- Email: [email protected]
Femur fractures in relatively low speed frontal crashes: the possible role of muscle forces
Accident Analysis and Prevention 34 (2002) 1 – 11 www.elsevier.com/locate/aap
Femur fractures in relatively low speed frontal crashes: the possible role of muscle forces Allan F. Tencer a,*, Robert Kaufman e, Kathy Ryan e, David C. Grossman d,e, M. Brad Henley a, Fred Mann c, Charles Mock b, Fred Rivara d,e, Stewart Wang f, Jeffery Augenstein g, David Hoyt h, Brent Eastman i, Crash Injury Research and Engineering Network (CIREN) a a
Department of Orthopedics, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA b Department of Surgery, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA c Department of Radiology, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA d Department of Pediatrics, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA e Harbor6iew Injury Pre6ention and Research Center, Uni6ersity of Washington, Seattle, WA 98104, USA f Uni6ersity of Michigan, Ann Arbor, MI, USA g Uni6ersity of Miami, Miami, FL, USA h Uni6ersity of California, San Diego, CA, USA i Scripps Institute, San Diego, CA, USA Received 20 July 2000; received in revised form 6 October 2000; accepted 9 October 2000
Abstract In a sample of relatively low speed frontal collisions (mean collision speed change of 40.7 kph) the only major injury suffered by the partly or fully restrained occupant was a femur fracture. However, femur load measurements from standardized barrier crash tests for similar vehicles at a greater speed change (mean of 56.3 kph) showed that in almost all the cases, the occupant’s femur would not have fractured because the loads were below fracture threshold. In order to address this discrepancy, the load in the femurs of the occupants in the crash sample were estimated and compared with the femur fracture threshold. Femur load was estimated by inspecting the scene and measuring deformations in each vehicle, defining occupant points of contact and interior surface intrusion, and calculating crash change in velocity and deceleration. From this data, the measured femoral loads from standardized crash test data in a comparable vehicle were scaled to the actual crash by considering crash deceleration, occupant weight, and restraint use. All the occupants (7 males, average age 26.7 years, 13 females, average age 36 years) sustained at least a transverse midshaft fracture of the femur with comminution, which is characteristic of axial compressive impact, causing bending and impaction of the femur. However, the estimated average maximum axial load was 8187 N (S.D. = 4343N), and the average probability for fracture was only 19% (based on the femur fracture risk criteria). In 13 crashes the fracture probability was less than 10%. Two factors were considered to explain the discrepancy. The occupant’s femur was out of position (typically the driver’s right front leg on the brake) and did not impact the knee bolster, instead hitting stiffer regions of the dashboard. Also, since most victims were drivers with their foot on the brake to avoid the collision, additional compressive force on the femur probably resulted from muscle contraction due to bracing for impact. Adding the estimated muscle load on the femur to the estimated external load increased the femur loads beyond threshold, explaining the fracture in all but one case. Since crash tests using dummies cannot simulate out of position occupants or muscle contraction loading, they may underestimate the total load acting on the femur during actual impacts where the driver is bracing for the crash. These results may have implications for altering knee bolster design to accommodate out of position occupants and the additional load caused by muscle forces during bracing. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Femur fracture; Frontal crashes; Biomechanics
* Corresponding author. Tel.: +1-206-3415601; fax: + 1-206-7313227. E-mail address: [email protected] (A.F. Tencer). 0001-4575/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0001-4575(00)00097-X
2
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
1. Introduction Fractures of the femur due to motor vehicle impacts can result in extensive medical care, and if they have an intra-articular component, may result in permanent disability (MacKenzie et al., 1993a,b). Two general approaches have been taken for studying the mechanisms of femur fractures in frontal collisions. Risk factor analyses have shown that occupant compartment intrusions act as direct mechanisms for contact injuries of the femur (Siegel et al., 1993; Daffner et al., 1988), that females are at higher risk Dischinger et al., 1995), and that seatbelts and airbags are not particularly effective in reducing femur injury risk (Dischinger et al., 1995). An extensive study by Karlson et al. (1998) showed that a frontal crash direction, being female, having a head-on vehicle to vehicle as opposed to a single-vehicle to-fixed-object collision, being in a smaller sized vehicle, and not using seatbelts were risk factors for serious femur fractures. However, the specific mechanisms of the injuries, including contact points of the occupant’s femur were not determined in that study. Extensive biomechanical studies of the cadaveric femur fracture mechanisms and forces in frontal impacts have been performed. A summary of the findings are given in Table 1 (Melvin et al., 1975; Roberts et al., 1987; Cheng et al., 1984; Leung et al., 1983; Powell et al., 1975; King et al., 1973; Morgan et al., 1990; NHTSA, 2000a) and shown in the Injury Risk Function Curve, Fig. 1, derived from those studies, relating femur axial compressive load to the probability of fracture (NHTSA, 2000b). The studies showed, from dynamic measurements, that the approximate range of axial compressive fracture load for cadaveric femurs was 7500–15000 N, with an accepted threshold load of 8900 N (Donnelly and Roberts, 1987). The probability
Fig. 1. Reproduction of the femur fracture injury risk curve published by the National Highway Traffic Safety Administration.
of fracture risk at these force levels was, from Fig. 1, about 6% at 7500 N and 90% at 15000 N. In an ongoing study of motor vehicle crashes, we have noticed that a significant proportion of relatively low speed change and deceleration frontal crashes resulted in isolated fracture of the occupant’s femur. The mechanism in each case, based on appearance of the fracture (a transverse fracture line indicating failure by bending combined with comminution indicating compression), was axial compressive loading by impact with the dashboard. The goal of the study was to compare the best estimate of the external force acting in the femur during impact in these crashes with two published criteria defining femur fracture tolerance, since by the first estimates the accelerations were low and the resulting femur loads below those which would produce this fracture. If the femur loads were below thresholds, the explanation for the fracture might lie in the additional load generated by internal muscle contractions, not normally simulated in dummy or cadaveric crash testing and out of position contact with the dashboard.
2. Methods
2.1. Hypothesis The hypothesis to be tested in this study was that for front seated, partly or totally restrained occupants in a sample of frontal crashes, who sustained an isolated femur fracture, the external force generated in the crash which acted on the occupant’s femur reached at least the threshold for fracture.
2.2. General approach The motor vehicle crash information included in this study was collected from CIREN (Crash Injury Research and Engineering Network, National Highway Transportation Safety Administration) Centers. Crashes in the CIREN database are sampled based on the fulfillment of several criteria. Among these are that the occupant must have been restrained and that an injury of AIS 3 or greater must have occurred. Each case was reviewed by a multidisciplinary team consisting of a crash investigator, a bioengineer, a research nurse, and the treating physicians. Each center employs a crash investigator who has been certified in vehicle crash reconstruction and data collection through the training program given by the Transportation Safety Institute’s National Automotive Sampling System (NASS). Each crash scene and vehicle investigation conducted by CIREN centers follow the data collection format established by NASS. Once the crash information was analyzed, and the injuries and hospital course documentation were obtained, the crash was reviewed
Had femur load cell Knee or supra condylar FMVSS 208 Patella, supracondylar femoral shaft fx Higher at 1–2 ms pulse Represents 5–95% probability Axial/obl 15 (10/5)
6.7–9.2 ms
10013–7565 4997–15998
7925 (1482) 7552 (2537) 7565 2.8–10.5 m/s Axial Axial 11 (9/2) 11 (10/1)
Unembalmed (21–63) Unembalmed (34–73) Unembalmed Embalmed
None, distal 1/3, supracondylar Knee or supra condylar 15045 (18343) 15588 (5927) 21–43 13.2–39.3 Axial Axial 22 (16/6) 18 (18/0)
841661 831629 770925 751160 730984 1990 IRCOBI
Melvin et al. (1975) Roberts et al. (1987) Cheng et al. (1984) Leung et al. (1983) Viano (1977) Powell et al. (1975) King et al. (1973) Morgan et al. (1990) 751159 876042
Unembalmed (45–90) Unembalmed (55–71)
Peak force mean (S.D.) (N) Velocity (ft/s) Direction Number tested (m/f) Author
Specimen type (age range)
3
by the multidisciplinary team to establish a probable mechanism. Of the cases reviewed, a subset of 20, involving a frontal collision and a femur fracture as the predominant injury, were selected for this study.
2.3. Case selection criteria The crashes selected all involved frontal or nearly frontal collisions (+20° to −10°). These included vehicles, which were impacted by other vehicles but also included about 1/3 that impacted trees, poles, barriers, or other stationary objects. The major resulting injury was limited to an isolated femur fracture, usually involving the midshaft but in some cases, including additional injuries about the knee or hip.
2.4. Crash in6estigation
SAE paper
Table 1 Summary of data available in the literature related to axial compressive fracture loads of human cadaveric femora (N = Newtons)
Type of fx produced/comments
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Each crash site had scaled documentation of the roadway, traffic controls, road surface type, conditions, and road grade at both pre- and post- impact locations. Physical evidence such as tire skid marks were located and referenced to establish the heading angle and post impact trajectory of the colliding vehicles. A scaled drawing with impact and final rest positions was completed to assist in calculation of the speed and force at impact. Exterior inspections of the vehicle were performed, which included detailed measurements of the direct and induced damage. For this study, all the crash damage involved the front of the vehicle. With a contour gauge, a damage crush profile was collected from the front bumper and a specific Classification Deformation Code (CDC), which includes the principal direction of force (PDOF) was assigned. These measurements were entered into crash analysis software (Win SMASH, U.S. Dept of Transportation) to calculate the change in velocity (Delta V) of the vehicle during impact and the energy dissipated during the crash event. An inspection of the interior of the vehicle from which the injured person had been removed was performed to determine points of contact and restraint system use. This inspection also included assessing the integrity of the passenger compartment, status of glazing, and intruding components. The exact locations and all evidence of contact by the occupant who had sustained the femur fracture with the interior surfaces were documented. The lower extremity contact evidence consisted of dents in the padding, fabric transfers, and minor cracking of instrument panels or knee bolsters. The dash panel was however not disassembled to identify deformation to the knee bolster. An examination of the restraint system was performed, including lap and shoulder belts and the air bag, if available, to confirm use by the injured occupant.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
4
2.5. Estimation of occupant femur load on impact
For a lap belt only restrained occupant with single femur contact, (add load from shoulder belt to femur):
We used data from the same vehicle as that involved in the actual crash, reported by the New Car Assessment Program (NCAP, US National Highway Traffic Safety Administration) testing in 56 kph frontal collisions with two front seat fully restrained instrumented anthropometric dummies. A set of data was produced for the specific vehicles of our collision sample, including change of velocity of the vehicle at impact, crush deformation of the vehicle, average deceleration during impact, and load on the femurs, lap belt, and shoulder belt, all measured parameters in these tests. The data set used, derived from the NCAP tests, is shown in Table 2. The relationships used to derive actual crash occupant femur load are given below. The average deceleration of the vehicle in the actual crash is: A = 3.93×
! " (DV)2 S
(1)
where, A is the average deceleration of vehicle (g); 3.93, unit conversion factor; DV, change in velocity of the vehicle (kph); S, average vehicle frontal crush distance (mm). The force acting on the occupant’s femur in the actual crash is scaled for weight and deceleration compared to the test crash conditions
(2)
where, Fo is the estimated axial force in femur of occupant in crash (N); Fd, the measured axial force in femur of dummy in test (N); Ao, the average vehicle deceleration in crash (g), from Eq. (1); Ad, the average vehicle deceleration in crash (g); Wo, the weight of occupant in collision; Wd, weight of dummy in test. The force in the occupant’s femur (Fd) depends on the type of restraints used. Considering a worst case, if the occupant was partly restrained, for example shoulder belt only, all the force normally taken by the lap belt was added to the femur force. The following conditions were considered. For a lap and shoulder belt restrained occupant with two femur contact, the mean femur load is the average of the loads recorded in each femur of both dummies for that test vehicle.
Fll +Flr +Frl + Frr 4
(2a)
For a lap and shoulder belt restrained occupant with single femur contact. Fd =
Fll +Flr +Frl + Frr 2
Fll + Flr + Flr + Frr 2Fsl+ 2Fsr + 2 2
(2c)
For a shoulder belt or airbag only restrained occupant with single femur contact, (add load from lap belt to femur): Fd =
Fll + Flr + Frl + Frr 2Fpl + 2Fpr + 2 2
(2d)
Where, Fll, Flr, Frl, Frr, are the measured femur loads from crash test for that vehicle; ll, the left dummy-left femur; lr, the left dummy–right femur; rl, the right dummy-left femur; rr, the right dummy-right femur; Fsl, Fsr, the measured shoulder belt load from crash test; sl, the shoulder belt left dummy; sr, the shoulder belt right dummy; Fpl, Fpr, the measured lap (pelvic) belt load from crash test; pl, the lap (pelvic) belt left dummy; pr, lap belt right dummy.
2.6. Probability of axial loading exceeding fracture le6el In order to assess the potential for femoral axial impact forces to have reached levels consistent with femur fracture, we employed two criteria. The probability of femur fracture, based on Fig. 1, (NHTSA, 2000a) is: p= (−1E− 10)Fo3 + (5E − 06)Fo2 − (0.0413)Fo
Wo Ao × Fo = Fd × Wd Ad
Fd =
Fd =
(2b)
+ 98.133, (r 2 = 0.9946, range p= 1− 95%)
(3)
where p, is the probability of fracture and Fo, the occupant femur force. Based on the criteria given in the literature, for impact durations greater than 20 ms, the threshold force for fracture is (Donnelly and Roberts, 1987): Fo \ 8900 N
(4)
3. Results A total of 20 crashes were studied, shown in Table 3. In the sample of occupants who sustained an isolated femur fracture, there were 13 females with an average age of 36 years, weight of 177 lbs (80.5 kg), and height of 64.8 in (1.65 m). There were seven males, with an average age of 26.7 years, weight of 179 lbs (81.5 kg), and height of 69.5 in. (1.77 m). In the sample, Table 4, the occupants used the following restraint types; six were restrained with shoulder and lap belts and an air bag; two used shoulder and lap belts, two used shoulder belts and an airbag; five used a shoulder belt only; one used a lap belt only, and three used an airbag only. None were completely unrestrained. Those using a shoulder belt with an airbag or a shoulder belt only
Table 2 Data from New Car Assessment Program (NCAP) tests used to develop relationship between vehicle change in velocity, crush distance, and total load on both femurs of occupant in frontal collision Year of vehicle in study
Year of vehicle in crash test
Closing speed (kph)
Damage to front of vehicle (mm)
Average deceleration (g)
Lap and shoulder, belted occupant, double femur contact load (N)
Hyundai Accent Honda Civic Toyota Celica Toyota Corolla Toyota Corolla Geo Prizm Mercury Sable Ford Escort Chevy Blazer Toyota Corolla VW Jetta Ford Taurus Acura Integra Mercury Tracer Ford Escort VW Jetta Nissan Maxima Toyota Camry Ford Mustang VOLVO 740 GLE
1995
1995
56
433
28.5
4278
1994 1990 1992
1994 1990 1993
56.5 55.8 56.3
577 502 489
21.7 24.4 25.5
1992
1993
56.3
489
1991 1991
1990 1988
56.5 56.5
1995 1996 1990
1995 1997 1991
1996 1991 1996 1994
Lap and shoulder, belted occupant, single femur contact load (N)
Lap belted occupant, single femur contact load (N)
Shoulder belted occupant, single femur contact load (N)
Airbag only occupant single femur contact load (N)
Unbelted occupant single femur contact load (N)
8557
14549
14023
14023
15737
3179 4199 4682
6359 8398 9364
15981 16056 16444
15060 12851 12043
15060 12851 12043
21503 16310 14442
25.5
4682
9364
16444
12043
12043
14442
502.9 474
24.9 26.5
4218 5597
8436 11194
16961 17684
12467 15818
12467 15818
16774 16711
56.4 56.3 56.3
502.7 556 491.1
24.9 22.4 25.4
5270 3658 5675
10541 7315 11350
19512 15416 23137
15497 13165 13580
15497 13165 13580
19198 17609 19693
1994 1991 1995 1989
56.2 56.3 56.3 56.5
536 502.9 455.2 493.6
23.2 24.8 27.4 25.4
5734 5039 5214 4498
11469 10077 10428 8997
18786 16827 18317 18213
14629 14032 14993 13125
14629 14032 14993 13125
16212 15743 17669 17843
1993 1991 1995
1994 1988 1995
56.3 46.8 46.5
412.5 391.2 292.7
30.2 22.0 29.0
5235 5037 5729
10470 10075 11458
16707 10075 11458
13058 10075 11458
13058 10075 11458
14059 5037 5729
1990
1992
47.6
550.3
16.2
3959
7918
7918
7918
7918
3959
1988
1987
56.7
590.6
21.4
3759
7518
14535
12524
12524
15782
1990
1988
56
574.9
21.4
4774
9548
17813
14728
14728
18219
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Vehicle type
5
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
6
Table 3 Demographics of the occupants involved in the crashes studied in which the occupant sustained an isolated fracture of the femura Case number
M/F
Age
Weight (lbs)
Height (in)
Vehicle class
Vehicle weight (lbs)
Vehicle type
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
F F M F F F F M M F M M F F M F F F M F
21 21 26 58 24 33 80 38 26 15 26 30 39 22 20 27 48 25 21 55
155 128 160 155 125 260 150 262 165 90 180 200 179 200 150 264 178 146 138 245
66 69 68 64 66 66 64 70 ? 56 67 73 70 63 70 66 61 65.5 69 66
01 02 01 01 01 01 03 01 14 01 01 03 02 01 01 01 03 02 02 03
2101 2313 2696 2253 2253 2435 3131 2404 3993 2330 2647 3276 2628 2393 2371 2275 3001 2690 2818 2854
95 94 90 92 92 91 91 95 96 90 96 91 96 94 93 91 95 90 88 90
a
Hyundai Accent Honda Civic Toyota Celica Toyota Corolla Toyota Corolla Geo Prism Mercury Sable Ford Escort Chevy Blazer Toyota Corolla VW Jetta Ford Taurus Acura Integra Mercury Tracer Ford Escort VW Jetta Nissan Maxima Toyota Camry Ford Mustang Volvo 740
Class, 01, subcompact; 02, compact; 03, intermediate; 14, compact utility.
Table 4 Vehicle crash data from the 20 crashes studied in which the occupant sustained a femur fracturea Case no
Restraint use
Delta V (mph)
PDOF (°)
Max crush (in.)
Average decel (g)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Shoulder, lap belt, airbag Airbag Shoulder, lap belt, airbag Shoulder belt only Lab belt only Shoulder, lap belts Shoulder, lap belts,airbag Shoulder belt, air bag Shoulder, lap belts,airbag Shoulder belt only Airbag only Shoulder, lap belts,airbag Airbag Shoulder belt, air bag Shoulder belt only Shoulder belt only Airbag Shoulder belt only Lap and shoulder belt Lap, shoulder belts, airbag
34 37 24 22 22 43 12 27.8 10 20 21.1 32 36 24 29 12 22 19.4 30 28.9
0 5 0 0 0 0 0 0 −10 −10 0 0 10 20 −10 −10 0 10 0 −10
24.4 22.4 17.7 19.3 19.3 43.7 8.7 38.8 n/a 10 16.1 30.7 17.7 24.4 28.3 11.75 13 16.5 15 37.75
19.0 24.4 13.0 10.0 10.0 16.9 6.6 8.0 n/a 16.0 11.1 13.3 29.3 9.4 11.9 4.9 14.9 9.1 24 8.8
a
PDOF, principal direction of force, (0°, 12 O’clock, direct frontal; −10°, 11 O’clock from left; 10°, 1 O’clock, from right.
were in vehicles with automatic shoulder belt devices, in which the lap belt had to be attached manually. All sustained at least a midshaft femur fracture, with five having additional supracondylar or femoral neck involvement. Contact points of the injured femur of the occupant with some part of the instrument panel, are shown in Table 5.
The principal direction of force ranged from +20° to − 10°, with 11 crashes having an estimated direction of 0° (directly frontal). The average delta V of the collisions was 25.3 mph (40.7 kph), (S.D.= 8.5 mph (13.7 kph)), with an average deceleration of 13.7 g (S.D.= 6.4 g), Table 3. The average vehicle curb weight of the cars involved was 2648 lbs (1203 kg) (S.D.=419
Table 5 Estimated load at fracture, contact points and injury sustaineda Weight (lbs)
Estimated femur load (N)
19 24.4 13 10 10 16.9 6.6 8 N/a 16 11.1 13.3 29.3 9.4 11.9 4.9 14.9 9.1 24 8.8
155 128 160 155 125 260 150 262 165 90 180 200 179 200 150 264 178 146 138 245
6722 13 540 4747 6720 7493 10 029 2260 9087 6243 8662 6071 22 737 8150 7739 5608 11 498 5760 7559 4921
a
\8900 N Yes/no
Lower leg contact points
Intru
Contact evidence
Femur fracture type
8 80 3 8 9 27 0 17
N Y N N N Y N Y
3 14 3 100 9 8 1 50 2 7 1
N N N Y N N N Y N N N
Knee bolster /steer col Knee bolster /steer col Knee bolster /ign cyl Inst panel /center dash Glove box at latch Inst panel/center dash Knee bolster /steer col Knee bolster Knee bolster Inst panel/center dash Knee bolster Knee bolster/steer col Knee bolster Knee bolster/center dash Inst panel/steer col Inst panel Knee bolster/steer col Inst panel Inst panel Knee bolster
None 2’’ None None None None None 4’’ 6’’ None None 2’’ None None 5.5’’ None None None None None
Scuffed Cracked, scuffed Deformed Scuffed Cracked Scuffed Small dent Scuffed Scuffed, deform Scuffed Deformed Scuffed Scuffed None Deformed Small dent Scuffed Dent Small dent Scuffed, dent
Trans, midshaft Trans, midshaft Trans, midshaft Trans, midshaft Trans, midshaft Midshaft, troch Medial cond Supracond dist Trans, midshaft Trans, midshaft Trans, midshaft Midshaft, neck Trans, midshaft Comminuted mid Comminuted mid Comminuted Femoral neck Trans, midshaft Comm shaft Comminuted
Probability (%)
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Average deceleration (g)
S, shoulder belt; L, lap belt; A, airbag seats; D, drivers; RF, front passenger’s.
7
8
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
lbs (190 kg)), Table 4. The estimated load on the occupant’s femur during the impact, based on Eqs. (1) and (2), are given in Table 5, along with the calculated probability of fracture from the equation (Siegel et al., 1993). The observed points of contact of the occupant’s lower limb and the resulting fracture type are also given in Table 5. The mean best estimate of the femur load was 8187 N (S.D.=4343 N) for the 19 cases for which complete data were available, with a mean probability of 19% that the femur load was in the range for fracture, based on Eq. (3) and Fig. 1. Of the sample, 13 had a probability of fracture less than 10%. Based on the criterion that an axial compressive force greater than 8900 N is required for fracture, the estimated force exceeded this in only five cases. Eighteen of the 20 occupants were drivers and of these, 13 fractured their right femur, Table 5. The remaining two occupants were seated in the right front passenger’s seat and both fractured their left femurs. The specific contact component only intruded in five of the cases and in two cases, the intrusion was minimal. In three cases the toe pan intruded, acting on the lower leg, but for the majority, the intrusion did not appear to play a role in the axial loading on the femur. The contacts to the knee bolster systems and the lower instrument panels consisted of only scuffs, small cracks or dents, with minimal observable deformation, although elastic rebound of the padding could have occurred as shown in Fig. 2. The narrow area of contact for the driver’s right knee, between the steering column and the center of the dashboard, was the location of contact for eight drivers who fractured their right femurs. Some drivers’ right knees contacted at or near the steering column and others became cornered into the vertical center part of the dashboard. Some knee contacts were with other rigid components such as the ignition cylinder and the lock/latch of the glove box. Overall, for half of the occupants, their knees appeared to strike the stiffer areas or became cornered within the center of the dashboard, with the area of contact not appearing to yield to the force of impact.
4. Discussion In this study, femur, lap belt, and shoulder belt forces measured during standardized frontal collisions performed as part of the New Car Assessment Program were used to develop relationships between vehicle closing velocity, crush distance, and the probable axial compressive load from direct femur contact with the dashboard, for different conditions of occupant restraint. These relationships were used in assessing the estimated probability of femur fracture by axial compressive loading in a series of frontal crashes in which
the major injury to the occupant was an isolated femur fracture. We found the mean probability for femur fracture by axial loading in these crashes to be 19%. Thirteen of the 19 in the group had a probability under 10%, based on the femur load-fracture risk curve published by NHTSA, and 14 of the 19 were below the 8900 N threshold. Although the estimated probability was low, all the fractures appeared to be transverse with comminution, which would be expected to result from axial impact. Therefore, despite the relatively low external forces, direct axial contact was the probable mechanism for these injuries. In the majority of cases the criterion for femur fracture force that we estimated was below accepted thresholds, and yet the femur fractures were characteristic of those due to axial compressive impact, therefore other axial forces must have been present. Muscle loading was not accounted for in estimating the compressive loads acting on the occupants’ femurs in this study. In a frontal crash when the driver forcefully presses on the brake and braces for the impact, the knee flexors and extensors are tensed. Estimates based on muscle crossectional area and muscle power demonstrate that these muscles can generate significant compressive loads on the femur. Quasistatic measurements of maximum voluntary flexor and extensor muscle torques (Borges, 1989; Lord et al., 1992; Murray et al., 1985), show that a mean extensor muscle torque around the knee for 26-year-old males (the average age of the male subjects in our study) is about 273 Nm. With an approximate moment arm of 5 cm about the knee, this moment results in a compressive force of 5383 N acting along the axis of the femur. For 36-year-old females (the average age in our study for the female subjects), the extensor muscles force acting on the femur is about 3356 N. The knee flexors produce about 3002 N for males and 1801 N for females. It is impossible to determine specifically the extent of muscle contraction that the occupants of the vehicles in the crashes that we studied actually exerted, but it is entirely reasonable to assume that forceful braking and anticipatory bracing for the impact will produce considerable contraction of the knee flexor and extensor muscles. This appears to be the most reasonable explanation for the discrepancy between the mechanism of the femur fracture, axial loading, and the relatively low loads predicted. Review of the studies describing femur ultimate loads (Melvin et al., 1975; Roberts et al., 1987; Cheng et al., 1984; Leung et al., 1983; Powell et al., 1975) do not appear to have considered dynamic muscle loading as a potential contributor to the overall load that the femur could be subjected to during a crash. For comparison, the predicted loads, without and with the contributions of muscles (taken as the force generated by the knee flexors and an equal load from the extensors at maximum contraction), are shown in Fig. 3.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
9
Fig. 2. Example documentation of a frontal crash. (a) Vehicle exterior deformation; (b) interior contact point near steering column.
This observation, if correct, has significant implications for the design of dashboards and knee bolsters. If internal muscle loads add considerably to the external loads experienced by the occupant’s femur in the crash, then the dashboard or knee bolster, if designed for an external load criterion of 8900 N, and tested using in position crash dummies, may be too stiff to prevent femur injury in some crashes involving human occupants, especially those at relatively low speeds. Considering also that many of the occupants’ knees did not strike the relatively less stiff part of the knee bolster, consideration should be given to (a) extending the
contact surface of the lower dashboard to include the junction with the steering column, and (b) reducing the crush load of the knee bolster to accommodate the internal component of the total force acting on the femur due to muscle loading. The assumptions upon which this study is based require discussion. The values provided for the femur forces experienced by occupants in the collisions we studied represent our best estimates, working from the quantitative data of the NCAP test results and considering the weight of the occupant, average deceleration in the frontal collisions, the contact points of the occu-
10
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Fig. 3. Comparison of the estimated loads acting on the femur due to external forces of contact with the dashboard during the crash, and combined with possible muscle contraction forces (considering males and females) due to muscle tensing around the knee resulting from braking and/or bracing for the impact (femur fracture threshold, 8900 N, indicated by dashed line).
pant with the interior of the vehicle, and the variable use of occupant restraints. The crash is a dynamic event with loads in the shoulder and lap belts and femur contact loads reaching peak values at slightly different times during the crash. In this case, we made the simplifying assumption that all restraint and femur loads reached their peak values together and that the forces acting on the occupant were all essentially horizontal, preventing forward movement, and therefore, acting in the same direction. Inspection of individual time histories from NCAP tests showed that peak loads occurred close together in time, so this assumption is probably reasonable. A second assumption in the analysis was to add the loads normally carried by components of the restraint system, if not used, to the load acting on the femur. In fact, this component of the total load was probably distributed to the other parts of the restraint system. A very important question relates to how well the hybrid III dummy represents the human femur in terms of its ability to predict femur loading during impact. The hybrid III dummy femur load measurement is used as a criterion for determining the potential for femur fracture during NCAP 35 mph frontal collisions. Donnelly and Roberts (1987) used load cells implanted into cadaveric femora and compared both the maximum force and force impulse. They found that the hybrid III dummy measured force was much higher at the same impact velocity, than that measured in the cadaver femur. For example, at a typical impact velocity of 30 ft/s), the hybrid III dummy femur measured a force of 12115 lbs compared with 4213 lbs for the femur, or a difference of 288%. However, Morgan et al. (1990) stated that the hybrid III response was close to that of the cadaver femora and that it correctly predicted injury and no injury conditions. Since the NCAP test
protocol is in fact accepted as a standard measure for the potential for injury, we assumed that it accurately represents crash conditions and forces and used the data accordingly. A number of studies, including those referred to in Table 1, have been used to develop failure criteria for the human femur to impact loading. Viano (1977) proposed an allowable force of F= 23.14–0.71 T (where F, is the peak impact force in Newtons, and T, time duration of impact load (ms), TB 20 ms) and F= 8900 N for T\20 ms, which is the case for impacts into padded dashboards. Leung et al. (1983) showed that ultimate load on the femur during impact is directly related to compact bone crossectional area and midsection, with bone area related directly to the weight of the subject, and to density. Based on their tests they proposed the same criterion as (Viano, 1977) for impact durations greater than 20 ms, F =8900 N. They also showed that femur midshaft crossectional area varied by a factor of 2 in their sample as did femur failure load. It appears that four factors led to axial impact fractures of the femora in the cases studied at relatively low speeds in which femoral fracture would have been considered unlikely from test crash predictions. First was single knee contact, second at least in most of the group, lack of complete restraint use, third was contact outside the central part of the knee bolster, and fourth muscle loading contributing to the total axial force on the femur due to bracing or braking.
Acknowledgements This study was funded through the NHTSA Crash Injury Research and Engineering Network and General Motors Inc. Cases supplied by the Universities of Washington, Michigan, Miami, California-San Diego and Scripps Institute are gratefully acknowledged.
References Borges, O., 1989. Isometric and isokinetic knee extension and flexion torque in men and women aged 20 – 70. Scandinavian Journal of Rehabilitation Medicine 21, 45 – 53. Cheng, R., Yang, K., Levine, R., King, A., 1984. Dynamic impact loading of the femur under passive restrained condition, SAE Technical Paper No. 841661. Daffner, R.H., Deeb, Z.L., Lupetin, A.R., Rothfus, W.E., 1988. Patterns of high speed impact injuries in motor vehicle occupants. Journal of Trauma 28, 498 – 501. Dischinger, P.C., Kerns, T.J., Kufera, J.A, 1995. Lower extremity fractures in motor vehicle collisions: The role of driver gender and height. Accident Analysis and Prevention 27, 601 – 606. Donnelly, B.R., Roberts, D.P., 1987. Comparison of cadaver and hybrid III dummy response to axial impacts of the femur. SAE 872204.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11 Karlson, T.A., Bigelow, W., Beutel, P., 1998. NHTSA Report No DOT HS 808 791, 09/98. King, J., Fan, W., Vargovick, R., 1973. Femur load injury criteria — a realistic approach. SAE Technical Paper No 730984. Leung, Y.C., Hue, B., Fayon, A., Tarriere, C., Hamon, H., Got, C., Patel, A., Hureau, J. Study of knee-thigh-hip protection criteria. SAE Technical Paper No. 831629. Lord, J.P., Aitkens, S.G., McCrory, M.A., Bernauer, E.M., 1992. Isometric and isokinetic measurement of hamstring and quadriceps strength. Archives of Physical Medicine and Rehabilitation 73, 324 – 330. MacKenzie, E.J., Burgess, A.R., McAndrew, M.P., et al., 1993a. Patient oriented functional outcome after unilateral lower extremity fracture. J. Orthop. Trauma 7, 393 –401. MacKenzie, E.J., Cushing, B.M., Jurkovich, G.J., 1993b. Physical imparement and functional outcomes six months after severe lower extremity fractures. Journal of Trauma 34, 528 – 561. Melvin, J.W., Stalnaker, R., Alem, N., Benson, J., Mohan, D., 1975. Impact response and tolerance of the lower extremities. SAE paper No 751159.
11
Morgan, R.M., Eppinger, R., Marcus, J., Nichols, H., 1990. Human cadaver and hybrid III responses to axial impacts of the femur. IRCOBI. Murray, M.P., Duthie, E.H., Gambert, S.R., Sepic, S.B., Mollinger, L.A., 1985. Age related differences in knee muscle strength in normal women. Journal of Gerontology 40, 275 – 280. NHTSA, 2000a. National Highway Traffic Safety Administration Web Site, http://www.nhtsa.dot.gov/ncap/cars. NHTSA, 2000b. Published femur fracture risk curve, http:// www.nhtsa.dot.gov/ncap/infopage.htm. Powell, W.R., Ojala, S., Advani, S., Martin, R., 1975. Cadaver femur responses to longitudinal impacts. SAE Technical Paper No. 751160. Roberts, D.P., Donnelly, B., Morgan, R., 1987. Cadaver response to axial impacts of the femur, SAE Technical Paper No. 876042. Siegel, J.H., Mason-Gonzales, S., Dischinger, P., et al., 1993. Safety belt restraints and compartment intrusions in frontal and lateral motor vehicle crashes: mechanisms of injuries complications, and acute care costs. Journal of Trauma 5, 736 – 759. Viano, D.C., 1977. Considerations for a femur injury criterion. SAE 770925.
Femur fractures in relatively low speed frontal crashes: the possible role of muscle forces Allan F. Tencer a,*, Robert Kaufman e, Kathy Ryan e, David C. Grossman d,e, M. Brad Henley a, Fred Mann c, Charles Mock b, Fred Rivara d,e, Stewart Wang f, Jeffery Augenstein g, David Hoyt h, Brent Eastman i, Crash Injury Research and Engineering Network (CIREN) a a
Department of Orthopedics, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA b Department of Surgery, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA c Department of Radiology, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA d Department of Pediatrics, Harbor6iew Medical Center, 325 Ninth A6e, Seattle, WA 98104, USA e Harbor6iew Injury Pre6ention and Research Center, Uni6ersity of Washington, Seattle, WA 98104, USA f Uni6ersity of Michigan, Ann Arbor, MI, USA g Uni6ersity of Miami, Miami, FL, USA h Uni6ersity of California, San Diego, CA, USA i Scripps Institute, San Diego, CA, USA Received 20 July 2000; received in revised form 6 October 2000; accepted 9 October 2000
Abstract In a sample of relatively low speed frontal collisions (mean collision speed change of 40.7 kph) the only major injury suffered by the partly or fully restrained occupant was a femur fracture. However, femur load measurements from standardized barrier crash tests for similar vehicles at a greater speed change (mean of 56.3 kph) showed that in almost all the cases, the occupant’s femur would not have fractured because the loads were below fracture threshold. In order to address this discrepancy, the load in the femurs of the occupants in the crash sample were estimated and compared with the femur fracture threshold. Femur load was estimated by inspecting the scene and measuring deformations in each vehicle, defining occupant points of contact and interior surface intrusion, and calculating crash change in velocity and deceleration. From this data, the measured femoral loads from standardized crash test data in a comparable vehicle were scaled to the actual crash by considering crash deceleration, occupant weight, and restraint use. All the occupants (7 males, average age 26.7 years, 13 females, average age 36 years) sustained at least a transverse midshaft fracture of the femur with comminution, which is characteristic of axial compressive impact, causing bending and impaction of the femur. However, the estimated average maximum axial load was 8187 N (S.D. = 4343N), and the average probability for fracture was only 19% (based on the femur fracture risk criteria). In 13 crashes the fracture probability was less than 10%. Two factors were considered to explain the discrepancy. The occupant’s femur was out of position (typically the driver’s right front leg on the brake) and did not impact the knee bolster, instead hitting stiffer regions of the dashboard. Also, since most victims were drivers with their foot on the brake to avoid the collision, additional compressive force on the femur probably resulted from muscle contraction due to bracing for impact. Adding the estimated muscle load on the femur to the estimated external load increased the femur loads beyond threshold, explaining the fracture in all but one case. Since crash tests using dummies cannot simulate out of position occupants or muscle contraction loading, they may underestimate the total load acting on the femur during actual impacts where the driver is bracing for the crash. These results may have implications for altering knee bolster design to accommodate out of position occupants and the additional load caused by muscle forces during bracing. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Femur fracture; Frontal crashes; Biomechanics
* Corresponding author. Tel.: +1-206-3415601; fax: + 1-206-7313227. E-mail address: [email protected] (A.F. Tencer). 0001-4575/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0001-4575(00)00097-X
2
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
1. Introduction Fractures of the femur due to motor vehicle impacts can result in extensive medical care, and if they have an intra-articular component, may result in permanent disability (MacKenzie et al., 1993a,b). Two general approaches have been taken for studying the mechanisms of femur fractures in frontal collisions. Risk factor analyses have shown that occupant compartment intrusions act as direct mechanisms for contact injuries of the femur (Siegel et al., 1993; Daffner et al., 1988), that females are at higher risk Dischinger et al., 1995), and that seatbelts and airbags are not particularly effective in reducing femur injury risk (Dischinger et al., 1995). An extensive study by Karlson et al. (1998) showed that a frontal crash direction, being female, having a head-on vehicle to vehicle as opposed to a single-vehicle to-fixed-object collision, being in a smaller sized vehicle, and not using seatbelts were risk factors for serious femur fractures. However, the specific mechanisms of the injuries, including contact points of the occupant’s femur were not determined in that study. Extensive biomechanical studies of the cadaveric femur fracture mechanisms and forces in frontal impacts have been performed. A summary of the findings are given in Table 1 (Melvin et al., 1975; Roberts et al., 1987; Cheng et al., 1984; Leung et al., 1983; Powell et al., 1975; King et al., 1973; Morgan et al., 1990; NHTSA, 2000a) and shown in the Injury Risk Function Curve, Fig. 1, derived from those studies, relating femur axial compressive load to the probability of fracture (NHTSA, 2000b). The studies showed, from dynamic measurements, that the approximate range of axial compressive fracture load for cadaveric femurs was 7500–15000 N, with an accepted threshold load of 8900 N (Donnelly and Roberts, 1987). The probability
Fig. 1. Reproduction of the femur fracture injury risk curve published by the National Highway Traffic Safety Administration.
of fracture risk at these force levels was, from Fig. 1, about 6% at 7500 N and 90% at 15000 N. In an ongoing study of motor vehicle crashes, we have noticed that a significant proportion of relatively low speed change and deceleration frontal crashes resulted in isolated fracture of the occupant’s femur. The mechanism in each case, based on appearance of the fracture (a transverse fracture line indicating failure by bending combined with comminution indicating compression), was axial compressive loading by impact with the dashboard. The goal of the study was to compare the best estimate of the external force acting in the femur during impact in these crashes with two published criteria defining femur fracture tolerance, since by the first estimates the accelerations were low and the resulting femur loads below those which would produce this fracture. If the femur loads were below thresholds, the explanation for the fracture might lie in the additional load generated by internal muscle contractions, not normally simulated in dummy or cadaveric crash testing and out of position contact with the dashboard.
2. Methods
2.1. Hypothesis The hypothesis to be tested in this study was that for front seated, partly or totally restrained occupants in a sample of frontal crashes, who sustained an isolated femur fracture, the external force generated in the crash which acted on the occupant’s femur reached at least the threshold for fracture.
2.2. General approach The motor vehicle crash information included in this study was collected from CIREN (Crash Injury Research and Engineering Network, National Highway Transportation Safety Administration) Centers. Crashes in the CIREN database are sampled based on the fulfillment of several criteria. Among these are that the occupant must have been restrained and that an injury of AIS 3 or greater must have occurred. Each case was reviewed by a multidisciplinary team consisting of a crash investigator, a bioengineer, a research nurse, and the treating physicians. Each center employs a crash investigator who has been certified in vehicle crash reconstruction and data collection through the training program given by the Transportation Safety Institute’s National Automotive Sampling System (NASS). Each crash scene and vehicle investigation conducted by CIREN centers follow the data collection format established by NASS. Once the crash information was analyzed, and the injuries and hospital course documentation were obtained, the crash was reviewed
Had femur load cell Knee or supra condylar FMVSS 208 Patella, supracondylar femoral shaft fx Higher at 1–2 ms pulse Represents 5–95% probability Axial/obl 15 (10/5)
6.7–9.2 ms
10013–7565 4997–15998
7925 (1482) 7552 (2537) 7565 2.8–10.5 m/s Axial Axial 11 (9/2) 11 (10/1)
Unembalmed (21–63) Unembalmed (34–73) Unembalmed Embalmed
None, distal 1/3, supracondylar Knee or supra condylar 15045 (18343) 15588 (5927) 21–43 13.2–39.3 Axial Axial 22 (16/6) 18 (18/0)
841661 831629 770925 751160 730984 1990 IRCOBI
Melvin et al. (1975) Roberts et al. (1987) Cheng et al. (1984) Leung et al. (1983) Viano (1977) Powell et al. (1975) King et al. (1973) Morgan et al. (1990) 751159 876042
Unembalmed (45–90) Unembalmed (55–71)
Peak force mean (S.D.) (N) Velocity (ft/s) Direction Number tested (m/f) Author
Specimen type (age range)
3
by the multidisciplinary team to establish a probable mechanism. Of the cases reviewed, a subset of 20, involving a frontal collision and a femur fracture as the predominant injury, were selected for this study.
2.3. Case selection criteria The crashes selected all involved frontal or nearly frontal collisions (+20° to −10°). These included vehicles, which were impacted by other vehicles but also included about 1/3 that impacted trees, poles, barriers, or other stationary objects. The major resulting injury was limited to an isolated femur fracture, usually involving the midshaft but in some cases, including additional injuries about the knee or hip.
2.4. Crash in6estigation
SAE paper
Table 1 Summary of data available in the literature related to axial compressive fracture loads of human cadaveric femora (N = Newtons)
Type of fx produced/comments
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Each crash site had scaled documentation of the roadway, traffic controls, road surface type, conditions, and road grade at both pre- and post- impact locations. Physical evidence such as tire skid marks were located and referenced to establish the heading angle and post impact trajectory of the colliding vehicles. A scaled drawing with impact and final rest positions was completed to assist in calculation of the speed and force at impact. Exterior inspections of the vehicle were performed, which included detailed measurements of the direct and induced damage. For this study, all the crash damage involved the front of the vehicle. With a contour gauge, a damage crush profile was collected from the front bumper and a specific Classification Deformation Code (CDC), which includes the principal direction of force (PDOF) was assigned. These measurements were entered into crash analysis software (Win SMASH, U.S. Dept of Transportation) to calculate the change in velocity (Delta V) of the vehicle during impact and the energy dissipated during the crash event. An inspection of the interior of the vehicle from which the injured person had been removed was performed to determine points of contact and restraint system use. This inspection also included assessing the integrity of the passenger compartment, status of glazing, and intruding components. The exact locations and all evidence of contact by the occupant who had sustained the femur fracture with the interior surfaces were documented. The lower extremity contact evidence consisted of dents in the padding, fabric transfers, and minor cracking of instrument panels or knee bolsters. The dash panel was however not disassembled to identify deformation to the knee bolster. An examination of the restraint system was performed, including lap and shoulder belts and the air bag, if available, to confirm use by the injured occupant.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
4
2.5. Estimation of occupant femur load on impact
For a lap belt only restrained occupant with single femur contact, (add load from shoulder belt to femur):
We used data from the same vehicle as that involved in the actual crash, reported by the New Car Assessment Program (NCAP, US National Highway Traffic Safety Administration) testing in 56 kph frontal collisions with two front seat fully restrained instrumented anthropometric dummies. A set of data was produced for the specific vehicles of our collision sample, including change of velocity of the vehicle at impact, crush deformation of the vehicle, average deceleration during impact, and load on the femurs, lap belt, and shoulder belt, all measured parameters in these tests. The data set used, derived from the NCAP tests, is shown in Table 2. The relationships used to derive actual crash occupant femur load are given below. The average deceleration of the vehicle in the actual crash is: A = 3.93×
! " (DV)2 S
(1)
where, A is the average deceleration of vehicle (g); 3.93, unit conversion factor; DV, change in velocity of the vehicle (kph); S, average vehicle frontal crush distance (mm). The force acting on the occupant’s femur in the actual crash is scaled for weight and deceleration compared to the test crash conditions
(2)
where, Fo is the estimated axial force in femur of occupant in crash (N); Fd, the measured axial force in femur of dummy in test (N); Ao, the average vehicle deceleration in crash (g), from Eq. (1); Ad, the average vehicle deceleration in crash (g); Wo, the weight of occupant in collision; Wd, weight of dummy in test. The force in the occupant’s femur (Fd) depends on the type of restraints used. Considering a worst case, if the occupant was partly restrained, for example shoulder belt only, all the force normally taken by the lap belt was added to the femur force. The following conditions were considered. For a lap and shoulder belt restrained occupant with two femur contact, the mean femur load is the average of the loads recorded in each femur of both dummies for that test vehicle.
Fll +Flr +Frl + Frr 4
(2a)
For a lap and shoulder belt restrained occupant with single femur contact. Fd =
Fll +Flr +Frl + Frr 2
Fll + Flr + Flr + Frr 2Fsl+ 2Fsr + 2 2
(2c)
For a shoulder belt or airbag only restrained occupant with single femur contact, (add load from lap belt to femur): Fd =
Fll + Flr + Frl + Frr 2Fpl + 2Fpr + 2 2
(2d)
Where, Fll, Flr, Frl, Frr, are the measured femur loads from crash test for that vehicle; ll, the left dummy-left femur; lr, the left dummy–right femur; rl, the right dummy-left femur; rr, the right dummy-right femur; Fsl, Fsr, the measured shoulder belt load from crash test; sl, the shoulder belt left dummy; sr, the shoulder belt right dummy; Fpl, Fpr, the measured lap (pelvic) belt load from crash test; pl, the lap (pelvic) belt left dummy; pr, lap belt right dummy.
2.6. Probability of axial loading exceeding fracture le6el In order to assess the potential for femoral axial impact forces to have reached levels consistent with femur fracture, we employed two criteria. The probability of femur fracture, based on Fig. 1, (NHTSA, 2000a) is: p= (−1E− 10)Fo3 + (5E − 06)Fo2 − (0.0413)Fo
Wo Ao × Fo = Fd × Wd Ad
Fd =
Fd =
(2b)
+ 98.133, (r 2 = 0.9946, range p= 1− 95%)
(3)
where p, is the probability of fracture and Fo, the occupant femur force. Based on the criteria given in the literature, for impact durations greater than 20 ms, the threshold force for fracture is (Donnelly and Roberts, 1987): Fo \ 8900 N
(4)
3. Results A total of 20 crashes were studied, shown in Table 3. In the sample of occupants who sustained an isolated femur fracture, there were 13 females with an average age of 36 years, weight of 177 lbs (80.5 kg), and height of 64.8 in (1.65 m). There were seven males, with an average age of 26.7 years, weight of 179 lbs (81.5 kg), and height of 69.5 in. (1.77 m). In the sample, Table 4, the occupants used the following restraint types; six were restrained with shoulder and lap belts and an air bag; two used shoulder and lap belts, two used shoulder belts and an airbag; five used a shoulder belt only; one used a lap belt only, and three used an airbag only. None were completely unrestrained. Those using a shoulder belt with an airbag or a shoulder belt only
Table 2 Data from New Car Assessment Program (NCAP) tests used to develop relationship between vehicle change in velocity, crush distance, and total load on both femurs of occupant in frontal collision Year of vehicle in study
Year of vehicle in crash test
Closing speed (kph)
Damage to front of vehicle (mm)
Average deceleration (g)
Lap and shoulder, belted occupant, double femur contact load (N)
Hyundai Accent Honda Civic Toyota Celica Toyota Corolla Toyota Corolla Geo Prizm Mercury Sable Ford Escort Chevy Blazer Toyota Corolla VW Jetta Ford Taurus Acura Integra Mercury Tracer Ford Escort VW Jetta Nissan Maxima Toyota Camry Ford Mustang VOLVO 740 GLE
1995
1995
56
433
28.5
4278
1994 1990 1992
1994 1990 1993
56.5 55.8 56.3
577 502 489
21.7 24.4 25.5
1992
1993
56.3
489
1991 1991
1990 1988
56.5 56.5
1995 1996 1990
1995 1997 1991
1996 1991 1996 1994
Lap and shoulder, belted occupant, single femur contact load (N)
Lap belted occupant, single femur contact load (N)
Shoulder belted occupant, single femur contact load (N)
Airbag only occupant single femur contact load (N)
Unbelted occupant single femur contact load (N)
8557
14549
14023
14023
15737
3179 4199 4682
6359 8398 9364
15981 16056 16444
15060 12851 12043
15060 12851 12043
21503 16310 14442
25.5
4682
9364
16444
12043
12043
14442
502.9 474
24.9 26.5
4218 5597
8436 11194
16961 17684
12467 15818
12467 15818
16774 16711
56.4 56.3 56.3
502.7 556 491.1
24.9 22.4 25.4
5270 3658 5675
10541 7315 11350
19512 15416 23137
15497 13165 13580
15497 13165 13580
19198 17609 19693
1994 1991 1995 1989
56.2 56.3 56.3 56.5
536 502.9 455.2 493.6
23.2 24.8 27.4 25.4
5734 5039 5214 4498
11469 10077 10428 8997
18786 16827 18317 18213
14629 14032 14993 13125
14629 14032 14993 13125
16212 15743 17669 17843
1993 1991 1995
1994 1988 1995
56.3 46.8 46.5
412.5 391.2 292.7
30.2 22.0 29.0
5235 5037 5729
10470 10075 11458
16707 10075 11458
13058 10075 11458
13058 10075 11458
14059 5037 5729
1990
1992
47.6
550.3
16.2
3959
7918
7918
7918
7918
3959
1988
1987
56.7
590.6
21.4
3759
7518
14535
12524
12524
15782
1990
1988
56
574.9
21.4
4774
9548
17813
14728
14728
18219
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Vehicle type
5
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
6
Table 3 Demographics of the occupants involved in the crashes studied in which the occupant sustained an isolated fracture of the femura Case number
M/F
Age
Weight (lbs)
Height (in)
Vehicle class
Vehicle weight (lbs)
Vehicle type
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
F F M F F F F M M F M M F F M F F F M F
21 21 26 58 24 33 80 38 26 15 26 30 39 22 20 27 48 25 21 55
155 128 160 155 125 260 150 262 165 90 180 200 179 200 150 264 178 146 138 245
66 69 68 64 66 66 64 70 ? 56 67 73 70 63 70 66 61 65.5 69 66
01 02 01 01 01 01 03 01 14 01 01 03 02 01 01 01 03 02 02 03
2101 2313 2696 2253 2253 2435 3131 2404 3993 2330 2647 3276 2628 2393 2371 2275 3001 2690 2818 2854
95 94 90 92 92 91 91 95 96 90 96 91 96 94 93 91 95 90 88 90
a
Hyundai Accent Honda Civic Toyota Celica Toyota Corolla Toyota Corolla Geo Prism Mercury Sable Ford Escort Chevy Blazer Toyota Corolla VW Jetta Ford Taurus Acura Integra Mercury Tracer Ford Escort VW Jetta Nissan Maxima Toyota Camry Ford Mustang Volvo 740
Class, 01, subcompact; 02, compact; 03, intermediate; 14, compact utility.
Table 4 Vehicle crash data from the 20 crashes studied in which the occupant sustained a femur fracturea Case no
Restraint use
Delta V (mph)
PDOF (°)
Max crush (in.)
Average decel (g)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Shoulder, lap belt, airbag Airbag Shoulder, lap belt, airbag Shoulder belt only Lab belt only Shoulder, lap belts Shoulder, lap belts,airbag Shoulder belt, air bag Shoulder, lap belts,airbag Shoulder belt only Airbag only Shoulder, lap belts,airbag Airbag Shoulder belt, air bag Shoulder belt only Shoulder belt only Airbag Shoulder belt only Lap and shoulder belt Lap, shoulder belts, airbag
34 37 24 22 22 43 12 27.8 10 20 21.1 32 36 24 29 12 22 19.4 30 28.9
0 5 0 0 0 0 0 0 −10 −10 0 0 10 20 −10 −10 0 10 0 −10
24.4 22.4 17.7 19.3 19.3 43.7 8.7 38.8 n/a 10 16.1 30.7 17.7 24.4 28.3 11.75 13 16.5 15 37.75
19.0 24.4 13.0 10.0 10.0 16.9 6.6 8.0 n/a 16.0 11.1 13.3 29.3 9.4 11.9 4.9 14.9 9.1 24 8.8
a
PDOF, principal direction of force, (0°, 12 O’clock, direct frontal; −10°, 11 O’clock from left; 10°, 1 O’clock, from right.
were in vehicles with automatic shoulder belt devices, in which the lap belt had to be attached manually. All sustained at least a midshaft femur fracture, with five having additional supracondylar or femoral neck involvement. Contact points of the injured femur of the occupant with some part of the instrument panel, are shown in Table 5.
The principal direction of force ranged from +20° to − 10°, with 11 crashes having an estimated direction of 0° (directly frontal). The average delta V of the collisions was 25.3 mph (40.7 kph), (S.D.= 8.5 mph (13.7 kph)), with an average deceleration of 13.7 g (S.D.= 6.4 g), Table 3. The average vehicle curb weight of the cars involved was 2648 lbs (1203 kg) (S.D.=419
Table 5 Estimated load at fracture, contact points and injury sustaineda Weight (lbs)
Estimated femur load (N)
19 24.4 13 10 10 16.9 6.6 8 N/a 16 11.1 13.3 29.3 9.4 11.9 4.9 14.9 9.1 24 8.8
155 128 160 155 125 260 150 262 165 90 180 200 179 200 150 264 178 146 138 245
6722 13 540 4747 6720 7493 10 029 2260 9087 6243 8662 6071 22 737 8150 7739 5608 11 498 5760 7559 4921
a
\8900 N Yes/no
Lower leg contact points
Intru
Contact evidence
Femur fracture type
8 80 3 8 9 27 0 17
N Y N N N Y N Y
3 14 3 100 9 8 1 50 2 7 1
N N N Y N N N Y N N N
Knee bolster /steer col Knee bolster /steer col Knee bolster /ign cyl Inst panel /center dash Glove box at latch Inst panel/center dash Knee bolster /steer col Knee bolster Knee bolster Inst panel/center dash Knee bolster Knee bolster/steer col Knee bolster Knee bolster/center dash Inst panel/steer col Inst panel Knee bolster/steer col Inst panel Inst panel Knee bolster
None 2’’ None None None None None 4’’ 6’’ None None 2’’ None None 5.5’’ None None None None None
Scuffed Cracked, scuffed Deformed Scuffed Cracked Scuffed Small dent Scuffed Scuffed, deform Scuffed Deformed Scuffed Scuffed None Deformed Small dent Scuffed Dent Small dent Scuffed, dent
Trans, midshaft Trans, midshaft Trans, midshaft Trans, midshaft Trans, midshaft Midshaft, troch Medial cond Supracond dist Trans, midshaft Trans, midshaft Trans, midshaft Midshaft, neck Trans, midshaft Comminuted mid Comminuted mid Comminuted Femoral neck Trans, midshaft Comm shaft Comminuted
Probability (%)
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Average deceleration (g)
S, shoulder belt; L, lap belt; A, airbag seats; D, drivers; RF, front passenger’s.
7
8
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
lbs (190 kg)), Table 4. The estimated load on the occupant’s femur during the impact, based on Eqs. (1) and (2), are given in Table 5, along with the calculated probability of fracture from the equation (Siegel et al., 1993). The observed points of contact of the occupant’s lower limb and the resulting fracture type are also given in Table 5. The mean best estimate of the femur load was 8187 N (S.D.=4343 N) for the 19 cases for which complete data were available, with a mean probability of 19% that the femur load was in the range for fracture, based on Eq. (3) and Fig. 1. Of the sample, 13 had a probability of fracture less than 10%. Based on the criterion that an axial compressive force greater than 8900 N is required for fracture, the estimated force exceeded this in only five cases. Eighteen of the 20 occupants were drivers and of these, 13 fractured their right femur, Table 5. The remaining two occupants were seated in the right front passenger’s seat and both fractured their left femurs. The specific contact component only intruded in five of the cases and in two cases, the intrusion was minimal. In three cases the toe pan intruded, acting on the lower leg, but for the majority, the intrusion did not appear to play a role in the axial loading on the femur. The contacts to the knee bolster systems and the lower instrument panels consisted of only scuffs, small cracks or dents, with minimal observable deformation, although elastic rebound of the padding could have occurred as shown in Fig. 2. The narrow area of contact for the driver’s right knee, between the steering column and the center of the dashboard, was the location of contact for eight drivers who fractured their right femurs. Some drivers’ right knees contacted at or near the steering column and others became cornered into the vertical center part of the dashboard. Some knee contacts were with other rigid components such as the ignition cylinder and the lock/latch of the glove box. Overall, for half of the occupants, their knees appeared to strike the stiffer areas or became cornered within the center of the dashboard, with the area of contact not appearing to yield to the force of impact.
4. Discussion In this study, femur, lap belt, and shoulder belt forces measured during standardized frontal collisions performed as part of the New Car Assessment Program were used to develop relationships between vehicle closing velocity, crush distance, and the probable axial compressive load from direct femur contact with the dashboard, for different conditions of occupant restraint. These relationships were used in assessing the estimated probability of femur fracture by axial compressive loading in a series of frontal crashes in which
the major injury to the occupant was an isolated femur fracture. We found the mean probability for femur fracture by axial loading in these crashes to be 19%. Thirteen of the 19 in the group had a probability under 10%, based on the femur load-fracture risk curve published by NHTSA, and 14 of the 19 were below the 8900 N threshold. Although the estimated probability was low, all the fractures appeared to be transverse with comminution, which would be expected to result from axial impact. Therefore, despite the relatively low external forces, direct axial contact was the probable mechanism for these injuries. In the majority of cases the criterion for femur fracture force that we estimated was below accepted thresholds, and yet the femur fractures were characteristic of those due to axial compressive impact, therefore other axial forces must have been present. Muscle loading was not accounted for in estimating the compressive loads acting on the occupants’ femurs in this study. In a frontal crash when the driver forcefully presses on the brake and braces for the impact, the knee flexors and extensors are tensed. Estimates based on muscle crossectional area and muscle power demonstrate that these muscles can generate significant compressive loads on the femur. Quasistatic measurements of maximum voluntary flexor and extensor muscle torques (Borges, 1989; Lord et al., 1992; Murray et al., 1985), show that a mean extensor muscle torque around the knee for 26-year-old males (the average age of the male subjects in our study) is about 273 Nm. With an approximate moment arm of 5 cm about the knee, this moment results in a compressive force of 5383 N acting along the axis of the femur. For 36-year-old females (the average age in our study for the female subjects), the extensor muscles force acting on the femur is about 3356 N. The knee flexors produce about 3002 N for males and 1801 N for females. It is impossible to determine specifically the extent of muscle contraction that the occupants of the vehicles in the crashes that we studied actually exerted, but it is entirely reasonable to assume that forceful braking and anticipatory bracing for the impact will produce considerable contraction of the knee flexor and extensor muscles. This appears to be the most reasonable explanation for the discrepancy between the mechanism of the femur fracture, axial loading, and the relatively low loads predicted. Review of the studies describing femur ultimate loads (Melvin et al., 1975; Roberts et al., 1987; Cheng et al., 1984; Leung et al., 1983; Powell et al., 1975) do not appear to have considered dynamic muscle loading as a potential contributor to the overall load that the femur could be subjected to during a crash. For comparison, the predicted loads, without and with the contributions of muscles (taken as the force generated by the knee flexors and an equal load from the extensors at maximum contraction), are shown in Fig. 3.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
9
Fig. 2. Example documentation of a frontal crash. (a) Vehicle exterior deformation; (b) interior contact point near steering column.
This observation, if correct, has significant implications for the design of dashboards and knee bolsters. If internal muscle loads add considerably to the external loads experienced by the occupant’s femur in the crash, then the dashboard or knee bolster, if designed for an external load criterion of 8900 N, and tested using in position crash dummies, may be too stiff to prevent femur injury in some crashes involving human occupants, especially those at relatively low speeds. Considering also that many of the occupants’ knees did not strike the relatively less stiff part of the knee bolster, consideration should be given to (a) extending the
contact surface of the lower dashboard to include the junction with the steering column, and (b) reducing the crush load of the knee bolster to accommodate the internal component of the total force acting on the femur due to muscle loading. The assumptions upon which this study is based require discussion. The values provided for the femur forces experienced by occupants in the collisions we studied represent our best estimates, working from the quantitative data of the NCAP test results and considering the weight of the occupant, average deceleration in the frontal collisions, the contact points of the occu-
10
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11
Fig. 3. Comparison of the estimated loads acting on the femur due to external forces of contact with the dashboard during the crash, and combined with possible muscle contraction forces (considering males and females) due to muscle tensing around the knee resulting from braking and/or bracing for the impact (femur fracture threshold, 8900 N, indicated by dashed line).
pant with the interior of the vehicle, and the variable use of occupant restraints. The crash is a dynamic event with loads in the shoulder and lap belts and femur contact loads reaching peak values at slightly different times during the crash. In this case, we made the simplifying assumption that all restraint and femur loads reached their peak values together and that the forces acting on the occupant were all essentially horizontal, preventing forward movement, and therefore, acting in the same direction. Inspection of individual time histories from NCAP tests showed that peak loads occurred close together in time, so this assumption is probably reasonable. A second assumption in the analysis was to add the loads normally carried by components of the restraint system, if not used, to the load acting on the femur. In fact, this component of the total load was probably distributed to the other parts of the restraint system. A very important question relates to how well the hybrid III dummy represents the human femur in terms of its ability to predict femur loading during impact. The hybrid III dummy femur load measurement is used as a criterion for determining the potential for femur fracture during NCAP 35 mph frontal collisions. Donnelly and Roberts (1987) used load cells implanted into cadaveric femora and compared both the maximum force and force impulse. They found that the hybrid III dummy measured force was much higher at the same impact velocity, than that measured in the cadaver femur. For example, at a typical impact velocity of 30 ft/s), the hybrid III dummy femur measured a force of 12115 lbs compared with 4213 lbs for the femur, or a difference of 288%. However, Morgan et al. (1990) stated that the hybrid III response was close to that of the cadaver femora and that it correctly predicted injury and no injury conditions. Since the NCAP test
protocol is in fact accepted as a standard measure for the potential for injury, we assumed that it accurately represents crash conditions and forces and used the data accordingly. A number of studies, including those referred to in Table 1, have been used to develop failure criteria for the human femur to impact loading. Viano (1977) proposed an allowable force of F= 23.14–0.71 T (where F, is the peak impact force in Newtons, and T, time duration of impact load (ms), TB 20 ms) and F= 8900 N for T\20 ms, which is the case for impacts into padded dashboards. Leung et al. (1983) showed that ultimate load on the femur during impact is directly related to compact bone crossectional area and midsection, with bone area related directly to the weight of the subject, and to density. Based on their tests they proposed the same criterion as (Viano, 1977) for impact durations greater than 20 ms, F =8900 N. They also showed that femur midshaft crossectional area varied by a factor of 2 in their sample as did femur failure load. It appears that four factors led to axial impact fractures of the femora in the cases studied at relatively low speeds in which femoral fracture would have been considered unlikely from test crash predictions. First was single knee contact, second at least in most of the group, lack of complete restraint use, third was contact outside the central part of the knee bolster, and fourth muscle loading contributing to the total axial force on the femur due to bracing or braking.
Acknowledgements This study was funded through the NHTSA Crash Injury Research and Engineering Network and General Motors Inc. Cases supplied by the Universities of Washington, Michigan, Miami, California-San Diego and Scripps Institute are gratefully acknowledged.
References Borges, O., 1989. Isometric and isokinetic knee extension and flexion torque in men and women aged 20 – 70. Scandinavian Journal of Rehabilitation Medicine 21, 45 – 53. Cheng, R., Yang, K., Levine, R., King, A., 1984. Dynamic impact loading of the femur under passive restrained condition, SAE Technical Paper No. 841661. Daffner, R.H., Deeb, Z.L., Lupetin, A.R., Rothfus, W.E., 1988. Patterns of high speed impact injuries in motor vehicle occupants. Journal of Trauma 28, 498 – 501. Dischinger, P.C., Kerns, T.J., Kufera, J.A, 1995. Lower extremity fractures in motor vehicle collisions: The role of driver gender and height. Accident Analysis and Prevention 27, 601 – 606. Donnelly, B.R., Roberts, D.P., 1987. Comparison of cadaver and hybrid III dummy response to axial impacts of the femur. SAE 872204.
A.F. Tencer et al. / Accident Analysis and Pre6ention 34 (2002) 1–11 Karlson, T.A., Bigelow, W., Beutel, P., 1998. NHTSA Report No DOT HS 808 791, 09/98. King, J., Fan, W., Vargovick, R., 1973. Femur load injury criteria — a realistic approach. SAE Technical Paper No 730984. Leung, Y.C., Hue, B., Fayon, A., Tarriere, C., Hamon, H., Got, C., Patel, A., Hureau, J. Study of knee-thigh-hip protection criteria. SAE Technical Paper No. 831629. Lord, J.P., Aitkens, S.G., McCrory, M.A., Bernauer, E.M., 1992. Isometric and isokinetic measurement of hamstring and quadriceps strength. Archives of Physical Medicine and Rehabilitation 73, 324 – 330. MacKenzie, E.J., Burgess, A.R., McAndrew, M.P., et al., 1993a. Patient oriented functional outcome after unilateral lower extremity fracture. J. Orthop. Trauma 7, 393 –401. MacKenzie, E.J., Cushing, B.M., Jurkovich, G.J., 1993b. Physical imparement and functional outcomes six months after severe lower extremity fractures. Journal of Trauma 34, 528 – 561. Melvin, J.W., Stalnaker, R., Alem, N., Benson, J., Mohan, D., 1975. Impact response and tolerance of the lower extremities. SAE paper No 751159.
11
Morgan, R.M., Eppinger, R., Marcus, J., Nichols, H., 1990. Human cadaver and hybrid III responses to axial impacts of the femur. IRCOBI. Murray, M.P., Duthie, E.H., Gambert, S.R., Sepic, S.B., Mollinger, L.A., 1985. Age related differences in knee muscle strength in normal women. Journal of Gerontology 40, 275 – 280. NHTSA, 2000a. National Highway Traffic Safety Administration Web Site, http://www.nhtsa.dot.gov/ncap/cars. NHTSA, 2000b. Published femur fracture risk curve, http:// www.nhtsa.dot.gov/ncap/infopage.htm. Powell, W.R., Ojala, S., Advani, S., Martin, R., 1975. Cadaver femur responses to longitudinal impacts. SAE Technical Paper No. 751160. Roberts, D.P., Donnelly, B., Morgan, R., 1987. Cadaver response to axial impacts of the femur, SAE Technical Paper No. 876042. Siegel, J.H., Mason-Gonzales, S., Dischinger, P., et al., 1993. Safety belt restraints and compartment intrusions in frontal and lateral motor vehicle crashes: mechanisms of injuries complications, and acute care costs. Journal of Trauma 5, 736 – 759. Viano, D.C., 1977. Considerations for a femur injury criterion. SAE 770925.