ANALYSIS OF MOVEMENT STRATEGIES DURING UNEXPECTED FALLS

S.N. Robinovitch, E. Hsiao, M. Kearny, and V. Frenk
Biomechanics Laboratory, Department of Orthopaedic Surgery,
San Francisco General Hospital and University of California,
San Francisco, San Francisco, CA 94110

Presented at the 20th Annual Meeting of the American Society of Biomechanics
Atlanta, Georgia. October 17-19, 1996

INTRODUCTION

Falls rank among the chief causes of accidental injury in both young and elderly populations, the most important of which are hip and wrist fractures [1]. However, the vast majority of falls cause no injury, despite the fact that the energy available in a typical fall from standing height far exceeds that required to fracture the hip or wrist [3, 4]. This suggests that specific motor control strategies exist to ensure Òsafe landingÓ during a fall. However, no previous study has examined human body movements during unexpected falls, and thus little is known regarding how injury is avoided in this potentially disastrous event. To address this issue, we measured the 3-dimensional movements of various body segments as young, healthy subjects underwent unexpected falls from standing height onto a gymnasium mat. We focused particularly on determining the temporal sequence in contact to the upper extremities and pelvis, and the effect on pelvic impact velocity of initial contact to the upper extremity.

REVIEW AND THEORY

Despite the fact that over 90 percent of hip and wrist fractures are caused by falls, few studies exist on the biomechanics of falling. Such studies have been limited to examining body movements during parachutist landings [2] and self-launched falls in young athletes [5]. Since these studies involved voluntary rather than unexpected falls, it is likely that observed motions were governed by pre-planned strategies for executing safe landings. Questions therefore arise regarding how well these motions represent actual falls from standing, which most often result from sudden and unexpected slips, trips, or loss-of-balance, and allow little time for pre-planning of a landing strategy.

Of central interest in the present study was assessing whether, rather than being random and unpredictable, the motions of the body segments during an unexpected fall involve a common sequence of coordinated movements which are employed in an attempt to land safely. In particular, based on the predominance of upper extremity fractures and the scarcity of hip fractures in the young, we hypothesized that young fallers (1) commonly ÒbreakÓ the fall by impacting the ground with an outstretched hand before contact occurs to the pelvis or trunk, and (2) avoid impact to the hip.

PROCEDURES

Three healthy, young volunteers participated in the study (two males and one female; aged 23, 28, and 22 yrs; body weight 881, 739, and 525 N; body height 1.73, 1.77, and 1.52 m). During the experiments, subjects stood on a large gymnasium mattress, which suddenly translated by means of a spring-actuated platform (Figure 1), initiating an unexpected ÒslipÓ. Throughout testing sessions, we randomly varied both the direction of the perturbation (achieved by having the subject stand forwards, backwards, or sideways with respect to the perturbation direction), and the strength of the perturbation applied to the platform (for each direction, two trials were conducted at horizontal accelerations of 4, 6, 8, and 9 m/s2). In order to evoke natural protective responses, the only instructions to the subject were that the platform might move, and that they should Òtry to prevent themselves from fallingÓ.

Figure 1: Experimental setup for simulated slipping experiments.

In all trials, a 6-camera, 60 Hertz motion analysis system (MacReflex, Qualisys Inc., Glastonbury, CT) was used to acquire the 3-dimensional positions of 20 soft foam, reflective markers located at the foot, ankle, shin, knee, thigh, anterior superior iliac spine (ASIS), sacrum, shoulder, elbow, wrist, and head. Custom routines written in MATLAB (The MathWorks, Natick, MA) were used to filter (recursive Butterworth filter, 10 Hz cut-off frequency) and differentiate position data.

In 73 percent of trials, subjects were able to regain balance through one or more steps. In four trials, subjects contacted the ground with one or both hands, but avoided impact to the trunk and knees. Fifteen trials resulted in ÒfallsÓ, defined as contact to a body part other than the feet and hands (distribution between subjects: 4, 3, and 8; direction of falls: 12 posterior, 3 lateral, 0 anterior), and were the subject of further analysis. Through close examination of stick figure animations and position-time graphs, we found that a reasonable criteria for the occurrence of ground contact to a given body part was passage of the corresponding body marker below a horizontal plane located 140 mm above the gym mattress. Contact velocities were then taken to equal the vertical velocity of the given marker at this instant.

RESULTS

In all falls, impact occurred to one or both wrists, consistently before impact to the ipsilateral elbow or shoulder. The average time for wrist impact after initiation of platform acceleration was 647±110 (s.d.) ms. In all falls, impact also occurred to at least one pelvic marker (right ASIS, left ASIS, or sacrum), at an average time of 732±166 ms. In all but one fall, pelvic impact occurred after impact to the wrist. However, the time difference between wrist and pelvic impacts was small, averaging 85±86 ms. In 47 percent of falls, the left or right ASIS was observed to impact before the sacrum, suggesting posteriolateral or laterally-directed impact towards the hip.

Wrist contact velocities were found to average 2.69±0.9 m/s, while pelvic contact velocities averaged 2.52±0.9 m/s. The time interval between wrist and pelvis contact was found to significantly affect pelvis contact velocity (Figure 2): as this time increased, the contact velocity of the pelvis decreased (r2=0.54; p<0.05).

Figure 2: Hip impact velocity as a function of the time interval between contact to the wrist and pelvis. Positive x-axis values reflect the wrist impacting before the pelvis.

Trajectories observed from the various falls showed strong similarity, involving steady downward movement of the pelvis, and a considerably more complex pattern for upper extremity movement (Figure 3). This involved an initial upwards movement of the wrist immediately following application of the perturbation (perhaps reflecting a startle response), followed by a rapid downward movement, and ending with another upward movement just prior to impact.

Figure 3: Variation in wrist and pelvic markers as functions of time after application of the perturbation to balance. w = time of wrist impact; p = time of pelvic impact.

DISCUSSION

In this study, we hypothesized that in the event of an unexpected fall, young, healthy individuals "break" their fall by contacting the ground with the upper extremity before the trunk or pelvis. Our experimental data suggest this to be true, since all but one fall involved impact to the wrist before the trunk or pelvis. We also hypothesized that young individuals avoid impact to the hip during a fall (thereby minimizing their risk for hip fracture). We found this to be untrue, based on the regular occurrence of impact to the pelvis, which was often oriented to predispose towards hip impact. It appears that rather than selecting for avoidance of hip impact, the motor control strategy governing falling seeks to reduce the likelihood of injury to any body part by: (1) distributing impact energy over several body regions through near-simultaneous impacts to the upper extremity, lower extremity, and pelvis; (2) imparting a relative upward motion to the distal upper extremity in the latter stages of the descent process, thereby reducing its kinetic energy at impact; and (3) reducing the velocity and kinetic energy of the pelvis at impact by ensuring the upper extremity contacts just prior to the pelvis.

This study appears to represent the first measures of movement strategies during unexpected falls. Limitations of the study include the possibility that subjects may have been less fearful of falling on a gymnasium mattress than on a hard surface. Furthermore, due to safety concerns, elderly subjects were not included. Whether elderly individuals exhibit similar falling patterns to those observed here is therefore unknown. However, it is interesting to consider our results in light of the fact that fall-related hip fractures are much more common in the elderly than the young, while upper extremity fractures are similar between the two populations. It may be that age-related changes in sensory and motor function cause deviations in falling behavior from those observed here, and that such deviations largely dictate hip fracture risk.

REFERENCES

(1) National Research Council. Injury in America, National Academic Press, 1985.

(2) Henderson, J. M. et al. Military Med., 158: 810-816, 1993.

(3) Lotz, J. C. et al. J, Bone Joint Surg., 72-A: 689-700, 1990.

(4) Myers, E. R. et al. Calcif. Tissue Int., 52: 199-204, 1993.

(5) van den Kroonenberg, A. et al. Trans. 39th ORS, Vol. 18, p. 24, 1993.

ACKNOWLEDGMENTS

This study was supported by a grant from the Academic Senate of the University of California, San Francisco.