During a fall, hip fracture risk increases 30-fold if there is direct impact to the hip. We conducted sideways falling experiments and found that subjects were able to avoid hip impact by rotating forward or by rotating backward during descent. These simple safe-landing strategies should be considered in designing hip fracture prevention programs.
Introduction: Ninety percent of hip fractures in the elderly are caused by falls. During a fall, hip fracture risk is increased 6-fold by falling sideways (instead of backward or forward) and 30-fold if direct impact occurs to the hip. Previous studies suggest that impact to the hip during a sideways fall can be avoided by rotating forward during descent to land on the outstretched hands. Presumably, an alternative strategy for avoiding hip impact is to rotate backward to land on the buttocks. We conducted sideways falling experiments to test the hypothesis that each of these falling strategies is equally effective in allowing one to avoid hip impact.
Materials and Methods: Twenty-two young adult women participated in trials where they were released from an inclined standing position into a sideways fall onto a foam mattress. Subjects were instructed to “land as softly as possible” and to “avoid impacting the hip” by either rotating forward or rotating backward during descent.
Results: We found that absolute values of the hip proximity angle, which described how close the impact site was to the lateral aspect of the pelvis, were not different in forward rotation and backward rotation trials (mean = 55.9 ± 22.4° versus 61.5 ± 15.8°, respectively). However, compared with forward rotation trials, backward rotation trials involved greater pelvis impact velocity (2.95 ± 0.25 versus 2.45 ± 0.77 m/s; p = 0.001) and greater whole-body kinetic energy at impact (238 ± 70 versus 156 ± 90 J; p = 0.001).
Conclusions: These results suggest that, during a sideways fall, individuals can avoid impact to the hip and thereby lower the risk for hip fracture by rotating forward or by rotating backward during descent. These simple yet effective safe-landing strategies should be considered in designing exercise-based hip fracture prevention programs.
Hip fractures are a major health problem for the elderly. In Canada, there are approximately 23,000 cases of hip fracture annually, with associated medical costs of about CDN$1 billion.(1) In the United States, these numbers are at least 10-fold greater.(2) Over 90% of hip fractures are caused by falls,(3) and the primary risk factors for hip fracture relate to bone strength, the frequency of falls, and the severity of falls.(4,5) Accordingly, hip fracture prevention strategies need to target each of these domains.
Growing evidence suggests that the strongest determinant of hip fracture risk in the event of a fall is the kinematic state of the body at the moment of impact.(5,6) Of primary importance is whether or not impact occurs to the lateral aspect of pelvis or the side of the leg, which increases fracture risk by 30-fold.(7,8) A related risk factor is the direction of the fall: sideways falls create a 6-fold greater risk for hip fracture than forward or backward falls,(9,10) presumably because of the greater risk for direct impact to the hip. Also, the velocity (and kinetic energy) of the body at impact influences fracture risk through its effect on the peak force generated during contact.(11–13)
Evidence also suggests that individuals use common protective mechanisms for avoiding injury in the event of a fall. For example, we previously found that, during unexpected sideways falls, young individuals tended to avoid hip impact by rotating their trunk about an inferior-superior axis during descent to land forward, facing the impact surface on the outstretched hands.(14) We also found that, by absorbing energy in the lower extremity muscles during descent (as is done during sitting or squatting), the impact velocity of the body during a fall is reduced by up to 70%.(15) These data help to explain why only about 2% of falls among the elderly result in hip fracture, and less than 10% result in any type of serious injury, despite the fact that the energy available during a fall greatly exceeds that required to fracture the elderly proximal femur.(11,13,16–19)
Presumably, impact to the hip during a sideways fall can be avoided not only by rotating forward to land on the hands but also by rotating backward to land on the buttocks. Indeed, while rotating forward during descent has the advantage of allowing visualization of the impact surface to coordinate landing, it generally creates the need for impacting the ground with the upper extremities to prevent impact to the head. This, in turn, increases one's risk for injury to the wrist, elbow, or shoulder.(7,20) By rotating backward during descent, one may presumably avoid impact and injury to the hip, head, and upper extremities. Therefore, the purpose of this study was to test the hypothesis that, in the event of an sideways fall (onto a gymnasium mattress), rotating backward versus rotating forward during descent has no affect on young females' ability to (1) avoid impacting their hip (as measured by the distance from the point of pelvis impact to the lateral aspect of the pelvis) and (2) impact the ground softly (as measured by the velocity and kinetic energy of the body at impact).
MATERIALS AND METHODS
Participants consisted of 22 females who ranged in age between 18 and 35 years (mean, 23 ± 5 years), had a body weight between 43.6 and 74.0 kg (mean, 56.5 ± 7.6 kg), and had a body height between 1.51 and 1.78 m (mean, 1.63 ± 0.07 m). This study was part of a larger project on fall injury mechanics, which involves only female subjects, based on the rationale that most injurious falls among the elderly occur in women.(21) Subjects were initially contacted through posting of advertisements at local universities. Respondents were then screened for eligibility through a telephone interview. Exclusion criteria included (1) exercising for fitness more than once per week; (2) prior participation in a program to train safe falling strategies (as is common in martial arts); (3) impairment of neuromuscular function secondary to previously diagnosed neurological disease (e.g., traumatic brain injury, cerebral palsy, multiple sclerosis, diabetic neuropathy); (4) amputation or other debilitating orthopedic conditions; and (5) inability to follow simple instructions. All participants provided written informed consent, and the experimental protocol was approved by the institutional review board.
During the experiment, the subject was made to fall sideways onto a gymnasium mattress by releasing (through an electromagnet) an inextensible tether that supported her at a 15° lean angle from the vertical (Fig. 1). Before each trial, we instructed the subject to “land as softly as possible” and to “avoid impacting the hip or side of the thigh” during the fall. We also instructed her to either rotate forward during descent to land on the outstretched hands (FR trials; Figs. 1A and 2) or to rotate backward during descent to land on the buttocks (BR trials; Figs. 1B and 2). Finally, we instructed her to keep her knees extended during descent. We conducted three FR trials and three BR trials, in a randomized order, with each subject. In all trials, the subject was positioned with the dominant side hip initially facing the impact surface (which, for 19 of the 22 subjects, was the right side). To increase the unexpectedness of the release, we inserted a random time delay between 1 and 10 s between the time that we verbally confirmed with the subject that she was “ready,” and the instant we released the tether.
For each trial, we used a seven-camera motion measurement system (ProReflex; Qualysis Inc., East Windsor, CT, USA) to acquire the three-dimensional positions of 18 reflective skin-surface markers placed bilaterally over the acromion process, lateral epicondyle of the humerus, distal end of the radius, anterior-superior-iliac spine (ASIS), greater trochanter, lateral epicondyle of the femur, lateral malleolus, and third metatarsal, and at the sacrum and top of the head. We also acquired simultaneous, synchronized measures with a force platform (model 4060H; Bertec Corp., Columbus, OH, USA) of the magnitude and point of application of contact forces on the subject's dominant foot (although these data are not reported here). Finally, to detect the instant of tether release (and fall initiation), we acquired force data from a load cell (model 31; Sensotec, Columbus, OH, USA) located in-line to the tether. We sampled force data at 960 Hz and motion data at 60 Hz.
We used a marker position technique to determine the occurrence and the time of impact between the ground and a given body part.(14) This involved acquiring “control trials” with each subject of body marker positions as the subject lay still on a 0.25-in-thick plywood board placed over the mat. We acquired measures in six different configurations: lying sideways on each side, sitting with the knees fully extended, lying prone, lying supine, and resting in a quadriped position on the hands and knees. We then used the data from these control trials to determine subject-specific look-up tables to indicate the occurrence of impact to the hands, pelvis, or knees during the fall trials. For example, the instant of pelvis impact was taken as the time frame where the vertical coordinate of the right or left trochanter marker (whichever was first to impact) descended below that of the corresponding marker in the appropriate control trial (e.g., the prone control trial if the subject landed prone).
Kinematic data and custom routines (MATLAB; The MathWorks, Natick, MA, USA) were used to determine four indices of fall severity specific to the instant of pelvis impact: (1) hip proximity angle, (2) pelvis tilt angle, (3) pelvis impact velocity, and (4) whole-body vertical kinetic energy.
Both the hip proximity angle and the pelvis tilt angle were determined by first considering the position of an ellipse whose circumference passed through the sacrum, right ASIS, and left ASIS markers (Fig. 3). This is an approach similar to that used by Smeesters et al.(22) to characterize pelvis orientation during falls. We then identified the site of pelvis impact by calculating the lowest point on the circumference of this ellipse at the time of impact. We also estimated the lateral aspects of the pelvis to coincide with the endpoints of the major axis of this ellipse. The hip proximity angle (α) reflected how near the point of impact was to the lateral aspect of the pelvis and was defined as the angle between the point of pelvis impact and the nearest lateral aspect of the pelvis, measured within the plane of the ellipse (Fig. 3). It was defined positive if impact occurred posterior to the lateral aspect of the pelvis and negative if impact occurred anterior to the lateral aspect of the pelvis (with α = 0° indicating direct impact to the lateral aspect of the pelvis). The pelvis tilt angle (β) indicated how near the pelvis ellipse was to being parallel to the ground at impact (Fig. 3). It was defined as the angle between the impact surface and the vector connecting the point of pelvis impact to the center of the pelvis ellipse. Therefore, β = 90° indicated that the trunk was horizontal (or parallel with the ground) at the moment of impact, and β = 0° indicated impact with the trunk in a vertical (sitting) position.
Pelvis impact velocity was calculated as the average vertical velocity of the left and right greater trochanter markers at the instant of pelvis impact (Fig. 4
∑i=111 ½ mivi2, where mi is the mass of segment i,(23) and vi is the vertical velocity of the center of gravity of segment i at the moment of pelvis impact.
In some trials, we had loss of data because of occlusion of markers from camera view. When possible, the missing marker positions were estimated using interpolation. However, there were several trials where markers dropped out and did not reappear for the remainder of the trial. Because of these lost data, we could calculate hip proximity angles and pelvis tilt angles for only 17 subjects and impact energies for 18 subjects. Hip impact velocities were calculated for all 22 subjects.
We used paired t-tests with an α level of 0.01 (to account for multiple comparisons) to determine whether differences existed between FR and BR trials in absolute values of hip proximity angle and pelvis tilt angle and magnitudes of pelvis impact velocity and vertical kinetic energy. All tests were performed with statistical analysis software (SPSS Inc.).
There was no difference in mean absolute values of hip proximity angle during FR and BR trials (p = 0.229). During FR trials, the mean hip proximity angle was −55.9 ± 22.4° (range, −13.0 to −86.0°), indicating that subjects tended to impact the anterior-lateral aspect of the pelvis (Fig. 5A; Table 1). During BR trials, the mean hip proximity angle was 61.5 ± 15.8° (range, 30.0–84.7°), indicating that subjects tended to land on the posterior-lateral aspect of the pelvis (Fig. 5B; Table 1).
Table Table 1. Mean Parameter Values
The mean pelvis tilt angle was significantly greater in FR trials than in BR trials (p < 0.001), reflecting that the transverse plane of the pelvis was more parallel to the ground during BR trials. The mean pelvis tilt angle was 50.8 ± 17.4° during FR trials (range, 13.9–78.9°), and 24.6 ± 14.9° (range, 6.3–54.4°) during BR trials (Fig. 6; Table 1).
Compared with BR trials, FR trials involved smaller vertical hip velocities (p = 0.001) and smaller vertical kinetic energies at impact (p = 0.001). Mean values of vertical hip velocity at impact were 2.95 ± 0.25 m/s (range, 2.49–3.46 m/s) for BR trials and 2.45 ± 0.77 m/s (range, 0.58–3.71 m/s) for FR trials. Mean values of vertical kinetic energy were 238 ± 70 J (range, 160–387 J) during BR trials and 156 ± 90 J (range, 6–295 J) during FR trials.
These differences may reflect the protective effect during FR trials of impacting the hands and knees before the pelvis (Fig. 4). During FR trials, the average time interval between fall initiation (tether release) and impact to the pelvis was 1135 ± 128 ms, which was slightly longer than the average time to pelvis impact in BR trials of 993 ± 102 ms. However, while the pelvis was the first body part (other than the feet) to impact the ground during BR trials, in FR trials, pelvis impact tended to be preceded by impact to the right knee (which occurred on average 66 ± 124 ms before impact to the pelvis), the right hand (which occurred on average 45 ± 75 ms earlier), and the left hand (which occurred on average 33 ± 84 ms earlier). Reflecting the fact that most subjects were released with the right hip facing the impact surface, impact to the left knee during FR trials occurred on average 15 ± 155 ms after impact to the pelvis.
Our results support the hypothesis that individuals are just as able to avoid impact to the lateral aspect of the pelvis during a sideways fall by rotating backward during descent to impact the buttocks (converting the sideways fall to a backward fall) as by rotating forward during descent and braking the fall with the outstretched hands (converting the sideways fall to a forward fall). There was no difference between FR and BR trials in the average (absolute) magnitude of hip proximity angle, which equaled 55.9 ± 22.4 in FR trials and 61.5 ± 15.8 in BR trials.
However, our results do not support the hypothesis that individuals are just as able to reduce the impact velocity of the pelvis and the kinetic energy of the body at impact by rotating forward as by rotating backward during descent. We found that BR trials resulted in average values of hip impact velocities that were 0.50 m/s greater and kinetic energies that were 88 J greater than average values in FR trials. This seemed to be because of the tendency in FR trials (but not BR trials) for impact to occur to the knee and outstretched hands before the pelvis.
There are several limitations to this study. We sought to directly test whether subjects' ability to avoid impact to the hip during a sideways fall is affected by the direction of rotation during descent. Accordingly, we instructed our subjects on the desired falling technique and provided them with the ability to plan their descent before the perturbation was applied. This is substantially different from a real-life loss of balance, where attempts to recover balance by sway or stepping would likely precede the onset of safe landing responses.(14) Also, we explored two specific falling strategies, and an individual's ability (or tendency) to use these strategies during a real-life fall may depend on a variety of intrinsic, environmental, and situational variables. For example, age-related changes in response time, muscle strength, and joint flexibility may affect one's ability to effectively use the FR or BR strategy. The protective value of each strategy may also depend on the nature of the perturbation and factors such as lighting and the presence of nearby obstacles. Our previous results suggest, among young individuals, the FR strategy is used more often than the BR strategy,(14) perhaps because it allows subjects' to coordinate their landing through visualization of the impact surface. However, the feasibility or appropriateness of the FR strategy clearly depends on one's ability to break the fall with the outstretched hands (to avoid head impact), which in turn should depend on both neuromuscular factors (e.g., reaction time and strength) and situational variables (e.g., the act of carrying an object).
The FR and BR strategies, as executed by our subjects, should reduce but not eliminate the risk for hip fracture during a fall. Anatomical considerations suggest that for the mean hip proximity angles we observed (∼55°), impact force and energy absorption would be shared between the pelvis bones (ilium and ischium) and the proximal femur. However, no study (to our knowledge) has directly examined how pelvis orientation at impact affects the portion of total force applied to the pelvis that is transmitted to the proximal femur. Furthermore, evidence suggests that, while over 90% of fall-related hip fractures involve impact to the hip and/or thigh, approximately 14–17% are caused by forward falls and 16–17% are caused by backward falls.(7,8,24) This suggests that impact during a fall to the anterior or posterior aspect of the pelvis, as opposed to the lateral, substantially reduces, but does not eliminate, the risk for hip fracture.
Risk for hip fracture during a fall should depend not only on impact energy but also on the contact force produced by a given impact energy and the force required to fracture the proximal femur under a similar loading configuration. With regard to the generation of contact force, the BR strategy should benefit from reductions in effective stiffness caused by the cushioning effect of the buttock soft tissues.(25,26) Conversely, the FR strategy would benefit from reductions in effective mass caused by impacting the ground with the trunk more horizontal(27) and the sharing of impact energy between the contacting knees, hands, and pelvis. With regard to the fracture force of the proximal femur, previous studies have shown this is greater when the load is directed anteriolateral as opposed to posteriolateral (which again favors the FR strategy over the BR). In particular, Pinilla et al. found that the average force required to fracture the proximal femur is 24% lower when the load is directed 15° posteriolateral compared with 15° anteriolateral,(28) and these trends have been supported by modeling studies by Keyak et al.(29) and Ford et al.(30)
On the other hand, an essential aspect of the FR strategy is impacting the ground with the outstretched hands, which increases one's risk for upper extremity injury. Among elderly individuals, impact to the outstretched hand during a fall increases the risk for wrist fracture by 20-fold, while reducing the risk for hip fracture by approximately 3-fold.(7,8) While wrist fractures are far less serious in terms of morbidity and mortality than hip fractures, they are nevertheless of no small consequence to their sufferers. Accordingly, in addition to training individuals in how to avoid hip impact, fracture prevention programs should teach participants how to minimize upper extremity contact forces. Recent studies have shown that the latter can be achieved by impacting the hands with the elbows flexed.(31,32)
Exercise-based programs to enhance safe landing responses would complement existing tools (e.g., osteoporosis medications, hip protectors, and fall prevention programs) for preventing fall-related hip fractures. We believe the development of such programs is a challenging but feasible task, for which the current study represents a crucial first step. In particular, we have demonstrated that, under controlled laboratory conditions, young adults can use the FR and BR strategies to avoid impacting the hip during a sideways fall. We have yet to examine whether elderly individuals are able to use these or other strategies to modify a fall as it happens. However, this seems to be a reasonable hypothesis, given that most falls in the elderly do not result in any serious injury(6) and commonly involve protective responses such as breaking the fall with the outstretched hands.(7–10)
Important goals for future studies are, first, to identify the feasibility of the FR and BR strategies for older individuals, and second, to identify those components of muscle strength, joint flexibility, and reaction time that govern the efficacy of these strategies, and thus represent appropriate targets for exercise programs. A starting point for achieving this would be to determine whether women ages 50–60 years are as able as younger women to use the BR or FR strategies to avoid hip impact during a sideways fall, through experiments similar to those used in the current study. Experiments with older women (ages 70–80 years) may be possible by employing additional safety precautions. These may include having subjects wear hip protectors and wrist guards during the trials, having subjects fall onto an extremely soft foam pit or a padded inclined surface; using a tether and harness system to control subjects' descent velocity; and using bone density imaging techniques (such as DXA) to screen individuals with osteoporosis.
An additional challenge is to use the knowledge gained from these experiments to design (in collaboration with fitness instructors, physiotherapists, and gerontologists) exercise programs that target key neuromuscular functions and incorporate safe techniques to improve specific fall protective responses (e.g., using a push-up exercise to simulate breaking a fall with the outstretched hands). Studies could then test whether such programs improve participants' fall protective responses and fear of falling, followed by larger efforts to determine whether training of safe landing responses reduces the frequency of fall-related injuries.
In summary, our results provide important new information on strategies for avoiding impact to the lateral aspect of the pelvis or thigh and thereby reducing the risk for hip fracture during a sideways fall. We found that, in the event of sudden loss of balance leading to a sideways fall, large hip proximity angles (averaging ∼55°) can be obtained by employing either a “forward rotation” (FR) or a “backward rotation” (BR) falling strategy during descent. These techniques allow one to convert a high-risk sideways fall to a lower-risk forward or backward fall. We also found that greater energy must be absorbed by the skin, fat, muscle, and skeletal structures of the pelvis during BR trials than during FR trials. However, unlike the FR strategy, the BR strategy does not require impact (and therefore the risk for injury) to the upper extremities. Further research is required to evaluate the appropriateness of the FR and BR falling strategies for frail elderly. However, our results provide strong support for the inclusion of training and education in safe landing strategies in exercise-based hip fracture prevention programs.
This study was supported by operating grants from the Centers for Disease Control and Prevention (R49CCR019355) and the National Institutes for Health (R01AR46890), a New Investigator Award from the Canadian Institutes for Health Research, and a Scholar Award from the Michael Smith Foundation for Health Research.