recent studies have compiled larger datasets in a more significant scenario. In order to optimize attempts to quantify the age effects more precisely. restraint performance for an aging population, it is Zhou et al. (1996) developed piece-wise linear necessary to understand and to be able to model the regression models of rib fracture risk using data from force-deflection and injury tolerance changes 107 cadaver tests compiled by Eppinger (1976), 195 associated with the aging thorax. living humans compiled by Foret-Bruno et al. (1978), and 66 cadaver tests performed by Kroell et al. (1971, Decreased Thoracic Injury Tolerance 1974). They found that the tolerance to belt force decreased substantially with age. The oldest subjects That thoracic injury tolerance decreases with aging is (age 66 to 85) retained only about 20% of the well established in the literature. Both cadaver youngest (age 16-35) group’s belt force tolerance. studies and field data have shown that older subjects The sternal deflection tolerance to blunt hub loading generally sustain more rib fractures and other did not decrease as markedly with age: the oldest thoracic injuries for a similar impact severity. Using group retained about 80% of the youngest group’s the cadaver blunt hub impact data collected by Kroell deflection tolerance. It is important to note that the et al. (1971, 1974), Neathery (1974) found both chest difference in loading condition (belt versus blunt deflection and age to be significant predictors of hub) is not the only factor contributing to this injury outcome as defined by the Abbreviated Injury difference. The loading rates were greater on average Scale (AIS). At a given sternal deflection magnitude, with the hub loading, and different criteria (force for the AIS level (dictated primarily by rib fractures) was the belt loading and deflection for the hub loading) found to increase by 0.031 for each year of age, were considered. In 2004, Kent and Patrie compiled which is similar to the 0.025 factor identified by

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 3 chest deflection data from 93 cadaver tests involving 1.0 belt, hub, distributed, and combined belt-and-bag 0.9 loading to assess the role of age using a consistent 0.8 injury criterion. They found age to be a significant 0.7 Age = 70 covariate in a logistic regression model of rib fracture 0.6 Age = 30 risk. The 50% risk of rib fracture onset occurred at 0.5

13% chest deflection for age 70 and at 35% P(Injury) 0.4 deflection for age 30 (Figure 3). The 50% 0.3 Rib fx. >6 probability of serious thoracic injury occurred at 33% 0.2 deflection and at 43% deflection for those ages. 0.1 Rib fx. >0 Interestingly, the loading condition did not 0.0 significantly alter the reduction in chest deflection 0 5 10 15 20 25 30 35 40 45 50 55 Percent Chest Deflection (Cmax) tolerance. Laituri et al. (2005) expanded upon that analysis by validating the cadaver-based findings FIGURE 3. Injury threshold sensitivity to age (from Kent et against a database of field crashes involving living al. 2003). Two ages (30 and 70) and two levels of injury humans and by transforming the cadaver-based chest (any rib fracture, more than six rib fractures) are shown. deflections to apply to the Hybrid III dummy loaded by a seatbelt. Their analysis showed that the 50% Thoracic Stiffness and Aging risk of AIS 3+ thoracic injury decreased from 74 mm The findings of Zhou et al. (1996) imply that the of Hybrid III chest deflection for a 20-year-old to 39 force tolerance and deflection tolerance of the chest mm for a 70-year-old. may decrease differentially with age, and that the restraint condition may play a role in that differential 100 1996 Error bars = 2 standard errors decrease. Unfortunately, it is difficult to separate the 90 2002 82 effect of restraint condition (concentrated belt load 80 76 69 69 versus distributed hub load) from the effect of criteria 70 62 (force versus deflection) using available cadaver data. 60 50 50 From the standpoint of restraint design, in which 40 restraint force is a modifiable specification but chest 30 deflection is a dependent outcome, it is useful to 20 understand how both of these thresholds change with 10 aging, and how the restraint condition (e.g., belt or

Overall Belt Use (Percent) . 0 bag) influences the change. One way to approach Young Adult Adult (25-69) Senior (70+) this is to study the effective stiffness of the chest, defined in this paper as the ratio of the applied force (16-24) over the resulting chest deflection, and how this FIGURE 1. Belt use rate in 1996 and in 2002 for three age stiffness changes with loading condition and with groups in the U.S. Data from Glassbrenner (2003) and age. Few studies have addressed this explicitly, but NHTSA (1997). some aspects of the relationship between aging and effective stiffness can be inferred from the literature. Injured Body Region (Fatal, All Drivers, Frontals) 50 The blunt hub data of Kroell et al. (1971, 1974) are Age Group 1 (16-33) 45 not well suited to evaluating how age influences the 40 Age Group 2 (34-64) effective stiffness since 1) the tests involved a range 35 Age Group 3 (65+) 30 of impactor masses and velocities, and 2) the 25 measured force is dominated by inertial and viscous 20 mechanisms, unlike restraint loading. A limited 15 assessment can be done, however, by comparing 10 Percent of Injuries . chest deflection across ages for those cadavers that 5 were subjected to similar impacts, though this value 0 is affected by the subject’s mass and is potentially Head Neck Chest Abd. Spine Up. Lo. Pelvis biased toward increased deflection with age simply Ex. Ex. because older subjects tend to sustain more rib FIGURE 2. Distribution of fatal injuries by body region for fractures. Despite this bias, an analysis of the 6 three age groups (modified from Kent et al. 2005a). cadavers (15FM, 18FM, 19FM, 20FM, 21FF, and 22FM) exposed to 6.7 m/s impacts with a 23.6 kg impactor (see Table 1 in Kroell et al. 1971) reveals

4 Kent et al. / Stapp Car Crash Journal 49 (November 2005) no relationship between chest deflection and age (p = designing “age-friendly” restraints. The purpose of 0.68, R2 = 0.047). this paper is to identify key factors associated with aging that influence thoracic injury tolerance and Additional insights can be gained by considering the stiffness, and to quantify the magnitude of their force and deflection measured at the onset of rib effects over realistic ranges of variation from fractures and by performing tests in which the inertial adulthood to senescence. forces are minimized. In a series of non-injurious quasistatic hub tests of 4 cadavers age 60-86, Nahum These key factors can be grouped into three general et al. (1970) did not find a strong correlation between areas: material changes, compositional changes, and the effective stiffness of the chest and age. Fayon et geometric changes. As background and motivation al. (1975) published a series of quasistatic belt tests for the current study, further description and a review in which they measured the force and deflection of the literature on these three general categories are when the first rib fracture occurred. Six cadavers presented below. from age 42 to 70 were tested. Interestingly, the ratio of force over the normalized chest deflection at Factor 1 – Material Changes with Aging fracture onset shows a significantly increasing trend It is well known that both cortical and cancellous with age (p=0.01, R2=0.95). Kent et al. (2003, 2004) bone exhibit a decrease in elastic modulus and other performed a series of tests on 15 cadavers using a changes in material response beyond adult middle loading arrangement similar to that of Fayon et al., age (e.g., Yamada 1970, Cowin 2001, Carter and but at a dynamic rate representative of restraint Spengler 1978). Carter and Spengler (1978) loading in a 48 km/h sled test. Diagonal belt, summarized literature from a number of sources and, distributed, and hub loading conditions were referring to a piece-wise linear elastic-plastic model considered. The studies showed inconsistent age of bone material behavior, concluded that the elastic effects, though it should be noted that the youngest modulus decreases by 1.5%, the yield stress by 2.2%, cadaver studied was 54 years old. When all the and the ultimate strain by 5.1% per decade after age loading conditions were grouped in the 2003 study, 20-29. These decreases are due to a number of the younger males (mean age 54 years) exhibited the morphologic and compositional changes that occur as greatest effective stiffness (15,317 N of posterior bone remodels throughout life. The number of reaction force per 100% chest deflection), followed osteons and of osteon fragments per unit of area is by the older males (78 years, 11,010 N), the older greater in bone from younger people. Furthermore, females (81 years, 10,368 N), and the younger while bone mineralization tends to increase with females (57 years, 8,170 N). This trend was similar aging (Black et al. 1974), remodeling causes a when the loading conditions were considered general increase in the porosity of cortical bone with individually. A pronounced size effect, which advancing age, with an accompanying decrease in confounded the assessment of age, was, however, cortical bone density (Evans 1975, Lindahl and present in the data, with larger subjects generally Lindgren 1967). It is this increased porosity that is exhibiting greater stiffness. In 2004, those authors generally regarded as most contributory to the above- added more data and attempted to scale the data to described response changes when bone is modeled as represent a 50th male, but were again unable to a continuum. identify a significant age effect on the effective thoracic stiffness. Some soft tissues exhibit the opposite trend, with an age-related increase in modulus. Kent et al. (2004) Age-Related Factors Affecting Chest Deflection summarized modulus trends with aging for several Tolerance and Effective Thoracic Stiffness thoracoabdominal soft tissues. The trends are not well established or understood, however, so this The literature seems to indicate, therefore, that while study does not consider the effects of changing soft the injury tolerance reduction with aging is well tissue. This is an important consideration, however, established, the exact mechanisms of the reduction that should be studied once data and a suitable model (material or structural) are not well described. of thoracic soft tissues become available. Furthermore, the age-related change in the restraining force required to reach a given level of chest Factor 2 – Compositional Changes with Aging deflection is poorly understood. The few studies Stein and Granik (1976) performed bending tests on documenting age-related changes in thoracic stiffness three ribs from each of 79 human donors having an have found small and inconsistent effects. age from 27 years to 83 years. They found a strong Furthermore, a comprehensive description of the age- inverse correlation between breaking force and donor related factors that influence thoracic stiffness is age at death (p<0.001), but concluded that the unavailable, yet this issue is of critical importance for

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 5 decreased strength was not due solely to increased is applied. Increased kyphosis of the thoracic spine is porosity of the bone material. Those authors perhaps the most well established of these geometric concluded that, like long bones, ribs apparently changes (e.g., Milne and Williamson 1983, Puche et undergo progressive circumendosteal resorption with al. 1995, Goh et al. 2000). Pulmonary pathologies advancing age; but, unlike long bones, ribs show no associated with aging may also generate age- evidence of continued subperiosteal apposition. This correlated geometric changes. Several studies have results in a general decrease in the percent of the rib hypothesized that rib cage depth may increase due to cross-section that is cortical bone. By fitting a linear hyperinflation associated with chronic obstructive regression to their data, Stein and Granik concluded pulmonary disease (COPD) and that a change in rib that the decrease in cortical bone area was angle (as measured in the sagittal plane) may be approximately 0.19 mm2 per year after age 25. Their associated with this change in depth (e.g., Hawes mechanistic explanation for this decrease is 1987, Simon et al. 1973). Findings have been supported by Smith and Walmsley (1957), who found mixed, however. For example, Walsh et al. (1992) that the bone near the endosteal surface in older performed a study of rib cage geometry comparing people is more porous than periostal bone, which is subjects with COPD against an age-matched control more recently formed. Kerley (1965) observed a group without COPD. Geometry at total lung similar effect in 126 human femora, tibiae, and capacity, functional residual capacity, and residual fibulae. Aging through age 95 was associated with a volume were measured using anteroposterior and reduction in the percentage of bone cross-sectional lateral chest radiographs. The only significant area composed of circumferential lamellar bone. structural change they found associated with COPD was a significant lowering of the diaphragm. Rib The costal cartilage also undergoes two distinct angle in the lateral plane was not significantly changes with aging. First, it frequently calcifies with different between the two groups. In a study looking aging (Ferguson et al. 2003). Second, the specifically at rib geometric influences on thoracic “amianthoid” change refers to the age-related stiffness, Kent et al. 2003 identified an anecdotal transition to a more fibrous and morphologically observation of ribs being more horizontal (for a structured material, with greater alignment of standing subject) with increasing age, but did not collagen fibers. Amianthoid regions, which are identify a mechanism for this geometric change visible to the eye as opaque white flecks in the (Figure 4). One purpose of the current work is to otherwise translucent cartilage, contain collagen evaluate this observation using a larger dataset, since fibrils that are thicker by 10 to over 100 times than this structural effect could dominate the thoracic those in normal costal cartilage, and are cell- and response to restraint or other crash loading. proteoglycan-depleted. This transition leads to greater anisotropy of the costal cartilage with aging, METHODS and toward a structure that resembles tendon (Hukins This is a two-part study intended to, first, evaluate the et al. 1976, Mallinger and Stockinger 1988). generality of the age-related change in rib angle The soft tissues also undergo compositional changes described above and, second, to assess the with aging. The arterial wall, for example, develops significance of this change. Particular attention was more collagen cross-linking with age, and fibroblasts paid to the bony rib cage since it is the primary load- produce more elastin and collagen fibers (Vitek and bearing structure in the thorax, rib fractures are Valenta 2001). Age-related changes in organ tissue particularly dangerous for older people, and state-of- include a reduction in the number and the volume of the-art soft-tissue and costal cartilage modeling is glomeruli in the kidney, which are lost to fibrosis currently insufficiently developed to evaluate age (McNelly and Dittmer 1976), and changes in the effects reliably. geometry and structure of the renal vasculature The first component of the study was a computed (Battisti et al. 1996), lung structure (Viidik 1979), tomography (CT) study of rib angle changes and liver (e.g., cirrhosis, Yeh et al. 2002). associated with aging. In the second component, a Factor 3 – Thoracic Geometric Changes with Aging computational study was performed using a full- In addition to the small-scale material and thorax finite element model in a three-factor, two- compositional changes that may affect thoracic injury level factorial experiment to assess the relative and stiffness, age-related changes in larger-scale importance of rib angle, thickness of the cortical thoracic geometry may influence both the force shell, and the bone material properties on the force- required to deflect the chest and the distribution of deflection response and injury tolerance of the strain that is generated within the thorax when a load anteriorly loaded thorax.

6 Kent et al. / Stapp Car Crash Journal 49 (November 2005)

square of the height (m)) who had undergone CT scanning for non-traumatic reasons, could not be obtained. Therefore, CT scans were selected from younger, thinner individuals who had undergone CT

d scanning for trauma evaluations but who were known not to have significant chest or abdominal injuries

CT Be θyounger that might affect chest geometry.

θolder At the University of Michigan, abdominal and pelvic CT scans have been part of the standard initial evaluation of blunt trauma for over a decade. Since 2000, thoracic CT scans have been used for routine evaluation of subjects at risk of chest injuries from blunt mechanisms. Thus, in contrast to the non- CIREN population, the CIREN data provided a FIGURE 4. Computed tomography reconstructed unique opportunity to study patients that were lateral views of the thoracic cage (bone window) of a selected somewhat randomly (inclusion criterion is 17 year-old female (left) and a 64 year-old male involvement in a car crash having characteristics (right) illustrating age-related change in rib slope necessary for CIREN enrollment). Thus, this data (modified from Kent et al. 2003). Definition of rib source has the advantage of containing many angle shown to the left. younger, thinner, and healthier subjects than the non- CIREN group. The disadvantage of the CIREN data, CT Study of course, is that some of the subjects had sustained thoracic loading and some had rib fractures. Clearly, Patients treated at the University of Michigan (UM) this is a confounder in the study of rib angle as a were used for this analysis. The University of function of age since older patients are predisposed to Michigan Institutional Review Board approved the sustaining rib fractures. This was dealt with in two use of patient CT scans for this study. 111 ways. First, a covariate for the presence of rib consecutive adult patients who underwent abdominal fractures was included in a statistical model of rib and/or chest CT evaluations from a motor vehicle angle. Second, the presence of rib fractures was used crash and who consented to be enrolled in the UM as an exclusion criterion and statistical models with CIREN study were included, as were 50 patients this reduced dataset were compared against the from outside the CIREN study, for a total sample size models based on the full dataset. of 161 patients. Immediately prior to initiation of CT scans of the All CIREN patients meeting the above criteria were chest, abdomen or pelvis, patients were given the included, but screening criteria were established for instructions “Take a deep breath and hold” in order to the non-CIREN population. The initial goal with the eliminate motion artifact. Hence, rib angles were non-CIREN population was to select 50 CT scans captured at near-maximal inspiration. Rib angles from individuals undergoing CT scanning for a were measured using Voxar 3D™ software (Voxar reason other than trauma. A sample was desired that Limited, Edinburgh, UK) and a consistent included a wide range of age as well as body mass methodology for all patients, as described below: index and was evenly divided between males and females. Subjects who had significant pulmonary 1. Load image in Voxar, 3D mode. pathology (e.g., thoracic tumors, prior surgery, and COPD) were excluded. After this initial selection 2. Center image to full frontal view, change color strategy, a large pool of older and heavier individuals setting to “Bone (General)”. of both sexes was obtained. However, individuals of normal or less than normal body mass index were 3. Sculpt out leads, internal organs and miscellaneous scarce. As expected, it was also difficult to locate material, using the “3D shape” tool, to get a clear younger subjects who had undergone CT scanning view of the patient’s bony structure. for a purpose other than trauma, as the younger population is generally healthier and therefore less 4. Rotate image 90° to full lateral view. likely to require a chest or abdominal CT. Over 1000 individuals were screened, but young patients of 5. Capture image and return to DICOM. normal or less than normal body mass index (BMI, defined as the total body weight (kg) divided by the 6. Load captured image in 2D.

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 7

7. Select the “measure angle” tool, and measure the was used as the baseline model for this study (Haug angle of the rib, from the superior-most posterior et al. 2004) (Figure 5). A three-factor, two-level point of the rib, to the superior-most anterior point of parametric study was performed to assess age-related the rib. This angle should be measured to a straight changes from this baseline, with two loading vertical normal line (i.e., the CT bed). conditions considered. The age-related factors were

Using this methodology, the rib angle is defined as 1. the angle of the ribs (baseline, “up”), the angle from the bed of the CT scanner, with 90° corresponding to the situation where the ribs are 2. the material properties of the cortical and perpendicular to the bed (i.e., horizontal for a trabecular bone of the ribs (baseline, “old”), and standing subject). Since these scans were analyzed retrospectively, after being obtained for diagnostic 3. the thickness of the cortical shell in the ribs purposes possibly unrelated to the thoracic cage (e.g., (baseline = 5 mm, “thin” = 3 mm). for imaging abdominal or pelvic injuries), not all ribs could be viewed in all scans. Furthermore, in cases See Table 1 and Table 2 for precise definitions of where a rib was fractured or where it could not be these factors and levels. The loading conditions adequately visualized (usually due to an artifact or considered were a single diagonal belt similar in size poor mineralization), that rib’s angle was not and orientation to an automotive seatbelt, and a measured. As a result of these missing data points, a double diagonal belt. The geometry and detailed hierarchical angle measurement strategy was used to description of these loading conditions were define two variables (Rib 7689 and Rib 9876) for presented by Kent et al. (2004), and the each patient. Each of these variables is a single angle computational study matched the experimental that was assumed to be a general descriptor of all ribs conditions described in that study (table-top test, in that patient’s thorax. The measurements used to nominal 1-m/s input deflection with peak magnitude define these two variables were 20% of initial chest depth). The identical input deflection-time history was used for all simulations Rib 7689: Angle of the right 7th rib (7R) used (Figure 5). The 20% input peak deflection (which preferentially, if angle of 7R unavailable, then use corresponds to 46 mm in the 50th percentile male H- angle of 7L. If angle of 7L unavailable, use 6R, then Model) is lower than, for example, the chest 6L, then 8R, then 8L, then 9R, then 9L. deflection threshold prescribed in FMVSS 208, but it represents approximately a 10% risk of severe Rib 9876: Angle of 9R used preferentially, if missing thoracic injury to a 70-year-old, and about 70% risk then use angle of 9L, 8R, 8L, 7R, 7L, 6R, 6L. of any rib fracture (see Figure 3), so it is not an unreasonable design goal for a safety system targeted Ideally, this hierarchical measurement strategy would at older occupants. not be required, and the angles of all ribs could be compared across all subjects. Unfortunately, in this Due to the difficulty in remeshing an entirely new study it was not possible to obtain a sufficient thoracic model to replicate the rib angle change number of scans visualizing the entire thoracic cage. associated with aging, a simplified approach was Future work should assess rib angle trends in the taken. A superiorly directed force was placed on the upper cage. sternum of the model, with the spine fixed, until the 9th rib had rotated superiorly by approximately 7°. Multiple linear regression models were used to assess All internal stresses were then set to zero and the the correlation between rib angle and age, while modified geometry was used as a reasonable controlling for gender, mass, height, BMI, and the representation of the increased rib angle found in the presence of rib fractures. CT component of the study. This geometrically modified model is referred to as the “up” model Computational Study of Factors Affecting (Table 1). As shown in Figure 5, the geometric Thoracic Response change generated by this process is not grossly A well-validated, commercially available finite different from the baseline model, and the sternum, element model of the 50th percentile male thorax (H- scapula, and other thoracic structures appear to be Thorax, H-Model, ESI Software S.A., Cedex, France) representative of a realistic range of human variation.

Author Name et al. / Stapp Car Crash Journal 48 (November 2004)

Force applied to sternum

Baseline “Up” Ribs (Rib 9 rotated ~7° superiorly)

45 40 35 30 25 20 15 10 5 Chest Deflection (mm). Deflection Chest 0 0 1020304050607080 Time (ms)

FIGURE 5. Finite element model used for computational parametric study (anterior tissue removed for visualization), with images of single and double diagonal belt loading. Input deflection-time history and technique for changing rib angle are also shown.

TABLE 1 – Abbreviations Used to Describe Changes from Baseline FE Thorax Model Abbreviation Description Up Force applied to sternum in the superior direction until 9th rib has rotated superiorly approximately 7°. Baseline and “up” ribs are compared in Figure 5. Old Material properties of trabecular and cortical bone modified as shown in Table 2 and Figure 6. Thin Thickness of cortical shell reduced from 5 mm to 3 mm for all ribs.

TABLE 2 – Description of Modified Material Model used for FE Ribs Baseline Properties “Old” Properties Property Cortical Trabecular Cortical Trabecular Reduction Elastic Modulus (Ee) 14.7 GPa 748.8 MPa 10.3 GPa 524.2 MPa 30 % Plastic Modulus (Ep) 600 MPa 1 MPa 420 MPa 0.7 MPa 30 % Yield Stress (σy) 88 MPa 9 MPa 61.6 MPa 6.3 MPa 30 % Failure Strain (εf) 0.03 0.13 0.018 0.078 40 %

The cortical and trabecular bone are modeled as The ribs of the H-Thorax are modeled using shell piece-wise linear, elastic-plastic materials in the FE elements having the properties of cortical bone model (Figure 6, Table 2). The coefficients in the around a core of beam elements having the properties material model were modified in accordance with the of trabecular bone. The thickness of the cortical shell age trends published by Carter and Spengler (1978), is 5 mm in the baseline model. In accordance with which are outlined in some detail above. The moduli the regression equation of Stein and Granik (1976), and yield stress were reduced by 30% from the this thickness was reduced to 3 mm to represent the baseline, and the failure strain was reduced by 40%. reduction in cortical bone area that occurs with aging. This modified model was defined as the “old” The model with the modified shell elements was material properties. named the “thin” model.

8 Kent et al. / Stapp Car Crash Journal 49 (November 2005) 9

Stress Typical of biological structures, there was a large Ep Tension σ amount of scatter in the rib angle data, indicating that y individual variability is a dominant factor. Despite this large scatter, there was a significant positive correlation between age and rib angle (rib angle 9876 Ee used unless specified, and trends between the two rib ε f angle definitions are the same unless noted) in the CIREN population (correlation 0.25, p=0.009), Strain εf Ee consistent with earlier observation that older individuals tended to have ribs closer to perpendicular to the spine. There was also noted to Comp. σy Ep be significant positive correlation between rib angle and weight (0.57, p<0.001) as well as BMI (0.667, p<0.001), again consistent with the clinical FIGURE 6. Symmetric bone material model used for observation that heavier individuals tend to have the H-Thorax ribs. more horizontal ribs. In the non-CIREN population, Two primary outcomes were studied with the FE a trend toward positive correlation between age and model. First, the posterior reaction force (see Kent et increased rib angle was observed, though this al. 2004) generated by the deforming thorax was correlation was not statistically significant (p=0.099). compared across the range of parameters for both There was noted to be a strong positive correlation loading conditions. By cross-plotting this force with between BMI and rib angle in the non-CIREN data, the input deflection, it was possible to assess the which had been observed anecdotally and was part of sensitivity of the effective thoracic stiffness to the the reason for care with subject selection. Consistent various parameter changes. Second, the onset and with the CIREN data, rib angle in the non-CIREN number of rib fractures were studied by counting as a group was significantly more horizontal with function of time the regions of elements that increasing BMI (correlation 0.45, p<0.001). This exceeded the failure strain. This gives an indication association was significant in both females (p=0.006) of how rib fracture threshold is affected by changes and males (p=0.025). As might be expected, similar in rib cage structure, bone material properties, and rib significant positive correlation was seen between rib constituency. angle and weight for the whole non-CIREN population (p=0.0012) as well as female subjects RESULTS (p=0.0129) and male subjects (p=0.0156). There was no correlation between rib angle and subject height. CT Study of Rib Angles – CIREN and non- CIREN Data CT Study of Rib Angles – Combined Dataset The final database included 111 CIREN cases and 50 Rib angle values in the combined dataset ranged from non-CIREN hospital patients, though some data 33.5° to 74.0° (Figure 8). Despite the large scatter, fields were missing from some cases. When all blank however, the slope of the regression line with age fields were removed, the total database was 152 was significantly different than zero (p=0.012), with cases. The age ranged from 18 to 89 years (mean of the rib angle increasing with increasing age. Weight 48.3 years) and there were more females (81) than and BMI were also found to be significantly related males (71). The height and weight ranges were 150 to rib angle (p<0.001), with a pronounced increasing cm to 193 cm (mean 171 cm) and 43 kg to 150 kg trend (Figure 8) that was much more pronounced (mean 77.9). The BMI (kg/m2) ranged from 16.0 to than the trend with age. Height, on the other hand, 55.1 (mean 26.6). The distributions were reasonably was not correlated with rib angle. The effect of close to normal for most of the relevant parameters gender differed depending on the rib angle (Figure 7), though there was a slight over- measurement used. Gender was not significantly representation of young subjects, and weight and related to Rib9876 (p=0.621), but it was significantly BMI were skewed slightly to the left. The influence related to Rib7689 (+3.1 degrees for males, p=0.017). of this skew was assessed by randomly removing This may indicate that gender plays a larger role in samples to normalize the distribution and re-fitting the slope of the more superior ribs, though more data statistical models with the reduced dataset. This was are needed to confirm this interpretation of the found to have a minimal effect on the resulting finding. statistical models so only the models with the full dataset will be presented.

Author Name et al. / Stapp Car Crash Journal 48 (November 2004)

Age Frequency Height Frequency 25.0% 21.7% 21.1% 20.0% 17.8% 15.8% 16.4% 15.0% 13.8% 13.2% 11.8% 11.8% 12.5% 11.8% 9.9% 9.9% 10.0%

Percent of Cases of Percent 5.3% 5.0% 3.9% 3.3%

0.0% 18-27 27-36 36-45 45-54 54-62 62-71 71-80 80-89 150- 155- 161- 166- 172- 177- 182- 188- Age Range (Years) 155 161 166 172 177 182 188 193 Height Range (cm)

Rib Angle Frequency Weight Frequency 35.0% 30.3% 30.0% 25.7% 23.7% 25.0% 22.4% 20.4% 20.0% 17.8% 15.8% 15.0% 9.2% Percent of Cases of Percent 10.0% 7.2% 7.9% 7.9% 5.9% 5.0% 1.3% 1.3% 1.3% 2.0% 0.0% 43-56 56-70 70-83 83-97 97- 110- 123- 137- 33-39 39-44 44-49 49-54 54-59 59-64 64-69 69-74 110 123 137 150 Rib Angle Range (deg) (Rib9876) Weight Range (kg)

FIGURE 7. Frequency distributions of subjects used in CT study.

From the univariate analysis, obesity is the strongest Rib angle (9876) (deg.) = 35.4 + 0.0412*age (years) predictor of rib angle. Age, however, was not + 0.572*BMI (kg/m2) + 1.03*gender [2]. correlated with weight (p=0.826) or BMI (p=0.282), so it is reasonable to presume that the age effect may To assess the importance of rib fractures as a be independent. To evaluate this presumption, a confounder, two more multiple regression models multiple linear regression model was developed using were developed. One model (Model 2) included a both age and BMI as predictors of rib angle (Model covariate for the presence of rib fractures (1 = 1). Model 1 reveals that both age and BMI (or fractures present), while the other (Model 3) was weight, which is highly collinear) contribute to the based only on the data from those patients not having variation in rib angle. The regression equation for rib fractures (i.e., the presence of any rib fracture was the model with BMI is an exclusion criterion). The exclusion of patients with rib fractures reduced the dataset to 88 patients. Rib angle (9876) (deg.) = 30.1 + 0.0584*age (years) Age remained a significant predictor of rib angle in + 0.749*BMI (kg/m2) [1]. both Model 2 (p = 0.015) and Model 3 (p = 0.018), and the presence of rib fractures was not a significant This is a significant regression (p<0.001) and both predictor of rib angle in Model 2 (p = 0.590). Thus, predictors are significant (Table 3). Adding gender it is reasonable to conclude that the presence of rib to the multivariate regression modified the model fractures in the CIREN population does not strongly slightly, but the gender term was not significant influence the relationship between age and rib angle (male=1, p=0.298): in this dataset.

10 Author Name et al. / Stapp Car Crash Journal 48 (November 2004)

Rib Angle vs. Age (CIREN and Non-CIREN data) Rib Angle vs. Height (CIREN and Non-CIREN data) 80 y = 0.0811x + 48.962 75 R2 = 0.0413, p = 0.012 70 65 60 55 50 45

Rib AngleRib 9876 (Deg) . 40 y = -0.0515x + 61.7 35 2 R = 0.0045, p=0.411 30 18 28 38 48 58 68 78 88 145 150 155 160 165 170 175 180 185 190 195 Age (Years) Height (cm)

Rib Angle vs. Weight (CIREN and Non-CIREN data) Rib Angle vs. BMI (CIREN and Non-CIREN data) 80 75 70 65 60 55 50 45

Rib AngleRib 9876 (Deg) . 40 y = 0.2257x + 35.296 2 y = 0.7637x + 32.571 R = 0.3559, p<0.001 2 35 R = 0.4353, p<0.001 30 40 50 60 70 80 90 100 110 120 130 140 150 160 15 20 25 30 35 40 45 50 55 Weight (kg) BMI

FIGURE 8. Rib angle as a function of age, height, weight, and BMI, with univariate linear regression models. The circled data points were the target for the FE model structural modification.

TABLE 3 –Multiple Regression Models of Rib Angle Based on these patients, there is evidence that the ribs are closer to perpendicular to the spine in older Predictor Coefficient p people than in younger. The univariate regression

model indicated a mean change of approximately Model 1 0.08°/year from age 20 to age 80, with the largest (No exclusions; no covariate for rib fractures) difference between subjects being 44.5° (smallest Constant 30.1 <0.001 angle was 33.5° in a 29-year-old, largest was 78° in a Age 0.0584 0.017 62-year-old). In the following section, the results of BMI 0.74878 <0.001 the finite element study shed some light on the importance of this structural change relative to other Model 2 changes associated with aging. (No exclusions; rib fracture in model) Constant 30.1 <0.001 Finite Element Parametric Study – Thoracic Age 0.0601 0.015 Stiffness BMI 0.75338 <0.001 The technique used for changing rib angle in the FE Rib Fractures -0.497 0.590 model generated a reasonable geometry for the “up”

ribs condition. The most representative end-point set Model 3 of CT patients (circled in Figure 8) were used as the (Patients with rib fractures excluded, n = 88) target for the FE modification. The geometry of the Constant 27.6 <0.001 baseline and “up” FE models is compared against CT Age 0.0733 0.018 scans from those patients in Figure 9. BMI 0.82799 <0.001

11 12 Kent et al. / Stapp Car Crash Journal 49 (November 2005)

FE Model Structural Modification shown in Figure 10 should be considered in the interpretation of that value.

Finite Element Parametric Study – Rib Fracture Threshold There is an apparent influence of element elimination (“fracture”) in the above-described stiffness trends. As described in the Methods, the change to “old” Baseline FE “Up” ribs FE model material properties included a decrease in the failure model (Rib 9 rotated ~7° superiorly) strain of the bone elements. Changing the thickness of the cortical shell or the rib angle would also Patients Closest to End Points of Regression Line logically change the relationship between the chest deflection magnitude and the strain distribution within the rib cage. The various thorax models therefore exhibited different chest deflection levels at the onset of element elimination (i.e., rib fractures). For both loading conditions, rib “fracture” onset was 18 mm of deflection (Figure 12) for both the “old” model and the “up + thin + old” model. The number of regions having eliminated elements also built 18 year-old, 50° 89 year-old, 57° relatively rapidly in time for these models. For rib angle rib angle example, the “up + thin + old” model exhibited 4

FIGURE 9. Comparison of rib angle changes in FE (single diagonal belt) and 8 (double belt) “fractures” model and patients nearest endpoints of regression at 30 mm of chest deflection. In contrast, the line (circled in Figure 8). baseline model, the “up” model, and the “thin” model tolerated nearly 30 mm of chest deflection prior to The simulation results showed sensitivity to all three any elements exceeding the failure strain, and they of the parameters studied. Relative to the baseline didn’t reach 4 and 8 “fractures” until chest deflection model, changing to the “old” material properties or reached 40-45 mm. the “thin” cortical shell resulted in a similar decrease One significant aspect of these element elimination in the force generated at each level of chest deflection patterns is the decrease in thoracic stiffness (Figure 10). This was true for both the single and the associated with them. At low-to-moderate double diagonal belts, and the magnitude of the magnitudes of chest deflection, the models with “up” reduction was similar for both loading conditions. In ribs exhibited a greater thoracic stiffness than the contrast, when the ribs were “up”, the effective baseline model, even when the “old” material stiffness of the thorax increased relative to the properties and the “thin” cortical shell were used. baseline. For example, at 30 mm of chest deflection Between 30 and 40 mm of deflection, the force with the single diagonal belt, the baseline model generated by the “thin + old + up” and the “old + up” generated 2.0 kN of force, while the “up” ribs model models became lower than the force generated by the generated 2.4 kN. A similar increase in stiffness was baseline model (see Figure 10). Presumably this is found with the double diagonal belt loading. related to the fact that element elimination becomes When factor interactions were studied, the model significantly greater in those models at that level of showed that the effects of all three changes resulted chest deflection. in a thoracic stiffness response very near the baseline Another significant result is that the material property (Figure 10). The “thin + old” model had the lowest description, the cortical shell thickness, and the rib effective stiffness, while the “old + up” model angle all influenced the number of “fractures” actually had slightly greater stiffness than the generated by the loading. For example, with the baseline. When all three changes were made (“old + single diagonal belt, the total number of “fractures” at up + thin”), the response was quite near the baseline 20% chest deflection ranged from 5 with the baseline for both loading conditions. The influences of all model to 21 with the “up + thin + old” model. The factors are summarized in Figure 11. In that figure, single biggest contributor to fracture outcome was the the force at 30 mm of deflection is used as a single material properties, as evidenced by the curves in indicator of stiffness trends, though the nonlinearity Figure 12. The change to “old” material properties increased the total number of “fractures” by a factor

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 13 of 2 to 3 for the two loading conditions. Changing There is a clear interaction, however, since the the single factors for rib angle and cortical shell combination of all three changes increased the thickness did not have such a large effect, increasing number of “fractures” by a factor of 4 to 6. the total number of “fractures” by only 20%-30%.

Effects of Single-Factor Changes - Single Diagonal Belt Effects of Multiple Factor Changes - Single Diagonal Belt

4000 Up Old + Up 3500

3000 Baseline model Thin + Old + Up 2500

2000 Old Thin Posterior Force (N) 1500

1000 Thin + Old Baseline model

500

Effects of Single-Factor Changes - Double Diagonal Belt Effects of Multiple Factor Changes - Double Diagonal Belt

Baseline model 6500 Up Thin + Old + Up 5500 Baseline model Old + Up 4500

3500 Posterior force (N) force Posterior 2500 Thin Old Thin + Old 1500 20 25 30 35 40 45 20 25 30 35 40 45 Chest Deflection (mm) Chest Deflection (mm)

FIGURE 10. Effect of single (left) and multiple (right) factor changes on the force-deflection response of the thorax loaded by a single (upper) and double (lower) diagonal belt.

DISCUSSION Puche et al. (1995) found spinal kyphosis in postmenopausal women to be related to reduced physical The first component of this study used CT scans of activity, which is correlated with increased weight patients over a range of ages and body compositions and BMI. to show a significant correlation between obesity, age and the angle of the ribs in the sagittal plane. There This spinal geometry hypothesis is not only a are several possible explanations for this finding. possible explanation for the finding, but also a The simplest is that the rib angles are associated in potential confounder of the angle measurement. The some way with the changes in spinal geometry that study used patients supine on a flat CT bed. This was occur with aging, and that a similar change may not only a necessity given the clinical protocol, but it occur with obesity. Segmental angulation, lordosis, also serves as a useful method for standardizing kyphosis, sacral inclination, and other factors all may spinal geometry to the extent possible across patients change with aging (Jackson and McManus 1994, having a range of ages and body types. If the patients Gelb et al. 1995, Vendantam et al. 1998, Korovessis had been, for example, standing when the rib angle et al. 1998, Hammerberg and Wood 2003). There is measurements were taken, the spinal geometry would evidence in the literature that body weight might also have confounded the interpretation of rib angle be correlated with changes in spinal geometry. Milne greatly. As taken, however, the rib angle and Williamson (1983) found size, in addition to age, measurements are reasonably well standardized to a to be a significant predictor of spinal geometry and

14 Kent et al. / Stapp Car Crash Journal 49 (November 2005)

common reference frame since the CT position is Number of Areas Eliminated - Single Diag. Belt more controlled than a standing position. It is, of (Represents Number of Rib Fractures) course, possible that the relationship between age and 20 rib angle would be different for a seated or standing Up + Thin + Old subject, since the spinal curvature would not be 15 standardized. This issue deserves further study since the ultimate goal of this research is the development of restraint systems optimized for people in a seated 10 position. On the other hand, it is not unreasonable to Number Up Old Thin conclude from the current data that rib angles vary 5 greatly among humans and that the angle between the ribs and a restraint’s line of action may vary with age Baseline and with obesity. 0

Number of Areas Eliminated - Double Diag. Belt 120% (Represents Number of Rib Fractures) Single Diag Belt 70 115% Double Diag Belt 60 110% 50 Up + Thin 105% + Old 100% 40

95% 30 Number Old

(Norm. to Baseline) to (Norm. 90% 20 Up

Force at 30 mm at Deflection. Force 85% Thin 10 80% Baseline 0 550% 15 20 25 30 35 40 45 Chest Deflection (mm) 500% 450% 400% FIGURE 12. Element elimination patterns from FE model. 350% 300% Tissue differentiation within the thoracic cage may 250% also play a role in changing rib geometry with both

(Norm. to Baseline) 200% obesity and aging. It is well established that aging is associated with a decrease in pulmonary reserve and Fractures at Max.Fractures . Deflection 150% 100% in tidal volume. In other words, older people take shallower breaths than younger. Numerous studies Up Old Old + Thin Thin + Thin + Up Old Old + have shown that the history of mechanical loading Up can influence the biosynthesis of connective tissue FIGURE 11. Summary of factor effects on thoracic constituents, such as collagen and proteoglycans in stiffness (top) and the number of rib fractures. Note the solid matrix of cartilage (e.g., Guilak et al. 1994, that all modifications increased the number of rib Davisson et al. 2002, Mikic et al. 2004). Over time, fractures, but the effects on stiffness were smaller and therefore, shallower breathing may result in a not monotonic. decrease in the functional range of motion of the costovertebral joints via morphological changes in A second possible explanation for the rib angle the connective tissues. While it is not clear how this finding is that pulmonary pathology, which is might result in a gradual change in rib angle, it is a associated with aging and obesity, may influence rib possible mechanism of geometric change associated angle. As described in the Background section with aging, and possibly obesity. above, several studies have hypothesized that rib cage depth may increase due to hyperinflation The second component of this study assessed the role associated with COPD and that a change in rib angle of this rib angle morphology relative to other changes may be associated (e.g., Hawes 1987, Simon et al. that occur with aging. Specifically, thinning of the 1973). cortical shell and a decrease in the moduli, yield stress, and failure strain of the bone material were

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 15 studied using a detailed and well validated FE model apportionment of responsibility for that decreased rib of the thorax. This modeling study showed that both fracture tolerance. The FE model shows that the age- the effective thoracic stiffness and the number and related change in material properties (predominantly onset of rib fractures are strongly influenced by all of failure strain) is the most important factor. Changing these factors. The change in bone material properties rib angle and cortical thinning both tend to decrease was shown to have the strongest influence on rib rib fracture tolerance, but the effect is smaller than fracture outcome, with rib angle and cortical thinning the effect of material property changes. The having a less pronounced, though still important, mechanism by which cortical thinning decreases rib effect. All three of the factors associated with aging fracture tolerance is obvious: there is less area over (rib angle increase, cortical thinning, and material which to distribute the force and the stress and strain property changes) tended to decrease the sternal increase accordingly. The mechanism by which a deflection threshold at which fractures began to occur changing rib angle decreases tolerance is likely the and to increase the number of fractures present at a same mechanism by which it increases thoracic given chest deflection magnitude. In contrast, the stiffness. The angle between the applied force vector thoracic stiffness was influenced differently by these and the axis of the rib (in the sagittal plane) is larger factors. As would be expected, the change in in the baseline model. Therefore, the component of material properties and the cortical thinning tended to the force vector acting in the plane of the rib (and decrease the effective thoracic stiffness. These hence generating bony deformation) is smaller and effects were offset, however, by the increase in rib the component acting normal to the rib plane (and angle, which tended to increase the thoracic stiffness. hence causing rotation of the rib) is larger. Thus, the In fact, the offsetting effects of all three changes structural response of the baseline model was less resulted in a model having virtually the same stiff and there was less bony deformation (and hence stiffness as the baseline model, even when the rib strain) at a given level of chest deflection. As the additional rib fractures were considered. rib was rotated closer to perpendicular to the spine, the bony deformation became greater and the rib The findings of the FE modeling study are consistent rotation under load became less. This tended to with the scant literature available on aging trends in increase the structural stiffness of the thorax, while thoracic biomechanics. There is very little data building strain faster as a function of chest deflection. documenting how thoracic stiffness changes with This phenomenon is illustrated in Figure 13. aging, but the available information shows virtually no relationship. In fact, earlier studies (Fayon et al. This study is, of course, subject to some important 1975, Kent et al. 2003) found aging, if anything, to limitations. The first is the supine position of the be associated with a slight increase in thoracic patients used in the CT study and possible selection stiffness. The rib angle change identified in this bias as a result of the retrospective nature of the CT study may be at least a partial explanation for the study. While careful screening and statistical finding that subjects having clearly lower bone analyzes were performed in an attempt to remove this moduli and mineral density do not exhibit markedly bias, a prospective study documenting the rib angles lower thoracic stiffness. The thorax is a complex and of a group of patients as they age would be more inhomogeneous structure, and its response to restraint robust. Unfortunately, the time required to observe a and other types of loading is dictated not only by the change in rib angle, and the radiation dose associated material properties of the constituents, but also by the with a full-thorax CT scan precludes a prospective structural arrangement of those materials. In the case study of asymptomatic patients. Future work should of an aging human, it appears that this structural include some assessment of the positional arrangement includes a slight but significant change dependency of the rib angle findings. Loading in rib cage geometry. This study has shown that this conditions other than the two considered here should structural change is, in terms of thoracic stiffness, as also be studied. Air bag loading, for example, is not important as the well-documented material changes as focused on the shoulders as belt loading and may that occur with aging. exhibit different stiffness and injury threshold trends than those seen in this study. Compared to the literature on thoracic stiffness, there is more data available on the role that aging plays in The lack of adequate visualization for all ribs, and the reducing the chest deflection at which ribs fracture accompanying necessity of using a hierarchical and in increasing the number of fractures that result strategy to define rib angle, are additional limitations from a given magnitude of chest deflection. Again, of this study. The use of two angle measurement the findings of this study are consistent with that strategies (9876 and 7689), which showed consistent literature. The contribution of this study is the

16 Kent et al. / Stapp Car Crash Journal 49 (November 2005) trends with respect to age and BMI, is, however, an that the moduli of thoracic viscera increase with indication that the findings are robust. aging, and that organs may atrophy or otherwise change structure. Recent papers have shown that the The FE study also has some important limitations. organs and superficial tissues bear up to 60% of the First, the maximum strain criterion and element force generated by a restraint system (Murakami et elimination strategy used here to predict rib fracture al. 2004, Kent et al. 2005b), so pronounced changes onset and distribution is probably too simple to in either the moduli or the structure of the thoracic capture all aspects of rib fracture in humans, soft tissues could have pronounced effects on including how rib fractures affect thoracic stiffness. thoracic stiffness. How these soft tissue factors Second, this model is a simplified representation of compare with or contribute to the effects of the the human thoracic cage. A more complete factors studied here is a topic worthy of additional description of the model is given by Haug et al. research. (2003), but important simplifications include the rib cross sectional shape, which is simplified as a Finally, while aging was the focus of this study, our hexagon, the costo-vertebral interface, and the findings relative to obesity deserve further research. internal organs, which are lacking detail. The H- Obesity rates are increasing significantly both in the Thorax is, however, a reasonable representation of U.S. and abroad, and restraint systems optimized for the state of the art and has gone through a validation the entire population must consider this. effort comparable to any commercially available FE thorax model. CONCLUSIONS Aging is associated with a variety of structural and material changes. A study of CT scans from 161 patients identified a significant relationship between rib angle and both age and BMI. With increasing age and increasing BMI, the ribs became closer to perpendicular to the spine. This important Younger rib cage Older rib cage morphological change was compared to two other age-related factors that could influence thoracic biomechanics and injury tolerance: changing material properties and thinning of the cortical shell. It is concluded that the rib angle change can offset the effects of material changes and cortical thinning when the effective stiffness of the thorax is More joint rotation, Less joint rotation, considered. Cortical thinning and decreased bone less bone strain more bone strain modulus both tend to decrease the stiffness of the thorax, while the change in rib angle tends to increase stiffness. As a result, it is concluded that the effective stiffness of the chest when loaded by a restraint system is not highly sensitive to age up to the point at which extensive rib fractures begin to affect the response. This conclusion is supported by the available literature. In contrast, the chest deflection threshold is decreased by all three factors. Joint rotation with Bony deformation The increased rib angle increases the bony “young” rib angle with “old” rib angle deformation that occurs at a level of chest deflection, cortical thinning increases stresses by decreasing the FIGURE 13. Illustration of mechanism by which area over which force is distributed, and the aging increased rib angle increases thoracic stiffness while material properties include a decrease in failure strain decreasing chest deflection tolerance. of the bone. Thus, older people would be expected to have decreased rib fracture tolerance due to multiple Finally, this study did not consider age-related consequences of aging. This is consistent with the changes in the thoracic soft tissues. Future work literature, which shows older people to have higher should assess the role of the soft tissues in age-related risk of rib fractures in a given field impact, and older changes in thoracic response since, as discussed cadavers to sustain more fractures for a given level of above and by Kent et al. (2004), there is evidence chest deflection in a laboratory test.

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 17

Rib fractures, particularly from restraint loading, are Cerelli, E. (1998) Research Note: Crash data and a common and dangerous injury for older people. rates for age-sex groups of drivers, 1996. National Furthermore, this injury scenario is likely to become Highway Traffic Safety Administration, U.S. an even higher priority in the future as restraint usage Department of Transportation, Washington, DC. climbs and the population continues to age. Biofidelic thoracic response is also required in order Cowin, S. (2001) Bone Mechanics Handbook. CRC to assess injury risk to other body regions, such as the Press, New York, NY. head, since the thorax is the primary structure Davisson, T., Kunig, S., Chen, A., Sah, R., Ratcliffe, through which restraint forces are transmitted to other A. (2002) Static and dynamic compression body regions. Accurate models of the aging thorax modulate matrix metabolism in tissue engineered are therefore a critical component of restraint cartilage. J. Orthop. Res. 20:842-8. optimization, and should consider all of the consequences of aging, not just the well-known Eppinger, R. (1976) Prediction of thoracic injury increase in bone porosity and resulting degradation of using measurable experimental parameters. Proc. material properties. 6th International Conference on Experimental Safety Vehicles (ESV), pp. 770-80. ACKNOWLEDGMENTS Eppinger, R., Marcus, J., Morgan, R. (1984) The authors gratefully acknowledge Toyota Motor Development of dummy and injury index for Corporation, which funded this study. We also NHTSA’s thoracic side impact protection research acknowledge our colleagues at the University of program. Paper 840885, Society of Automotive Virginia Center for Applied Biomechanics and at the Engineers, Warrendale, PA. University of Michigan Program for Injury Prevention and Education (UMPIRE), who spent Evans, F. (1975) Mechanical properties and histology countless hours compiling and reducing the data of cortical bone from younger and older men. Anat. presented in this paper. Finally, we acknowledge the Rec. 185:1-11. support of NHTSA and of other sponsors of the Crash Injury Research Engineering Network Fayon, A., Tarriere, C., Walfisch, G., Got, C., Patel, (CIREN), without which the valuable CT data A. (1975) Thorax of 3-point belt wearers during a analyzed here would not have been available. The crash (experiments with cadavers). Paper 751148, views expressed in the paper are those of the authors Proc. 19th Stapp Car Crash Conference. and do not necessarily represent the views of the Ferguson, V., Bushby, A., Boyde, A. (2003) sponsors or of others acknowledged here. Nanomechanical properties and mineral REFERENCES concentration in articular calcified cartilage and subchondral bone. J. Anat. 203(2):191-202. Battisti, A., Barili, P., Ferrante, F., Valsecchi, B., Amenta, F. (1996) Effect of treatment with L- Foret-Bruno, J., Hartemann, F., Thomas, C., Fayon, deprenyl on age-dependent microanatomical A., Tarriere, C., Got, C., Patel, A. (1978) changes in the rat kidney. Mechanisms of ageing Correlation between thoracic lesions and force and development 89:1-10. values measured at the shoulder of 92 belted occupants involved in real accidents. Paper Black, J., Mattson, R., Korostoff, E. (1974) 780892, Proc. 22nd Stapp Car Crash Conference. Haversian osteons: size, distribution, internal structure, and orientation. J. Biomed. Mater. Res. Gelb DE, Lenke LG, Bridwell KH, et al. (1995) An 8:299-319. analysis of sagittal spinal alignment in 100 asymptomatic middle and older aged volunteers. Carter, D. and Spengler, D. (1978) Mechanical Spine 20:1351–8. properties and composition of cortical bone. Clinical Orthopedics and Related Research, Goh, S., Price, R., Song, S., Davis, S., Singer, K. (135):192-217. (2000) Magnetic resonance-based vertebral morphometry of the thoracic spine: age, gender, Cavanaugh, J., Zhu, Y., Huang, Y., King, A. (1993) and level-specific influences. Clinical Injury and response of the thorax in side impact Biomechanics 15:417-25. cadaveric tests. Proc. 37th Stapp Car Crash Conference, pp. 199-222. Guilak, F., Meyer, B., Ratcliffe, A., Mow, V. (1994) The effects of matrix compression on proteoglycan

18 Kent et al. / Stapp Car Crash Journal 49 (November 2005)

metabolism in articular cartilage explants. Kerley, E. (1965) The microscopic determinates of Osteoarthritis Cartilage 2:91-101. age in human bone. Am. J. Phys. Anthropol. 23:149. Hammerberg, E. and Wood, K. (2003) Sagittal profile of the elderly. J. Spinal Disorders and Korovessis PG, Stamatakis, Baikousis AG. (1998) Techniques 16(1):44-50. Reciprocal angulation of vertebral bodies in the sagittal plane in an asymptomatic Greek Haug, E., Choi, H., Robin, S., Beaugonin, M. (2004) population. Spine 23:700–5. Human Models for Crash and Impact Simulation, Computational Models for the Human Body, Kroell, C., Schneider, D., Nahum, A. (1971) Impact Special Volume Handbook of Numerical Analysis, tolerance and response of the human thorax. Paper Volume 12, Elsevier, New York, NY. number 710851, Society of Automotive Engineers, Warrendale, PA. Hawes, L. (1987) A history of radiology of the chest from 1895 to 1920. Radiology, Taveras and Gray Kroell, C., Schneider, D., Nahum, A. (1974) Impact (eds.) Lippincott, Philadelphia, PA, p. 1-7. tolerance and response of the human thorax II. Paper number 741187, Society of Automotive Hukins DW, Knight DP, Woodhead-Galloway J. Engineers, Warrendale, PA. (1976) Amianthoid change: orientation of normal collagen fibrils during aging. Science. Laituri, T., Prasad, P., Sullivan, K., Frankstein, M., 194(4265):622-624. Thomas, R. (2005) Derivation and evaluation of a provisional, age-dependent, AIS 3+ thoracic risk Jackson RP, McManus AC. (1994) Radiographic curve for belted adults in frontal impacts. Paper analysis of sagittal plane alignment and balance in 2005-01-0297, Society of Automotive Engineers, standing volunteers and patients with low back Warrendale, PA. pain matched for age, sex and size. Spine 19:1611– 8. Lindahl, O. and Lindgren, A. (1967) Cortical bone in man – I. Variation in the amount and density with Kallieris, D., Mattern, R., McIntosh, A., Boggasch, F. age and sex. Acta Orthop. Scand. 38:133-140. (1992) The biofidelity of EuroSID 1 and BioSID. Paper 922518. Proc. 36th Stapp Car Crash Mallinger, R. and Stockinger, L. (1988) Amianthoid Conference. (asbestoid) transformation: electron microscopical studies on aging human costal cartilage. American Kent, R, Sherwood, C, Lessley, D, Overby, B, Journal of Anatomy. 181(1):23-32. Matsuoka, F. (2003) Age-related changes in the effective stiffness of the human thorax using four Marcus, J., Morgan, R., Eppinger, R., Kallieris, D., loading conditions. Proc. IRCOBI Conference on Mattern, R., Schmidt, G. (1983) Human response the Biomechanics of Impact. to and injury from lateral impact. Proc. 27th Stapp Car Crash Conference, pp. 419-432. Kent, R, Lessley, D, Sherwood, C. (2004) Thoracic response corridors for diagonal belt, distributed, McNelly, N. and Dittmer, J. (1976) Glomerular four-point belt, and hub loading. Stapp Car Crash basement membrane width and proteinuria in the Journal 48:495-519. aging hamster kidney. Exp. Geront. 11:49-55. Kent, R, Patrie, J. (2004) Chest deflection tolerance Mikic, B., Isenstein, A., Chhabra, A. (2004) to blunt anterior loading is sensitive to age but not Mechanical modulation of cartilage structure and to load distribution. Forensic Science International function during embryogenesis in the chick. 149(2-3):121-128. Annals of Biomedical Engineering 32(1):18-25. Kent, R., Henary, B., Matsuoka, F. (2005a) On the Milne, J. and Williamson, J. (1983) A longitudinal fatal crash experience of older drivers. Proc. study of kyphosis in older people. Age and Ageing AAAM (in press). 12:225-233. Kent, R., Murakami, D., Kobayashi, S. (2005b) Morris, A., Welsh, R., Frampton, R., Charlton, J., Frontal thoracic response to dynamic loading: the Fildes, B. (2002) An overview of requirements for role of superficial tissues, viscera, and the rib cage. the crash protection of older drivers. Proc. AAAM Proc. IRCOBI (in press). 46:141-56.

Kent et al. / Stapp Car Crash Journal 49 (November 2005) 19

Morris, A., Welsh, R., Hassan, A. (2003) chronic bronchitis and emphysema. Thorax 28:15- Requirements for the crash protection of older 23. vehicle passengers. Proc. AAAM 47:165-80. Smith, J. and Walmsley, R. (1957) Elastic after- Murakami, D, Kitagawa, Y, Kobayashi, S, Kent, R, effect, plasticity, and fatigue in bone. J. Anat. Crandall, J. (2004) Development and validation of 91:603. a finite element model of a vehicle occupant. Paper Stein, I. and Granik, G. (1976) Rib structure and 2004-01-0325, Society of Automotive Engineers, bending strength: an autopsy study. Calcif. Tissue Warrendale, PA. Research, 20:61-73. Nahum, A., Gadd, C., Schneider, D., Kroell, C. Vendantam R, Lenke LG, Keeney JA, et al. (1998) (1970) Deflection of the human thorax under Comparison of standing sagittal spinal alignment sternal impact. Paper 700400, Proc. 14th Stapp Car in asymptomatic adolescents and adults. Spine Crash Conference. 23:211–5. National Highway Traffic Safety Administration Viidik, A. (1979) Connective tissues – possible (1993) Traffic safety plan for older drivers. U.S. implications of the temporal changes for the aging Department of Transportation, Washington, DC. process. Mechanisms of ageing and development Neathery, R. (1974) Analysis of chest impact 9:267-285. response data and scaled performance Vitek, K. and Valenta, J. (2001) Age-related changes recommendations. Paper 741188, Proc. 18th Stapp in pressure-radius relation of human coronary Car Crash Conference. arteries. Proc. 39th International Conference of OECD (2001) Ageing and Transport – Mobility Experimental Stress Analysis, Tabor, Czech Needs and Safety Issues. Organization for Republic, pp. 355-356. Economic Co-operation and Development. Paris, Walsh, J., Webber, C., Fahey, P., Sharp, J. (1992) France. Structural change of the thorax in chronic Oskvig, R. (1999) Special problems in the elderly. obstructive pulmonary disease. J. Appl. Phys. Chest 115:158S-164S. 72(4)1270-1278. Puche, R., Morosano, M., Masoni, A., Jimeno, N. Wang, S. (2001). Personal Communication, Quarterly Bertoluzzo, S., Podadera, J., Podadera, M., CIREN Meeting, National Highway Traffic Safety Bocanera, R., Tozzini, R. (1995) The natural Administration, Washington, DC. history of kyphosis in postmenopausal women. Yamada, H. (1970) Strength of Biological Materials, Bone 17(3):239-246. Evans, F.G. (ed.), Williams and Wilkins Co., Schmidt, G., Kallieris, D., Barz, J., Mattern, R., Baltimore, MD. Klaiber, J. (1975) Neck and thorax tolerance levels Yeh, W., Li, P., Jeng, Y. et al. (2002) Elastic of belt-protected occupants in head-on collisions. modulus measurements of human liver and Paper 751149, Proc. 19th Stapp Car Crash correlation with pathology. Ultrasound in Med. & Conference. Biol. 28(4):467-74. Simon, G., Pride, N., Jones, N., Raimondi, A. (1973) Zhou, Q., Rouhana, S., Melvin, J. (1996) Age effects Relation between abnormalities in the chest on thoracic injury tolerance. Paper 962421, Proc. radiograph and changes in pulmonary function in 40th Stapp Car Crash Conference