Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Apr 1.
Published in final edited form as: Comput Methods Biomech Biomed Engin. 2020 Oct 29;24(5):517–526. doi: 10.1080/10255842.2020.1837119

Investigation of Femur Fracture Potential in Common Pediatric Falls using Finite Element Analysis

Keyonna McKinsey a,†, Angela Thompson b,*,†, Gina Bertocci a
PMCID: PMC8720240  NIHMSID: NIHMS1758758  PMID: 33115286

Abstract

A finite element (FE) femur model of an 11-month-old was developed to evaluate fracture risk in short-distance feet-first falls and bed falls. Pediatric material properties were applied to the FE model. Femur loading was derived from previously conducted fall experiments using a child surrogate where fall conditions (e.g. fall height, impact surface) were varied. Fracture thresholds based on principal stress and strain were used to examine potential for fracture. Peak stress/strain were significantly greater for increased fall heights and harder surfaces in feet-first falls. Feet-first falls exceeded some, but not all fracture thresholds. Bed falls did not exceed any thresholds.

Keywords: children, femur fracture, falls, biomechanics, computer model

Introduction

In 2017, an estimated 674,000 children were abused or neglected in the United States (U.S. Department of Health & Human Services et al. 2019). Since household falls are commonly provided as false explanations for abusive injuries, it is important for clinicians to distinguish between injuries resulting from abuse and those from true accidents. Bone fractures are the second most common presentation of abuse (Clarke et al. 2012), but are also common in accidental trauma.

The likelihood of fracture in children due to either accidental or abusive causes has been evaluated through retrospective clinical studies and biomechanical studies. Clinical studies compare femur fracture incidence in abuse and accidental causes and have shown that femur fractures are more likely due to abuse in cases where the child is non-ambulatory and that falls from a height are commonly offered as the cause of injury in cases of abuse (Dalton et al. 1990; Pandya et al. 2009). Retrospective clinical studies often involve unwitnessed falls or unclear histories that lack detailed fall descriptions. Thus, assessing femur fracture risk using clinical studies with unknown or lacking fall conditions can result in misleading conclusions. On the other hand, biomechanical studies allow for investigation of the likelihood of femur fracture under various fall conditions. These studies conduct experiments using surrogates or anthropomorphic test devices (ATDs) to simulate commonly reported causes of injury (Bertocci et al. 2003; Deemer et al. 2005; Thompson et al. 2013). However, assessment of fracture risk in these experiments has been limited by both the scarcity of pediatric femur fracture thresholds and the biofidelity of the ATDs. Thompson et al. (2018) modified a CRABI ATD for measurement of femur loading and improved biofidelity of the femur and used the modified CRABI to simulate common short-distance fall scenarios. The modified CRABI included two load cells at the proximal and distal boundaries of the diaphysis and a geometric representation of the diaphysis based off a CT scan of a child’s femur. Thompson et al. measured loads that were experienced in the femur during these falls, but the analysis was limited in that the femur diaphysis was solid (did not include a cortex) and loading was only measured at the load cells rather than the stress and strain distribution across the femur surface. In addition, Thompson et al. predicted fracture risk by comparing each load type (i.e. bending and compression) independently to fracture thresholds rather than examining fracture risk as a sum of the total combined loading experienced in these ATD fall simulations.

Finite element (FE) modeling has been used to evaluate various loading conditions such as lateral falls in elderly populations (Bryan et al. 2009; Bessho et al. 2009). However, FE models of infant femurs have been limited to the evaluation of simple loading scenarios such as pure bending or torsional loads (Li et al. 2015; Altai et al. 2018; Castro et al. 2019). In addition, these studies examining infant bone used adult mechanical properties. These infant FE studies had also used an adult tensile yield strain to evaluate fracture risk whereas a study by Ambrose et al. (2018) suggests that the yield strain for infants (3.97%) should be greater than that of the applied adult values (0.73%; Bayraktar et al. 2004).

Fracture risk assessment of pediatric bone has been limited due to the scarcity of pediatric bone specimens. Pediatric fracture thresholds are mostly based on properties of adult bone or that of older children which may not be valid for infants. The distinction between the properties of adult and infant bone is important due to the rapid changes in the structure of the bone especially in the first 12 months of age. Zimmermann et al. (2019) reported that there is enhanced remodeling around 1–2 years of age suggesting this remodeling is associated with increased ambulatory function. Although the sample size was limited, Zimmermann also reported markedly different mechanical properties for a 12-month old compared to adolescent bone including lower bone mineral density and transversely oriented collagen fibers in infant bone. The findings by Zimmermann et al. suggest that it may be inappropriate to use adult bone properties to represent 1-year old bone.

The current study will address two limitations that currently exist within the infant femur FE model literature: (1) examining loading conditions relevant to common real-world scenarios, i.e. short-distance falls; and (2) use of pediatric mechanical properties rather than adult properties along with pediatric-specific criteria to examine fracture risk. A pediatric femur model simulating complex loading associated with common fall scenarios while also using pediatric-specific mechanical properties can provide further biomechanical evidence as to whether short-distance falls are likely to result in a fracture. The purpose of this study was to determine the likelihood of femur fracture in an infant due to common fall scenarios, specifically feet-first falls and falls from a bed, using FE analysis.

Methodology

An FE model of an 11-month-old pediatric femur was developed to investigate femoral stress and strain under loading conditions associated with short-distance falls. Loads measured in prior fall experiments using an anthropomorphic test device (ATD) (Thompson et al. 2018) were evaluated using the FE femur model. Loading scenarios included feet-first falls from two different heights (119cm and 69cm measured from ground to ATD center of mass (COM), or 73cm and 23cm from ground to the bottom of the ATD feet, respectively) and falls from a horizontal surface (61cm from edge of platform to ground) representing a bed from a side-lying position. Two different impact surfaces, carpet and linoleum, were evaluated for each fall scenario. Fracture potential was evaluated by comparing stress and strain outcomes to published thresholds.

FE Femur Model Segmentation

A deidentified whole-body CT scan of an 11-month-old child was obtained from the University of New Mexico (UNM) Radiology-Pathology Center for Forensic Imaging. CT imaging was performed using a Philips Brilliance Big Bore scanner, with a scan resolution of 0.499 × 0.499 × 1.00 mm3. CT images of the femur provided the FE model geometry and was also the basis for the diaphysis geometry for the modified ATD femur assembly used in previous fall experiments (Thompson et al. 2018). The left femur was segmented in Mimics v15.0.1 (Materialise, Ann Arbor, MI) using a 156 Hounsfield Unit (HU) threshold to create the 3D femur model.

FE Meshed Model Development

A 10-node tetrahedral mesh was generated in Materialise 3-matic v7.01 from the 3D model created in Mimics. The outer surface of the 3D model was smoothed to minimize the number of elements while still retaining accuracy. The surface mesh was then converted to a volume mesh. A mesh refinement study was performed to minimize the number of elements needed, thus reducing computational resources required while also still maintaining geometric accuracy; maximum edge length was independently varied in the cortical and trabecular regions to account for differing geometric profiles. Edge length was reduced until stress and strain outcomes varied less than 10%. The final mesh (Figure 1) had a maximum edge length of 1 mm and 1.4 mm in the cortical and trabecular regions of the femur, respectively and had a total of 341,736 elements.

Figure 1.

Figure 1.

Open in a new tab

Anterior view of meshed femur model. The mesh consisted of 341,736 10-node tetrahedral elements.

Application of Material Properties

Material properties applied to the meshed femur model were based on the elastic modulus of cortical tibia specimens from 10–11 month old children as determined by Ambrose et al. (2018). The density was determined using the relation of modulus (E) to apparent density (ρapp; Equation 1) based on the equation by Morgan et al. (2003). A summary of the properties applied are provided in Table 1.

ρapp=(E6850) 0.671 (1)

Table 1.

Applied Material Properties

Property Value Reference
Apparent density (ρapp, kg/m3) 344.5 (Morgan et al. 2003)
Elastic modulus (MPa) 1400 (Ambrose et al. 2018)
Poisson’s ratio (ν) 0.3 (Schileo et al. 2007)
Open in a new tab

Boundary Conditions and Constraints

A coordinate system representative of the ATD femur load cell orientation used in previous experiments (Thompson et al. 2018) was applied to the meshed femur model to apply loads from fall experiments. Constraints implemented in the model corresponded to ATD femur constraints (Figure 2). The proximal constraint representing the hip was a universal joint located at a node (Figure 2; point A) where translation was fixed in all directions and rotation was constrained about the longitudinal (z) axis of the femur. At the distal end of the femur, displacement of nodes in the intercondylar region was fixed (Figure 2; region C).

Figure 2.

Figure 2.

Open in a new tab

Applied boundary conditions and constraints: fixed constraints were applied at the most superior point of the proximal metaphysis (A) and the intercondylar region (C). The forces and moments measured from ATD load cells were applied to groups of elements corresponding to each load cell location (B).

Applied Femur Loading Representative of Short-Distance Falls

Simulations were conducted using the transient structural module in ANSYS v17.1 (Ansys Inc., Canonsburg, PA). Femur loading measured in previously conducted ATD experiments (Thompson et al. 2018) was applied to the FE model at the experimental load cell locations (Figure 2; locations B). Loading time histories derived from the two 6-axis load cells were applied as measured except for the moments. The measured moments were adjusted such that forces that would induce a moment at the opposing load cell were accounted for in the applied loading. Loads were applied in the same manner for all loading scenarios since the loads were derived from the load cells integrated into the ATD femur in experimental fall simulations (Thompson et al. 2018). Simulated fall scenarios included vertical feet-first free falls and falls representing a side-lying child rolling off a horizontal platform, such as bed. Left femur loading from fall trials was selected since the left leg impacted the ground first in bed falls (Figure 3) and similar left and right leg dynamics were found in feet-first falls (Figure 4). ATD femur loading profiles were clipped to only include the primary impact of the ATD with the ground. Three ATD fall trials were randomly selected for each of the 4 fall scenarios (Table 2) for a total of 24 fall simulations. Large deformation effects were used in FE analysis.

Figure 3.

Figure 3.

Open in a new tab

Representative ATD position at impact for bed falls with different fall dynamics condition: either the upper (A) or lower (B) leg impacted the ground first.

Figure 4.

Figure 4.

Open in a new tab

Representative ATD position for feet-first falls at initial foot impact (A) and secondary pelvis impact (B).

Table 2.

Fall scenarios simulated using the FE model. Each scenario is a bed fall or feet-first fall and combination of condition 1 and condition 2 resulting in 4 scenarios for bed falls and 4 scenarios for feet-first falls. 3 trials of each scenario were simulated.

Fall Type Condition 1 Condition 1 Categories Condition 2 Condition 2 Categories Total Number of Trials
Bed Fall Impact Surface Carpet (n=3) Fall Dynamic Upper leg impacts first (n=3) 12
Linoleum (n=3) Lower leg impacts first (n=3)
Feet-First Fall Impact Surface Carpet (n=3) Fall Height* 69 cm (n=3) 12
Linoleum (n=3) 119 cm (n=3)
Open in a new tab
*

Fall heights were measured from the ground to center of mass of the ATD in the initial position. Fall heights as measured from ground to feet are 23cm and 73cm for the 69 and 119cm falls, respectively.

Validation

The FE model was validated using the output of a strain gauge located posterio-distally on the modified ATD femur shaft (Thompson et al. 2018). The FE model-predicted maximum principal strain at this location was compared to the experimental maximum principal strain. For the purpose of this validation, the applied material properties were that of the ATD femur shaft (AlSi10Mg). Five measures were used to compare the FE predicted and experimental strain curves: mean value ratio, difference in peak value and its time occurrence, Pearson correlation coefficient (r), coefficient of determination (r2), and area under the curve (Dsouza & Bertocci 2010; Vavalle et al. 2013). The acceptance criteria for validation were ≤ 20% difference between experimental and FE predicted values or 0.8 (Pipkorn & Eriksson 2003; Dsouza & Bertocci 2010). The validation was completed using one feet-first fall from 119cm onto carpet.

Evaluation of FE Predicted Outcomes

The femur diaphysis peak maximum principal stress and strain during the interval of applied loading were determined for each trial. Statistical analysis was completed using Minitab v18.1 (Minitab Inc., State College, PA). Descriptive statistics, such as the mean and standard deviation of the stress and strain, were compared across the fall scenarios. For each fall type (i.e. feet-first and bed), separate two-way ANOVAs (α = 0.05) were performed for each outcome measure (maximum principal stress or strain). For feet-first fall simulations, the factors assessed were impact surface and fall height, and in bed fall simulations, factors assessed were impact surface and fall dynamics. Tukey’s test was used for post-hoc analysis to identify significant differences (α = 0.05).

Fracture Potential Evaluation

Three criteria (Table 3) were used to determine if there was a potential for femur fracture. The first criterion was dependent upon the maximum principal strain theory (Schileo et al. 2008). This theory indicates failure (fracture) will occur when the maximum principal strain is greater than or equal to that of the yield strain (tensile or compressive). A tensile yield strain threshold of 3.97% for infant tibial bone (Ambrose et al. 2018) was used. Additionally, two stress-based criteria derived from mechanical testing of pediatric bone were also used (Table 3): ultimate tensile strength (Zimmermann et al. 2019) and ultimate flexural strength (Ambrose et al. 2018). Potential fracture was defined as having maximum principal stress or strain exceeding any criterion threshold in conjunction with exceeding an additional element criterion (Table 3).

Table 3.

Fracture Thresholds.

FE Model Predicted Outcome Threshold Description Threshold Value Additional Element Criterion
Maximum Principal Strain Yield Strain 3.97 % 1 A minimum of 5 contiguous elements must have exceeded the threshold.
Maximum Principal Stress Ultimate Flexural Strength 64.7MPa 1
Ultimate Tensile Strength 90MPa 2
Open in a new tab

Results

Validation

The maximum principal strain time history demonstrated good agreement between the FE model and fall experiment, aside from the initial impact (first peak) of the ATD’s feet (Figure 5, experimental peak at approximately 0.01 s). The model-predicted strain met the acceptance criteria (≤ 20% or ≥ 0.8) with respect to the experimentally measured strain (Table 4). The model was considered valid to be used for further evaluation.

Figure 5.

Figure 5.

Open in a new tab

Time history of maximum principal strain (με) (black, solid) of the corresponding strain gauge region in the FE model and the experimental strain (red) for a representative feet-first fall.

Table 4.

Validation results. Percent difference of FE prediction for the peak value, time occurrence of the peak, mean value, Pearson correlation (r), and coefficient of determination (r2) when comparing the FE prediction to the experimental maximum principal strain.

Peak Value (%difference) Time occurrence of peak (% difference) Mean Value (% difference) Pearson Correlation (r) Coefficient of Determination (r2)
2.4% 0% 3.5% 0.97 0.94
Open in a new tab

FE Model Predicted Stress and Strain

Figure 6 shows mean peak stress and strain values observed for each fall scenario. FE stress and strain outcomes were generally greater for feet-first falls than for bed falls, except for the lowest height falls (69cm) onto carpet, which were associated with the lowest stress and strain outcomes of all the fall scenarios.

Figure 6.

Figure 6.

Open in a new tab

The peak maximum principal stress and strain observed in the bed falls (A) and feet first falls (B). Error bars represent the range of values observed. In bed falls (a), fall dynamic is indicated by whether the lower or upper leg impacted first in the ATD experiment. n=3 for each fall scenario.

In ATD bed fall experiments, two different impact dynamics were observed: either the lower or upper leg impacted the ground first (Figure 3). However, there was no significant difference in stress and strain outcomes with respect to the bed fall conditions (fall dynamics and/or impact surface). Maximum FE-predicted outcomes were coincident with peak force in the medio-lateral direction following the initial impact.

In feet-first falls, significant differences in stress and strain outcomes were found for all fall conditions (fall height, impact surface, and interaction) (p<0.001). Falls from greater heights (119 cm) and/or onto linoleum generally led to greater outcomes compared to falls from lower heights and other surfaces. Peak stresses and strains in 119cm falls were significantly greater than those in 69cm falls (p<0.05). Stress and strain outcomes in 69cm falls onto linoleum were greater than those onto carpet (p<0.01); however, no significant differences were found in the 119cm falls onto different impact surfaces (p=0.986). For the 69cm feet-first falls and one 119cm fall onto linoleum, maximum FE-predicted outcomes were coincident with the peak bending moment which occurred during the secondary impact of the pelvis with the ground (Figure 4B). For the remaining 119cm falls (5/6; 83%), the maximum FE-predicted outcomes were coincident with peak torsional loads (8–10Nm) which occurred between initial (foot) and secondary (pelvis) impacts with the ground.

Fracture Potential

Eight feet-first falls exceeded at least one threshold and two of those exceeded 2 thresholds (Table 5). No falls exceeded the ultimate tensile strength threshold. All 119cm feet-first falls and two 69cm linoleum feet-first falls exceeded the tensile yield strain threshold. Two of the 119cm falls also exceeded the ultimate flexural strength threshold. None of the thresholds were exceeded in bed falls. Since a greater number of falls from greater heights and those onto linoleum exceeded thresholds, these fall conditions are likely to be associated with a greater potential for fracture.

Table 5.

Number of fall trials exceeding fracture criteria. (n=3 for each fall scenario).

Fall Type Fall Conditions Number of Trials Exceeding Threshold Number of Trials Exceeding All Thresholds
Tensile Yield Strain Ultimate Tensile Strength Ultimate Flexural Strength
Feet-First 119cm Linoleum 3 0 1 0
119cm Carpet 3 0 1 0
69cm Linoleum 2 0 0 0
69cm Carpet 0 0 0 0
Bed All conditions 0 0 0 0
Open in a new tab

Discussion

An FE model representing an 11-month-old child’s femur was developed and validated to evaluate fracture risk in simulated fall scenarios (horizontal bed falls and vertical feet-first falls). To the best of our knowledge, this study was the first to examine femur fracture risk in fall scenarios using FE analysis and the first infant FE femur model to use mechanical properties and fracture criteria based on infant bone. Fracture risk (indicated by FE predicted stress and strain outcomes) was greater in the 119cm vertical feet-first falls compared to falls from lower heights and bed falls. No falls exceeded all fracture criteria, but 8 of the 12 simulated feet-first falls exceeded at least one stress or strain threshold. In feet-first falls, falls from greater height and falls onto linoleum resulted in greater outcomes (and thus a higher fracture risk). There were no significant differences in outcomes or fracture potential in simulated bed falls associated with different impact surfaces or fall dynamics.

Two-thirds of the simulated feet-first falls exceeded the tensile yield strain fracture threshold, and two of those also exceeded the ultimate flexural strength threshold (Table 5). While five (83%) of 119cm falls exceeded at least one threshold, the maximum FE-predicted outcomes in these falls were associated with peak torsional loads which likely occurred due to extensive plantar flexion and internal rotation of the ankle. This dynamic appears unnatural and may be due to limitations in the biofidelity of the CRABI ATD ankle (Thompson et al. 2018). However, the increase in maximum principal stress and strain for falls from a greater height is consistent with studies evaluating fall height and injury (Wang et al. 2001; Chaudhary et al. 2018). While falls from a height may result in femur fractures, reported fractures are generally due to falls from heights greater than those in our study which was at most 119cm from the ground for the center of mass or 73cm from the ATD’s feet to the ground. Chaudhary et al. (2018) reported 19% of falls from a caregiver’s arms resulted in femur fractures for children less than 5 years old but did not specify the height of these falls. Wang et al. (2001) reported lower extremity fractures resulting from falls in children <15 years old: 24% were due to falls 4–15ft (122–457cm) and 76% occurred in falls from heights >15ft (>457cm). However, it is difficult to compare the results of this study to clinical studies since they often lack detail regarding history and injury mechanisms (e.g., fall type, fall height, impact surface, and fall dynamics). In addition to fall height, we also found greater injury potential in falls onto linoleum compared to falls onto carpet. This is consistent with ATD studies indicating injury risk may vary for different impact surfaces (Bertocci et al. 2003; Bertocci et al. 2004; Thompson et al. 2009; Ibrahim & Margulies 2010; Thompson et al. 2013).

No bed falls exceeded fracture criteria (Table 5); the peak FE model-predicted outcomes for these falls were approximately half of the tensile yield strain and ultimate flexural strength thresholds. This is consistent with published clinical studies where few femur fractures are reported due to bed falls or falls from other furniture, especially from a height of only 2ft (61cm). Chaudhary et al. (2018) and Pomerantz et al. (2012) both evaluated fall injuries resulting in hospitalization in children up to 4 and 5 years old, respectively. Pomerantz found femur fractures in 21% of cases involving falls from furniture. Similarly, Chaudhary reported 20% of falls from beds or couches in this hospitalized cohort resulted in femur fracture. Another study observed only 3% of falls from furniture (<90cm) in children up to 5 years old resulted in a fracture (Helfer et al. 1977). The disparity in fracture incidence between these studies is likely due to differences in data collection methods: Pomerantz and Chaudhary used hospital records which would skew toward more severe falls, while Helfer surveyed parents in a primary care setting to identify witnessed falls. Other studies of in-hospital falls report limited occurrences of fractures; Lyons and Oates (1993) reported only 2/207 falls resulted in fracture (1 clavicle, 1 skull) and Nimityongskul and Anderson (1987) reported 2/76 falls resulted in fracture (1 skull fracture, and 1 tibia fracture occurring in a child with osteogenesis imperfecta).

This study found a higher fracture risk in feet-first falls with high torsion loads (8–11Nm) and bending loads (9–10Nm). To our knowledge, only a few studies have investigated infant femur fracture through FE analysis (Li et al. 2015; Altai et al. 2018; Castro et al. 2019). Our findings are similar to Altai (2018) who estimated failure loads in FE femur models of infants and young children. Altai et al. reported failure loads of approximately 1–30 Nm under static torsion loading and approximately 1–28 Nm under static 4-point bending loads for children ranging from 0–3 years of age. For the 2 subjects around 1-year of age, failure loads of approximately 7–13 Nm and 7–10 were reported for torsion and bending, respectively. Whereas this study included complex, dynamic loading that was not purely bending (i.e. also included torsional and axial loading), Altai evaluated pure bending and pure torsional loads under static loading conditions. Altai and Li used adult bone material properties and employed a tensile yield strain criterion to identify fractures that was derived for adult bone (Bayraktar et al. 2004). Additional differences between Altai and Li’s femur FE models and ours include greater elastic moduli (peak 20GPa vs 1.4GPa in our study), use of a density-modulus relationship versus a uniform modulus, and a lower strain threshold (0.73% vs 3.97% used in our study). Castro et al. (2019) also evaluated infant FE femur models but for younger subjects (4- and 7-month-olds). Additionally, Castro et al. developed their model using a combination of CT and MRI scans which allowed for the inclusion of the epiphyses in the model. Comparing children of similar ages (4- and 7-month old) as in the Li and Altai studies, Castro predicted similar failure moments using the same application of mechanical properties and evaluation of fracture potential.

Our FE model has some limitations regarding the application of material properties used to represent pediatric bone. A uniform, linear model was used to characterize mechanical response since there are no validated mechanical models for pediatric bone response to loading. The primary limitation of the linear material model is neglecting the post-yield response of bone and any viscoelastic effects. While pediatric bone is more ductile than adult bone, characterization of this mechanical behavior is limited (Ambrose et al. 2018). Additionally, uniform application of material properties does not account for the dependency of the elastic modulus and the ultimate and/or yield stress on bone density, a relationship which has been employed in FE studies evaluating adult bone (Bessho et al. 2007; Trabelsi et al. 2011; Schileo et al. 2014). Single-value injury thresholds for infant bone were used rather than the stress-density relationship often used in adult bone studies; the youngest specimen in studies developing stress-density relationships was from a 4-year-old child (Ohman et al. 2011). A modulus-density relationship may exist for infant bone, but it is unclear whether the relationships defined for adult bone also apply to infant bone. Due to developmental differences demonstrated by Zimmerman et al. (2019), bone material properties of adult and older children are likely not extensible to infants who have lower bone mineral density and transversely oriented collagen fibers. Thus, using a uniform linear material model incorporating infant bone mechanical properties and single-value stress-thresholds provides an initial estimation of fracture potential.

Infant bone mechanical properties used in this study were based on quasi-static loading (Ambrose et al. 2018; Zimmermann et al. 2019) but femur loading in falls occurs dynamically. This may lead to an overestimation of fracture potential since mechanical behavior varies with strain rate. Increasing strain rates influence mechanical response of adult bone; changes with increasing strain rate include varying ultimate stresses (Hansen et al. 2008 found that ultimate stress increases under compressive loading and decreases under tensile loading at higher strain rates), more brittle behavior (Zioupos, Hansen, et al. 2008), and increasing elastic modulus (Carter & Hayes 1976; Hansen et al. 2008). Currently, characterization and extension of these observed behaviors to infant bone are limited. Bending stiffness has been demonstrated to be dependent on strain rate for immature porcine bone indicating that immature bone has similar strain-rate dependencies as mature bone (Cheong et al. 2017). Unlike stress-based thresholds, yield strain has been found to not vary with density and vary minimally at low strain rates (Morgan et al. 2003; Hansen et al. 2008; Zioupos, Cook, et al. 2008; Ohman et al. 2011). In addition, using a tensile yield strain criterion (versus stress-based criteria) has resulted in improved identification of fracture location and fracture potential in in vitro testing of adult bone when using a linear mechanical response model (Schileo et al. 2008). Because of the uncertainty of pediatric bone fracture thresholds, in this study we used both stress-based and strain-based thresholds for a conservative estimate of fracture potential.

Other model limitations include use of a clinical resolution CT scan and limitations related to the input loading. A lower resolution CT may result in reduced femur geometry accuracy compared to FE models developed from higher resolution scans (Schileo et al. 2007; Li et al. 2015; Altai et al. 2018). The 5 contiguous element criterion (Table 3) was included to reduce the effect of geometric modeling inaccuracies. In addition, limitations in applied femur loading due to the biofidelity of the modified CRABI ATD (Thompson et al. 2018) could affect the accuracy of fracture predictions. Biofidelity of ATD joints could influence fall dynamics and femur loading, which in turn can influence FE-model predicted stress and strain.

Conclusion

A pediatric femur FE model was developed and used to assess femur fracture potential in two common fall scenarios: bed falls and feet-first falls. This model is unique in that it applied complex, dynamic loading conditions measured in previous ATD studies simulating falls. To our knowledge, this study is the first FE femur model to use mechanical properties of infant bone and apply fracture criteria based on mechanical response of infant bone rather than adult bone. Fall conditions such as fall height and impact surface influenced potential for femur fracture; feet-first falls from 119cm resulted in the greatest potential for fracture. This FE analysis advances our understanding of infant femur fracture risk by providing a more accurate representation of the infant femur mechanical response by incorporating infant-specific mechanical properties and criteria. Future work should include validation of our FE model using mechanical testing of pediatric bone specimens.

Acknowledgements

This work was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number R03HD078491. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

Declaration of Interest Statement

The authors have no potential conflicts of interest to report.

References

  1. Altai Z, Viceconti M, Offiah AC, Li X. 2018. Investigating the mechanical response of paediatric bone under bending and torsion using finite element analysis. Biomech Model Mechanobiol. 17:1001–1009. [DOI] [PubMed] [Google Scholar]
  2. Ambrose CG, Soto Martinez M, Bi X, Deaver J, Kuzawa C, Schwartz L, Dawson B, Bachim A, Polak U, Lee B, Crowder C. 2018. Mechanical properties of infant bone. Bone. 113:151–160. [DOI] [PubMed] [Google Scholar]
  3. Bayraktar HH, Morgan EF, Niebur GL, Morris GE, Wong EK, Keaveny TM. 2004. Comparison of the elastic and yield properties of human femoral trabecular and cortical bone tissue. J Biomech. 37:27–35. [DOI] [PubMed] [Google Scholar]
  4. Bertocci GE, Pierce MC, Deemer E, Aguel F, Janosky JE, Vogeley E. 2003. Using test dummy experiments to investigate pediatric injury risk in simulated short-distance falls. Arch Pediatr Adolesc Med. 157:480. [DOI] [PubMed] [Google Scholar]
  5. Bertocci GE, Pierce MC, Deemer E, Aguel F, Janosky JE, Vogeley E. 2004. Influence of fall height and impact surface on biomechanics of feet-first free falls in children. Injury. 35:417–424. [DOI] [PubMed] [Google Scholar]
  6. Bessho M, Ohnishi I, Matsumoto T, Ohashi S, Matsuyama J, Tobita K, Kaneko M, Nakamura K. 2009. Prediction of proximal femur strength using a CT-based nonlinear finite element method: Differences in predicted fracture load and site with changing load and boundary conditions. Bone. 45:226–231. [DOI] [PubMed] [Google Scholar]
  7. Bessho M, Ohnishi I, Matsuyama J, Matsumoto T, Imai K, Nakamura K. 2007. Prediction of strength and strain of the proximal femur by a CT-based finite element method. J Biomech. 40:1745–1753. [DOI] [PubMed] [Google Scholar]
  8. Bryan R, Nair PB, Taylor M. 2009. Use of a statistical model of the whole femur in a large scale, multi-model study of femoral neck fracture risk. J Biomech. 42:2171–2176. [DOI] [PubMed] [Google Scholar]
  9. Carter D, Hayes W. 1976. Bone compressive strength: the influence of density and strain rate. Science (80- ). 194:1174–1176. [DOI] [PubMed] [Google Scholar]
  10. Castro APG, Altai Z, Offiah AC, Shelmerdine SC, Arthurs OJ, Li X, Lacroix D. 2019. Finite element modelling of the development infant femur using paired CT and MRI scans. PLoS One. 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chaudhary S, Figueroa J, Shaikh S, Mays EW, Bayakly R, Javed M, Smith ML, Moran TP, Rupp J, Nieb S. 2018. Pediatric falls ages 0–4: understanding demographics, mechanisms, and injury severities. Inj Epidemiol. 5:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cheong VS, Karunaratne A, Amis AA, Bull AMJ. 2017. Strain rate dependency of fractures of immature bone. J Mech Behav Biomed Mater. 66:68–76. [DOI] [PubMed] [Google Scholar]
  13. Clarke NMP, Shelton FRM, Taylor CC, Khan T, Needhirajan S. 2012. The incidence of fractures in children under the age of 24 months - In relation to non-accidental injury. Injury. 43:762–765. [DOI] [PubMed] [Google Scholar]
  14. Dalton HJ, Slovia T, Helfer RE, Comstock J, Scheurer S, Riolo S. 1990. Undiagnosed abuse in children younger than 3 years with femoral fracture. Arch Pediatr Adolesc Med. 144:875. [DOI] [PubMed] [Google Scholar]
  15. Deemer E, Bertocci G, Pierce MC, Aguel F, Janosky J, Vogeley E. 2005. Influence of wet surfaces and fall height on pediatric injury risk in feet-first freefalls as predicted using a test dummy. Med Eng Phys. 27:31–39. [DOI] [PubMed] [Google Scholar]
  16. Dsouza R, Bertocci G. 2010. Development and validation of a computer crash simulation model of an occupied adult manual wheelchair subjected to a frontal impact. Med Eng Phys. 32:272–279. [DOI] [PubMed] [Google Scholar]
  17. Hansen U, Zioupos P, Simpson R, Currey JD, Hynd D. 2008. The effect of strain rate on the mechanical properties of human cortical bone. J Biomech Eng. 130:011011. [DOI] [PubMed] [Google Scholar]
  18. Helfer RE, Slovis TL, Black M. 1977. Injuries resulting when small children fall out of bed. Pediatrics. 60:533–535. [PubMed] [Google Scholar]
  19. Ibrahim NG, Margulies SS. 2010. Biomechanics of the toddler head during low-height falls: an anthropomorphic dummy analysis. J Neurosurg Pediatr. 6:57–68. [DOI] [PubMed] [Google Scholar]
  20. Li X, Viceconti M, Cohen MC, Reilly GC, Carré MJ, Offiah AC. 2015. Developing CT based computational models of pediatric femurs. J Biomech. 48:2034–2040. [DOI] [PubMed] [Google Scholar]
  21. Lyons TJ, Oates RK. 1993. Falling out of Bed: A relatively benign occurence. Pediatrics. 92:125–127. [PubMed] [Google Scholar]
  22. Morgan EF, Bayraktar HH, Keaveny TM. 2003. Trabecular bone modulus-density relationships depend on anatomic site. J Biomech. 36:897–904. [DOI] [PubMed] [Google Scholar]
  23. Nimityongskul P, Anderson LD. 1987. The likelihood of injuries when children fall out of bed. J Pediatr Orthop. 7:184–6. [DOI] [PubMed] [Google Scholar]
  24. Ohman C, Baleani M, Pani C, Taddei F, Alberghini M, Viceconti M, Manfrini M. 2011. Compressive behaviour of child and adult cortical bone. Bone. 49:769–776. [DOI] [PubMed] [Google Scholar]
  25. Pandya NK, Baldwin K, Wolfgruber H, Christian CW, Drummond DS, Hosalkar HS. 2009. Child abuse and orthopaedic injury patterns: analysis at a level I pediatric trauma center. J Pediatr Orthop. 29:618–625. [DOI] [PubMed] [Google Scholar]
  26. Pipkorn B, Eriksson M. 2003. A method to evaluate the validity of mathematical models. In: 4th Eur MADYMO Users Meet. Brussels, Belgium. [Google Scholar]
  27. Pomerantz WJ, Gittelman MA, Hornung R, Husseinzadeh H. 2012. Falls in children birth to 5 years. J Trauma Acute Care Surg. 73:S254–S257. [DOI] [PubMed] [Google Scholar]
  28. Schileo E, Balistreri L, Grassi L, Cristofolini L, Taddei F. 2014. To what extent can linear finite element models of human femora predict failure under stance and fall loading configurations? J Biomech. 47:3531–3538. [DOI] [PubMed] [Google Scholar]
  29. Schileo E, Taddei F, Cristofolini L, Viceconti M. 2008. Subject-specific finite element models implementing a maximum principal strain criterion are able to estimate failure risk and fracture location on human femurs tested in vitro. J Biomech. 41:356–367. [DOI] [PubMed] [Google Scholar]
  30. Schileo E, Taddei F, Malandrino A, Cristofolini L, Viceconti M. 2007. Subject-specific finite element models can accurately predict strain levels in long bones. J Biomech. 40:2982–2989. [DOI] [PubMed] [Google Scholar]
  31. Thompson A, Bertocci G, Pierce MC. 2013. Assessment of injury potential in pediatric bed fall experiments using an anthropomorphic test device. Accid Anal Prev. 50:16–24. [DOI] [PubMed] [Google Scholar]
  32. Thompson A, Bertocci G, Smalley C. 2018. Femur loading in feet-first fall experiments using an anthropomorphic test device. J Forensic Leg Med. 58:25–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Thompson AK, Bertocci G, Pierce MC. 2009. Assessment of head injury risk associated with feet-first free falls in 12-month-old children using an anthropomorphic test device. J Trauma Inj Infect Crit Care. 66:1019–1029. [DOI] [PubMed] [Google Scholar]
  34. Trabelsi N, Yosibash Z, Wutte C, Augat P, Eberle S. 2011. Patient-specific finite element analysis of the human femur—A double-blinded biomechanical validation. J Biomech. 44:1666–1672. [DOI] [PubMed] [Google Scholar]
  35. U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children Youth and Families, Children’s Bureau. 2019. Child Maltreatment 2017. Washington, D.C. [Google Scholar]
  36. Vavalle NA, Jelen BC, Moreno DP, Stitzel JD, Gayzik FS. 2013. An evaluation of objective rating methods for full-body finite element model comparison to PMHS tests. Traffic Inj Prev. 14:S87–S94. [DOI] [PubMed] [Google Scholar]
  37. Wang MY, Kim KA, Griffith PM, Summers S, McComb JG, Levy ML, Mahour GH. 2001. Injuries from falls in the pediatric population: An analysis of 729 cases. J Pediatr Surg. 36:1528–1534. [DOI] [PubMed] [Google Scholar]
  38. Zimmermann EA, Riedel C, Schmidt FN, Stockhausen KE, Chushkin Y, Schaible E, Gludovatz B, Vettorazzi E, Zontone F, Püschel K, et al. 2019. Mechanical competence and bone quality develop during skeletal growth. J Bone Miner Res. 00:1–12. [DOI] [PubMed] [Google Scholar]
  39. Zioupos P, Cook RB, Hutchinson JR. 2008. Some basic relationships between density values in cancellous and cortical bone. J Biomech. 41:1961–1968. [DOI] [PubMed] [Google Scholar]
  40. Zioupos P, Hansen U, Currey JD. 2008. Microcracking damage and the fracture process in relation to strain rate in human cortical bone tensile failure. J Biomech. 41:2932–2939. [DOI] [PubMed] [Google Scholar]

RESOURCES