Balance control during walking in the older adult: research and its implications.

[Woollacott MH, Tang P-F. Balance control during walking in the older adult: research and its implications. Phys Ther. 1997;77:646-660.]

Key Words: Assessment, Balance, Older adults, Training, Walking.

Much research has been done on balance control in the older adult population. The ability of healthy and frail older adults to maintain postural stability during quiet standing, perturbed standing, and voluntary movement while standing has been well documented in ahe literature.[1-7] Research using functional mobility assessments has also helped to identify older adults who are at a high risk of falling during activities of daily living such as rising from a chair, bending over, and turning.[8,9]

Epidemiological studies, however, have shown that 30% to 70% of older adults' falls are due to trips, slips, and missteps; these events mostly take place during walking.[10-15] These statistics convey the important information that walking is the main daily activity in which the majority of the falls of community-dwelling older adults occur. Even though these older adults are capable of independent walking, there could be a substantial decline in their ability to control equilibrium, which does not become evident until a slip or trip happens. Although previous research on balance control has advanced our understanding regarding various aspects of static balance control ability of older adults, there are at least two limitations to generalizing such knowledge to balance control during walking.

First, the task of maintaining in-place balance (ie, "static" balance during standing and sitting) is different from maintaining balance when a person is mOving from point A to point B (ie, "dynAmic" balance during walking). In static balance, the base of support (BOS) remains stationary and only the body center of mass (COM) moves. The balance task in this case is to maintain the COM within the BOS or the limit of stability (the maxImal estimated sway angle of the COM).[16, 17] The activity of the ankle muscles is sufficient to maintain static balance during quiet standing.[17] In dynamic balance, however, both the BOS and COM are moving, and the COM is never kept within the BOS during the single-limb support periods. Ankle muscle activity alone has been found to be insufficient to maintain balance of the whole body during walking.[17] Thus, there has to be a different control mechanism for balance during walking. Second, researchers have reported that currently available functional assessment instruments are limited in predicting falls precipitated by slips or trips.[15] Thus, the factors that cause frequent falls during walking in older adults are yet to be identified.

The purpose of this article is threefold. First, we would like to address the nature of the walking task and how this task challenges dynamic balance in humans. Biomechanical analysis of human bipedal locomotion can provide valuable insight into the understanding of the balance dEmands during human walking. Second, a theoretical framework of the control mechanisms used to achieve balance during walking is introduced. This "proactive and reactive control" framework was originally Put forth by Patla.[18] Evidence that relates to this framework is discussed in this article. Studies investigating these two control mechanisms in older adults are introduced. Although research on dynamic balance is just beginning, some valuable information is starting to become available regarding why older adults have a higher tendency to fall during walking. Lastly, the clinical implications of these recent research findings are discussed.

Our discussion will be focused on community-dwelling older adults. These older adults are more likely to be engaged in regular walking activity than frail older adults or nursing home residents. Therefore, they are also more often challenged by walking in various environments.

Biomechanical Challenges to Balance Control in Bipedal Human Locomotion

Locomotion consists of multiple subtasks that have to be fulfilled at the same time for this behavior to be considered successful. The four basic subtasks during locomotion are (1) generation of continuous movement to progress toward a destination, (2) maintenance of equilibrium during progression, (3) adaptability to meet any changes in the environment or other concurrent tasks, and (4) initiation and termination of locomotor movements.[19,20] Although the first subtask involves repetitive lower- and upper-extremity movements to propel the body, the second and third subtasks require a complex integration of locomotor and balance abilities to maintain an upright posture and to properly modify the ongoing locomotor behavior to suit the environmental changes or other task demands.[21,22] The fourth subtask relates to the ability to switch between one status of motion to another. Although this last subtask also involves postural adjustments, it will not be discussed here because of a limitation in space. Readers may refer to recent work by other researchers for in-depth information.[23-26]

In humans, the control of balance during steady-state walking (ie, the second and third subtasks of locomotion) is not an easy task. Compared with other species, humans have two biomechanical disadvantages that make walking an especially challenging task. One disadvantage arises from the use of a bipedal locomotor pattern, which consists of two single-limb support periods. These two periods are relatively long and together take up 75% to 80% of the whole gait cycle duration.[27] During these two periods, the vertical projection of the body's center of mass (COM) travels forward and outside the medial border of the supporting foot.[28] Although this COM a d BOS spatial relationship facilitates weight transfer between the two lower extremities, it also inevitably creates potential mediolateral instability during single-limb support periods. That is, the product of the mass of the whole body and the distance between the COM a d BOS results in a gravitational moment about the ankle joint that makes the body fall toward the midline. A counterbalancing moment around the hip and lower trunk is required to prevent the whole body from falling toward the midline and at the same time ensure proper weight transfer to the other leg.[29] This counterbalancing moment is largely generated by the hip abductors and trunk lateral flexors, and fine tuned by the ankle evertors and invertors.[29-31]

Foot placement at heel-strike also determines the magnitude of the gravitational moment of the COM about the supporting foot during the single-limb support periods. Winter[17] hypothesized that the foot placement in the mediolateral direction was primarily controlled by the activity of the hip abductors during the swing phase. He tested this hypothesis by asking young adults to walk with wider or narrower step widths than normal. When the young adults widened their step widths, the gluteus medius muscle activity during the swing phase also increased. When they narrowed their step widths, the gluteus medius muscle activity decreased. Thee ankle evertors and invertors, however, did not show concurrent changes in activity level with the widened or narrowed step widths.[17] These findings support Winter's hypothesis that the hip abductors play an important role in regulating step width and balance adjustments associated with different step widths.

Studies on gait patterns of healthy older adults and older adults with recent histories of falls have shown that these older adults have narrower step widths as compared with young adults.[32,33] We believe that older adults may use a narrowed BOS to reduce the gravitational moment of the COM in the mediolateral direction in attempt to minimize lateral instability. It is also possible, however, that the narrowed BOS reflects a decreased control ability of the hip abductors. In addition, Gabell and Nayak[34] noted that the step-to-step variability in step width was greater in healthy older adults during comfortable-speed walking as compared with young adults.[34] The finding of increased within-subject variability in stride width suggests a lack of consistency in the control of lateral stability amo g older adults.

Traditional gait analysis of older adults has been focused on the sagittal-plane motion; little attention has been paid to lateral stability control. Recent studies on standing balance control of older adults have indicated that the ability to control lateral stability during quiet stance can be used to predict older adults' likelihood of falling when the COM or BOS is disturbed.[15] Moreover, when older adults experience large and rapid perturbations of the support surface in the anteroposterior direction during standing, they more often take steps laterally than do young adults.[4] The more frequent use of lateral steps implies greater lateral instability in older adults when their stance is highly challenged. Whether older adults also have difficulty with regulating lateral stability when their gait is disturbed, however, is unknown. Future research is needed in this area.

A second biomechanical disadvantage of human locomotion has to do with the human body structure--two thirds of the total body weight is centered in the upper body (head-arm-trunk) segment.[35] With this form of weight distribution, the upper body can store a large amount of potential energy. If the upper body is not controlled in an upright position, this potential energy can easily be converted to kinetic energy during a fall and result in a serious injury. Winter and associates[31] noted that at heel-strike and push-off during the gait cycle, it is particularly difficult for a person to maintain the upright posture of the upper body. At heel-strike, a backward hip acceleration is caused by the ground reaction force. Due to this acceleration, the upper body leans forward. Similarly, at push-off, the hip accelerates forward, and as a result, the upper body leans backward. Therefore, during the gait cycle, the upper body continues to oscillate forward and backward because of the changes in the hip acceleration. To overcome this upper-body instability, a counterbalancing torque has to be generated around the hip and trunk. Winter and colleagues[31] found that during normal walking, a hip extensor torque is required at heel-strike to prevent the upper body from falling forward. Similarly, a hip flexor torque is required at push-off to prevent the upper body from falling backward.

During walking, older adults often assume a more rigid and guarded posture than do young adults.[36] Is this how older adults preserve the ability to control the upright posture of the upper body during normal walking? Crowninshield et al[37] found that adults over the age of 60 years showed a decrease in the peak hip joint moments when compared with adults aged 22 to 30 years. This decrease was found to be related to the shorter stride length in the older adults. Therefore, it is possible that because of the decreased ability to control an upright posture of the upper trunk in the sagittal plane, older adults adopt a smaller stride length to reduce the ground reaction force. This reduction in ground reaction force, in turn, decreases the balance challenges to the upright stability of the upper body in the sagittal plane.

The maintenance of the upright posture of the upper body not only prevents a potential fall preceded by an unsteady upper-body movement, it also assists in stabilizing the head and gaze. Pozzo et al[38] found that during various dynamic tasks, such as free walking, walking in place, running in place, and hopping, the angular displacement of the head from a horizontal plane in line with the semicircular canals was less than 20 degrees, regardless of the large limb movement. They concluded that head stabilization is necessary for gaze orientation. To further understand how head stabilization is achieved during walking, Prince et al[39] examined the activity of paraspinal muscles at nine different spinal levels (C-7, T-2, T4, T-6, T-8, T-10, T-12, L-2, and L-4) during normal walking. Interestingly, the muscle activation profile revealed that activity of the higher-level paraspinal muscles preceded that of the lower-level paraspinal muscles in a top-down propagation manner. Because the lower-level paraspinal activity coincided well with heel-strike to attenuate the hip posterior acceleration, Prince et al suggested that the paraspinal activity from the higher spinal levels is controlled in an anticipatory fashion to attenuate the expected hip acceleration at heel-strike.

Although similar studies have not been performed on older adults, Winter[40] noted a greater acceleration of the head in older adults during walking as compared with young adults. The impact of this greater acceleration of the head on gaze stabilization in older adults is yet to be studied.

Proactive and Reactive Balance Control During Walking: Theoretical Framework and Research Evidence

Our discussion so far has been centered on balance control during normal walking. Most of the walking-related falls in older adults, however, result from trips, slips, or missteps. How do researchers attack this problem?

Recently, Shumway-Cook and Woollacott4' pointed out that when studying motor control, it is important to consider the interaction between the individual, the task, and the environment (Fig. 1, graph on the left). According to this point of view, the study of balance control during walking in older adults should take into account the nature of the task (walking) given to an older adult and the environment that puts the older adult at a higher risk of falling. With this perspective in mind, it becomes evident that the investigation of balance control mechanisms during walking in various environments is necessary to understand how older adults perform such a daily task (walking) in a challenging context (such as a slippery surface or a path full of obstacles) (Fig. 1, graph on the right).

[Figure 1 ILLUSTRATION OMITTED]

This approach is different from research that typically treated balance and gait disorders as two different risk factors for falls or two clinical problems in older adults.[42-44] The approach based on Shumway-Cook and Woollacott's model implies that difficulty with balance and gait abilities are intertwined. Impairments in gait function jeopardize the control of balance during walking. For example, gait disorders resulting from lesions in the central nervous system, such as hyperreflexia of the ankle planter flexors, could drastically reduce the BOS during walking, which in turn would increase the difficulty of maintaining balance.

Patla[18] suggested two balance control mechanisms for maintaining equilibrium during human walking. The first mechanism, the proactive control mechanism, refers to the balance control mechanism that takes place before the body encounters a potential threat to stability. This mechanism functions in two modes. One mode is to activate muscles or generate joint torques to reduce the inherent biomechanical threats to balance during normal walking. The aforementioned control for the upright posture of the upper body belongs to this mode of proactive control. That is, this mode of proactive control is already integrated into the normal walking pattern. A second mode of proactive control involves an early detection of potential environmental hazards and the implementation of postural and locomotion adjustments prior to the actual contract with the hazards. A good example for this mode of proactive control is when a person detours to avoid stepping onto an icy surface in the winter. If the balance threats are not detected in advance, the reactive control mechanism is needed. Then, a person has to evoke automatic postural responses to quickly regain balance.

Over the past few years, extensive studies on the second mode of proactive control of balance during walking have been done by Patla and collaborators[18,22,45-47] and Chen and associates.[48-50] The visual system and vigilance, or attention, are the keys to an early detection of potential balance threats.[22,50] Once the type and extent of the balance threats are recognized, complex sensorimotor integration processes are carried out to promptly implement appropriate modifications to the ongoing walking behaviors. For example, Patla and Rietdyk[46] found that the movement of the swing limb was modulated according to the obstacle height, but not the obstacle width, provided that the goal of the person was to step over, rather than around, the obstacle.

Noting the important role of vision in proactive balance control during walking, Patla[18] hypothesized that the high incidence of walking-related falls in older adults might be due to a decrease in the ability to use visual information during walking. To test this hypothesis, he first examined whether young and older adults differed in the ability to visually gather relevant information from their surroundings. In this study, both young adults (undergraduate college students) and older adults (65-85 years of age) were asked to wear opaque liquid-crystal eyeGlasses and to press a switch to make the glasses transparent whenever they wanted to sample the environment. The floor across which the subjects walked was either unmarked or had footprints marked at regular intervals on which the subjects were supposed to step. The young and olderadults showed no difference in visual sampling frequency or duration when they walked over the unmarked floor. When asked to walk on the footprints,however, the older adults sampled the tErrain for longer periods and less often than the young adults, suggesting that they perceived the need for more information to make accurate foot placement in a constrainedenvironment.

In another set of experiments, Patla[18] examined whethEr older adults required more time than young adults to implement an avoidance strategy when walking. Subjects were aSked to walk along a walkway and, when given a visual cue, to either lengthen or shorten their stride to match the position of the cue. He found that both young adults (undergraduate college students) and older adults (65-85 years of age) were able to perform the task when the cue was given two steps in advance. Older adults, however, had more difficulty than young adults in modifying step length when the cue was given only one step in advance. The lengthened visual monitoring in older adults, as previously discovered, or a longer time to implement gait modifications could contribute to older adults' need for longer periods of time to adjust step length. Furthermore, older adults were successful 60% of the time when lengthening the step length and only 38% of the time when shortening the step length, whereas young adults were able to accomplish both tasks 80% of the time. Patla[18] proposed that older adultS had more difficulty in shortening the step because of balance problems. He nOted that shortening a step requires modulating the forward pitch of the trunk and that older adults have difficulty controlliNg this aspect of balance when walking.

Similarly, Chen et al[49] investigated whether the minimum response time allowed to modify gait patterns to avoid an obstacle during walking would be different between older and young adults. The obstacle used was a hand of light located at a predicted foot placement. This virtual obstacle could be lit up at different times prior to the subject's heel-strike. The time allowed for gait modification before heel-strike ranged between 200 and 450 milliseconds, with increments of 50 milliseconds. The results showed that the rate of success in avoiding the obstacle was higher for young adults than for older adults, regardless whether the time allowed to make a modification to gait was long (96% for young adults, 92% for older adults) or short (21% for young adults, 16% for older adults). These findings are consistent with those of Patla[18] in that older adults need a longer period of time to implement gait modifications to prevent running into an obstacle on the walking path. One question that remains unsolved, however, is whether this longer response time in older adults can be ascribed primarily to a longer time to detect the obstacle, a longer time to execute the avoidance response, or both.

Because visual attention is important for avoiding obstacles, Chen et al[50] hypothesized that older adults may be more affected by a visual distraction presented concurrently with the obstacle as compared with young adults. To test this hypothesis, the subjects were asked to perform a vocal reaction-time task in response to a visual stimulus while performing the same obstacle-avoidance task described above.[50] They found that when the visual distraction was present synchronously with the obstacle appearance, the success rate in avoiding the obstacle decreased to a greater extent in older adults (36%) than in the young adults (20%), as compared with the no-distraction condition. These results indicate that attention demand is greater in older adults than in young adults for successful implementation of a proactive balance control mechanism.

Teasdale and colleagues[51] also reported that attentional demand during normal walking was higher in older adults than young adults, regardless of the phases of the gait cycle. When we consider the findings of Teasdale et al[51] and Chen and colleagues[50] together, it becomes clear that obstacle avoidance is not an easy task for older adults because a large amount of attention has to be allocated not only to the normal gait pattern but also to its modification.

Aside from the differences in the successful rate of implementing the obstacle-avoidance strategy, what are the differences between older and young adults in their actual obstacle-avoidance behaviors? Chen et al[48] investigated whether older adults (mean age=71 years) use different strategies than do young adults (mean age=22 years) in their actual avoidance of obstacles of heights created to simulate natural obstacles in the environment such as a 2.5- to 5.1-cm (1- to 2-in) door threshold or a 15.2-cm (6-in) curb. The control condition was a 0-mm condition in which tape was simply attached on the walkway. The investigators noted no differences between the age groups in foot clearance over the obstacles, but they found that older adults used a more conservative strategy when crossing obstacles, including a slower crossing speed, shorter heel-to-obstacle distance after crossing, and a shorter crossing-step length. They also found that because of the shorter crossing-step length, 4 of the 24 older adults, compared with none of the young adults, accidentally stepped on an obstacle and increased the risk of tripping.

Research by Patla and collaborators[18,45-47] and that by Chen and associates[48-50] have led to more in-depth understanding of proactive balance control during walking in older adults and how this control can be implemented successfully. To date, however, balance responses when an older adult actually trips over an obstacle (ie, the reactive control mechanisms during trips) have not been investigated.

The reactive control mechanism that is different from the proactive control balance mechanism primarily relies on the somatosensory and vestibular systems to determine the extent and type of the stimulus (threat) and to trigger appropriately scaled postural responses.[18] These postural responses are mainly polysynaptic spinal reflexes and supraspinal responses.[52,53]

In 1980, Nashner[53] performed an experiment to examine the reactive balance control mechanisms during perturbed walking. In the study, he perturbed the gait of young adults as they walked across a movable platform incorporated into a 4-m-long walkway. The perturbations were forward or backward translations of the support surface of 8 cm causing an ankle rotation at approximately 40 [degrees]/s. To examine whether muscle response characteristics changed according to the phase of the step cycle in which the perturbation occurred, he gave perturbations at heel-strike, the beginning of single-limb support phase, mid-stance, and the beginning of double-limb support.

The results showed that in response to perturbations at heel-strike, there were muscle responses in the stretched ankle muscles that began approximately 95 to 110 milliseconds after the onset of platform movement and lasted for about 100 to 400 milliseconds. For example, a forward platform displacement at heel-strike elicited an excitatory response in the tibialis anterior muscle that served to resist the change in the rotational trajectory of the ankle joint. The effects were strongest at heel-strike and the beginning of the single-limb support phase, were weaker at mid-stance, and were absent at the beginning of the double-limb support phase.

Nashner[53] hypothesized that a deviation in the movement of the ankle from its normal stepping trajectory provided a principal source of input to the muscles of the legs. He noted that the responses helped to slow or speed up the body's rate of forward progression to realign the body's COM. He also noted that the leg muscle postural responses during perturbed walking were similar to those observed when equivalent platform movements were given to subjects standing quietly on the platform. Neither response was monosynaptic, and both responses appeared to involve longer-latency response pathways, because the responses were activated at about 90-millisecond latencies.[54,55] Given the similarity in the patterns and onset latencies of the postural responses between perturbed standing and walking, Nashner speculated that the locomotor center uses the same postural synergies that are activated during stance perturbations to respond to perturbations during walking. In particular, he hypothesized that the spinal stepping generators continuously receive input regarding postural instability from both peripheral and central pathways and then incorporate the necessary postural adjustments into the output of the locomotor patterns.

Nashner's hypothesis that the discrepancy between the actual and planned ankle joint trajectories triggers reactive postural responses in platform-perturbed gait has gained additional support from other studies. Dietz and colleagues[56] have examined the ability of young adults to respond to postural perturbations during gait using a treadmill paradigm. In this paradigm, subjects walked on a treadmill that was accelerated (accelerations were 2.5-14 m/[s.sup.2] with amplitudes of 3 [degrees]-6 [degrees] of ankle angle change) or decelerated at heel-strike at unexpected moments in time. Responses to these perturbations were recorded from the thigh and leg muscles (gastrocnemius, hamstring, tibialis anterior, and quadriceps femoris).

Dietz and associates[56] found that latencies of responses in the gastrocnemius muscle to accelerations of the treadmill ranged from 90 milliseconds for slower accelerations (2.5 m/[s.sup.2]) down to 70 milliseconds for the fastest accelerations (14 m/[s.sup.2]). They noted that the duration of the response increased with higher-magnitude perturbations. They also noted that the hamstring muscles were activated about 100 milliseconds after perturbation onset (about 20 milliseconds after onset of gastrocnemius muscle activity). The gastrocnemius muscle's electromyographic response was closely correlated to the duration of the acceleration impulse and ankle joint displacement.

Dietz and colleagues proposed that group II afferents from the ankle muscles were responsible for the activation of the ankle muscle responses because a monosynaptic stretch reflex was not activated, even though stretch velocities were high, and also because previous experiments had shown that the response was preserved after ischemic blockage of group I afferents of the leg and foot.

An important finding in Nashner's study[53] was that postural adjustments were most prominent when the platform movement was imposed at heel-strike and the beginning of single-limb support, and these adjustments became much smaller or absent if the platform perturbation occurred in the later stance phases. This finding suggests that due to the changes in body orientation and support surface, the balance demands differ between the different phases of the gait cycle. Accordingly, postural responses evoked by the same perturbation also differ because of the different times at which the perturbation is imposed. Neurophysiologists have referred to this phenomenon as "phase-dependent modulation of reflexes."[57] The existence of this phenomenon suggests that spinal reflexes are modifiable and are adaptive to task requirements.

This "phase-dependent reflex reversal" phenomenon has an important function in dynamic balance control during locomotion. For instance, Stein[58] used a pneumatic system to apply a mechanical stretch to the gastrocnemius muscle at various times during the stance phase of human walking. He found that the gain of the stretch reflex (as measured by the ratio between the magnitude of gastrocnemius muscle activity in response to the stimulus and the magnitude of the stimulus) was small in early stance, but gradually increased toward late stance. Stein suggested that if the gain of this stretch reflex was set high in early stance by the nervous system, then the reflex would prevent the shank from efficiently rolling over the foot and would impede the continuation of walking. A high stretch reflex gain in late stance is useful, however, because it reinforces the push-off and thus assists in forward progression.

Yang and Stein[59] observed cutaneous reflex reversal in the tibialis anterior muscle when they applied an electrical stimulation to the tibial nerve at the ankle during human walking. In reaction to this stimulus, the medium-latency response (70-120 milliseconds) of the tibialis anterior muscle changed from excitatory, when the stimulus was applied in early swing, to inhibitory, when the stimulus was applied during the transition from swing to stance phase. This inhibitory response of the tibialis anterior muscle during late swing allowed the perturbed leg to fulfill its weight acceptance obligation after the transition to the stance phase. In this study, the most pronounced phasic modulation was in the

medium-latency responses, in contrast to the short- and long-latency responses. This finding indicates that this modulation in humans requires supraspinal neural substrates or polysynaptic spinal pathways for control, because medium-latency responses are known to involve these neural substrates.[52]

Thus, the phasic modulation of spinal reflexes is thought to be important for dynamic balance control in human walking. To further understand this contribution, researchers[60,61] have extended the investigation to the reflexive responses of both legs. Eng et al [61] applied a mechanical obstruction to the swing leg during the early and late swing phases of human walking. When the disturbance was imposed in early swing, young adults predominantly presented a flexion pattern of the swing leg (via activation of the biceps femoris and rectus femoris muscles of the perturbed leg) and an enhanced extension of the support leg (by activating the gastrocnemius muscle of the support leg). In contrast, when the swing leg was tripped in late swing, the swing leg showed a lowering response and there was no extension movement of the support leg. Eng et al explained that when the swing leg was perturbed in early swing, the extension of the support leg led to an early heel-off. This action raised the height of the total-body COM a d permitted a longer period of time for the perturbed leg to move away from the obstacle. The extension' movement of the support leg, however, was unnecessary when the obstacle appeared in late swing. In this case, the subject had already lowered the body's COM to prepare for landing with the perturbed leg. This study provides an illustration that reflexive responses to a realistic trip during human walking are coordinated between both lower extremities.

Because of the risky nature of a tripping task, reactive balance control mechanisms of older adults in response to a trip during walking have not been investigated. Recently, in our laboratory, a study was conducted to explore age-related differences in the reactive balance control mechanisms to slips during walking.[62] A slip was defined as a sudden increase in the horizontal velocity of any part of the foot that is in contact with the floor.[62] Healthy young adults (n=33, 20-35 years of age) and older adults (n=32, 70-85 years of age) were tested while walking across a movable platform incorporated into a walkway. A translational platform movement of 10 cm at 40 cm/s, timed to move at one of the three phases of the gait cycle (right heel-strike, mid-stance, or late stance) was used to simulate slips occurring at various times during the gait cycle. This platform movement was chosen because it was compatible with the slips when a person is walking on an icy surface.[62] Subjects were given a total of 48 trials, with the first 12 trials being no-perturbation control trials that were blocked. The rest of the trials were in random order and consisted of 12 forward-perturbation trials, 12 backward-perturbation trials, and 12 no-perturbation trials. Only the right foot was perturbed during the experiment.

The first question being addressed in this research was whether the locus of balance control during slips resides in the leg musculature, as Nashner[53] had suggested, or in the trunk and hip musculature, as had been discovered from nonperturbed normal human walking.[31] To answer this question, we analyzed postural responses from [15] muscles of the bilateral lower legs, thighs, hips, and trunk muscles in the young adults. In responding to the most challenging slip, a forward slip occurring at heelstrike, we found that the postural activity of the tibialis anterior and rectus femoris muscles of the perturbed leg was very important to dynamic balance control. The postural responses from these muscles were not only consistent (occurrence rate of [is greater than] 98%), but also had very short onset latencies ([bar] x=91 milliseconds for the tibialis anterior muscle and 140 milliseconds for the rectus femoris muscle) and long burst durations ([bar] x= 133 milliseconds for the tibialis anterior muscle and 203 milliseconds for the rectus femoris muscle). Furthermore, the magnitude of these postural responses were about seven times what would be observed in the muscle activity during normal walking. The activity of the gastrocnemius muscle was suppressed when the tibialis anterior muscle was highly active. This suppression of the gastrocnemius muscle (antagonist) activity indicates an important organizational function of the nervous system to enhance the postural recovery function of the tibialis anterior muscle (agonist) activity in regaining ankle joint trajectory. From the kinematic analysis, it was evident that the tibialis anterior muscle's postural activity served to restore the disrupted ankle joint trajectory and to realign the foot and leg segments of the perturbed lower extremity (Fig. 2). In this process, the knee joint underwent a flexion motion, which indicated eccentric contraction of the rectus femoris muscle to prevent a collapse at the knee (Fig. 3).

[Figure 2 ILLUSTRATION OMITTED]

Another consistent reactive response was found in the biceps femoris muscle of the perturbed leg. The onset of this postural response ranged between 80 and 130 milliseconds after platform perturbation onset, and its duration lasted about 100 milliseconds (Fig. 3). These findings indicate an important role of this muscle in restoring the normal knee trajectory (knee flexion) after heel-contact. Along with the long burst of the rectus femoris muscle, there was therefore a high prevalence of coactivation between the rectus femoris and biceps femoris muscles to control the knee joint stability. Moreover, the biceps femoris muscle activity during perturbed walking was not reported by previous researchers.

[Figure 3 ILLUSTRATION OMITTED]

Perturbed level walking versus treadmill walking may have had a different influence on the results.

On the nonperturbed side, consistent and early onset of the tibialis anterior, rectus femoris, and biceps femoris muscle responses was also observed after the platform perturbation. Although supplementary kinematic information was not available, it appeared from the video-based analysis that the function of these muscles' activity is to secure the foot lift-off in the early swing phase of the nonperturbed leg. Because the forward platform movement had stretched the right foot forward, the COM of the whole body was likely to drop in the vertical direction after the platform onset. Thus, the activity of the tibialis anterior, rectus femoris, and biceps femoris muscles of the nonperturbed leg may assist directly in ankle dorsiflexion, knee flexion, and hip flexion, respectively, in the early swing phase. These flexion movements in turn prevent the nonperturbed leg from being tripped when the COM drops in the vertical direction. These findings suggest that interlimb coordination is needed in such a dynamic balance task.

Based on the analysis of the frequency at which a muscle became active in response to the platform perturbation, the trunk muscle activity was not needed as frequently as the activity of the above-mentioned six leg muscles (de, bilateral tibialis anterior, rectus femoris, and biceps femoris). We hypothesized that when the balance threats can be attenuated sufficiently by the lower-extremity muscles during a slip, trunk muscle activity may not be needed.

Next, a comparison of postural responses between the older and young adults was made. Older adults were found to activate the same predominant postural muscles as did the young adults. The primary age-related differences in the postural responses were found to be a combination of a longer onset latency, a longer burst duration, and a smaller burst magnitude in older adults' postural responses as compared with those of young adults (Fig. 4). The combination of these differences resulted in a less effective balance recovery strategy in the older adults. The composite effect of all three factors, rather than a single factor, resulted in the difficulty with dynamic balance in older adults. This composite effect of the postural muscle activity in the older adults probably led to a slower rate in generating postural activity. Age-related losses of fast-twitch muscle fibers or decreased ability in recruiting motoneurons may have contributed to this slower rate of generating postural activity. Furthermore, these insufficient postural responses from the leg muscles often led to a backward lean of the trunk in older adults. As a result, older adults were found to more frequently use arm movement to assist in trunk stabilization and prevent a fall (Fig. 5).

[Figures 4-5 ILLUSTRATION OMITTED]

In the analysis of walking speed and stride length, we found that older adults shortened the stride length after the slip, whereas young adults did not shorten their stride length. This shortened stride length could be partially be accounted for by the longer coactivation time between the rectus femoris and biceps femoris muscles of the nonperturbed leg in the older adults as compared with the young adults. Shortening stride length is a common strategy people adopt while walking on icy or slippery surfaces in order to decrease the heel-strike speed and prevent a slip. However, shortening the stride after experiencing a slip, as found in our study, may have a different effect on balance control than shortening the stride while stepping on an icy surface. In particular, in our study, the whole-body COM was passively moved forward by the platform. In this case, a shortened stride length would cause the location of the COM to be closer to the front border of the BOS in the shortened step following the slip. Subsequently, if a subject shortened the step following the slip, he or she would need at least an extra step to regain the normal spatial relationship between the BOS and the COM after the slip. Therefore, stride shortening, as a common conservative balance strategy during walking, may not be an effective strategy for older adults in the current experimental context. Interestingly, as described earlier, Chen et al[48] also found that older adults used a shorter step to cross an obstacle than did young adults. This shortened step length while crossing an obstacle increased the likelihood of tripping for the older adults. Step length regulation appears to be important during both slips and trips. Questions as to why older adults tend to use a smaller step length after a trip or a slip are yet to be answered.

Given the known functional significance of phase-dependent modulation of reflexes, we believed that it would be interesting to examine whether older adults preserve the phase-dependent modulation ability in their postural responses to slips occurring at different times during the stance phase. Therefore, we examined the postural responses from both age groups in reaction to slips occurring at heel-strike and at mid-stance. Because the mid-stance slips were less challenging than the heel-strike slips, both young and older adults showed less frequent postural responses during the mid-stance slips than during the heel-strike slips. The young adults modulated their postural responses by shortening the burst duration and burst magnitude in the mid-stance slips. The older adults, however, showed decreases in burst duration, but not in burst magnitude, of their postural response in the mid-stance slips. We hypothesized that psychological fear, changes in motoneuron recruitment patterns, and changes in muscle physiology may have contributed to the inflexibility of the older adults' postural response.

As a whole, based on this series of experiments, we found that the reactive balance control mechanisms, when encountering an unexpected slip, were less efficient and less flexible in older adults as compared with young adults. More importantly, the less efficient control in the lower extremities often led to upper-trunk instability. Without concurrent arm movements, older adults may be very likely to fall.

Clinical Implications

Assessment of Proactive and Reactive Balance Control Mechanisms During Walking

What might be the appropriate balance tests for the clinicians to use to evaluate balance control during walking? Can gait evaluation predict the likelihood of falls or the type of falls in older adults? We suggest that to glean a complete understanding of a person's ability to control balance during walking, both proactive and reactive control mechanisms have to be assessed. In particular, for the proactive control, the control of upper-body stability in both the sagittal and frontal planes, the ability to stabilize the head, the coordination between the upper and lower extremities, and foot placement are all important features for clinicians to consider in their assessment of a patient's balance control ability during walking. Furthermore, to be able to successfully navigate through various environments, attention and the use of visual information to detect environmental changes are important items to be evaluated. At present, only the gait assessments pertaining to the first form of proactive control are available; assessment tools that evaluate a patient's visual perception during motion or attentional demand during walking are yet to be developed. Structured evaluation of reactive postural mechanisms in response to trips, slips, missteps, or uneven terrain are also yet to be developed.

Among the existing balance and gait assessment instruments, there are a few that may be particularly useful for clinicians. One such instrument is Wolfson's Gait Abnormality Rating Scale (GARS).[63] This scale consists of observational evaluation of 16 features of gait pattern based on videotaped records. The GARS uses a four-point rating scale: 0=normal, 1=mildly impaired, 2=moderately impaired, and 3=severely impaired. A higher score indicates more impaired gait. All except three items (head forward, shoulder elevation, and upper trunk forward) of the GARS show high interrater reliability (r=.73-.95). When the validity of this scale was tested by relating GARS results to fall history in 49 nursing home residents (27 had a history of recent falls) and 22 control subjects, the results showed that the GARS score correlated well with walking speed and stride length, and GARS scores were higher for older fallers than for the control subjects. Among the 16 items of rating, arm-swing amplitude, upper- and lower-extremity synchrony, and guardedness of gait were found to best distinguish the older fallers from the other subjects. Thus, the GARS appeared to be a valid, simple assessment tool for predicting the history of falls. These findings appeared to be congruent with our hypothesis that assessment of the upper-body balance is important for balance control during gait. Unfortunately, the types of falls were not documented in this study, and therefore it is difficult to make an association between performance on the GARS and the type of falls.

Recently, VanSwearingen et al[64] developed a modified version of the GARS (GARS-M). The intent was to simplify the GARS and to predict fall history of community-dwelling, frail older adults. In this study, frail older adults were defined as adults over 60 years of age with difficulty in one to three activities of daily living and decreased physiologic reserve. Seven of the 16 GARS items remained in the OARS-M test. These items were variability, guardedness, staggering, foot contact, hip range of motion, shoulder extension, and arm heel-strike synchrony. They found that the OARS-M scores distinguished between these older adults with and without a history of recurrent falls. Similar to the OARS, however, the validity of using this test to predict the types of falls is unknown.

Why should we predict the type of falls? Topper et al[15] classified falls into three categories based on the biomechanical characteristics: (1) falls due to perturbation to the BOS, (2) falls due to perturbation to the COM, and (3) falls with no apparent biomechanical perturbation. Falls due to perturbation of the BOS may indicate an impaired reactive balance control mechanism in an older individual. Falls due to perturbations to the COM can be due to an impairment in the proactive or reactive, or both, mechanisms. Thus, knowledge of the type of fall may assist clinicians in identifying the balance control mechanisms, proactive or reactive, that precipitated the fall. The hypothesis that the type of fall may be highly related to impairment in different balance control mechanisms is supported by the results of the study by Topper and associates.[15] They found that Tinetti's activity-based balance and gait tests were predictive of falls with no obvious biomechanical precipitant and of falls precipitated by a COM perturbation, but not of falls precipitated by a BOS perturbation.

The Berg Balance Scale has been found to be limited in predicting the likelihood of falls among older adults living in independent life care communities in a recent longitudinal study by Thorbahn and Newton.[65] This finding was not surprising because the Berg Balance Scale does not include any items pertaining to gait, and most falls among older adults occur during walking.

Dynamic Balance Training

Winstein[66] documented that balance skills developed through training that involves standing tasks have limited effects on balance control during walking in patients with stroke. Therefore, exercise training focused on improving balance control during walking should include dynamic tasks that involve multisegmental control. Training should also include sensory, motor, and cognition components to improve the acuity and integration ability of the sensory system, effectiveness of the motor system, and vigilance of older adults. Specifically, for the sensory system, visual acuity, depth perception, and motion perception are particularly important for a person to detect an unexpected obstacle in advance. The vestibular system interacts closely with the visual system during walking to stabilize the gaze. An impaired somatosensory system could affect the reactive postural control mechanism. General health, mental status, and certain medications can influence vigilance, attention, or function of the vestibular system.

We suggest that exercise that emphasizes fast and powerful muscle activity generation, such as brisk walking, is necessary for the reactive balance control mechanism to be effective and efficient. Training should include interlimb coordination as well as the coordination between the lower-extremity and upper-body movements. This multisegmental coordination will ensure better safety in case the early postural responses fail to lead to complete balance recovery. This type of training would increase the number of balance response repertoires that older adults could use to supplement the inefficient early postural responses. For example, Wolf and colleagues [67] reported that tai chi exercise improves balance control in older adults. Although tai chi exercise uses slow movements, the beneficial effect of tai chi exercise on balance control could be due to the dynamic nature of this activity, in that it requires complex whole-body coordination.

References

[1] Bohannon RW, Larkin PA, Cook AC, et al. Decrease in timed balance test scores with aging. Phys Ther. 1984;64:1067-1070. [2] Duncan PW, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol. 1990;45:M192-M197.

[3] Inglin B, Woollacott MH. Age-related changes in anticipatory postural adjustments associated with arm movements. J Gerontol. 1988;41: M105-M113.

[4] McIlroy WE. Maki BE. Age-related changes in compensatory stepping it response to unpredictable perturbations. J Geriatr 1996;51:M289-M296.

[5] Sheldon JH. The effect of age on the control of sway. Gerontologica Clinica. 1963;5:129-138.

[6] Stelmach GE, Teasdale N, Di Fabio RP, Phillips J. Age-related decline in postural control mechanisms. Int J Aging Hum Dev. 1989;29:205-223.

[7] Woollacott MH, Shumway-Cook A, Nashner LM. Aging and posture control: changes in sensory organization and muscular coordination. Int J Aging Hum Dev. 1986;23:97-114.

[8] Berg K, Wood-Dauphinee S, Williams JI, Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiotherapy Canada. 1989;41:304-311.

[9] Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc. 1986;34:119-126.

[10] Gabell A, Simons MA, Nayak USL. Falls in the healthy elderly: predisposing causes. Ergonomics. 1985;28:965-975.

[11] Lord SR, Ward JA, Williams P, Anstey KJ. An epidemiological study of falls in older community dwelling women: the Randwick falls and fractures study. Aust J Public Health. 1993;17:240-245.

[12] Overstall PW, Exton-Smith AN, Imms FJ, Johnson AL. Falls in the elderly related to postural imbalance. BMJ. January 1977:261-264.

[13] Sheldon JH. On the natural history of falls in old age. BMJ. December 1960:1685-1690.

[14] Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988;319:1701-1707.

[15] Topper AK, Maki BE, Holliday Pl. Are activity-based assessment of balance and gait in the elderly predictive of risk of falling and/or type of fall? J Am Geriatr Soc. 1993;41:479-487.

[16] Nashner LM. Practical biomechanics and physiology of balance. In: Jacobson GP, Newman CW, Kartush JM, eds. Handbook of Balance Function Testing. St Louis, Mo: Mosby-Year Book Inc; 1993:261-279.

[17] Winter DA. A B C (Anatomy, Biomechanics, and Control) of Balance During Standing and Walking. Waterloo, Ontario, Canada: Waterloo Biomechanics; 1995.

[18] Patla AE. Age-related changes in visually guided locomotion over different terrains: major issues. In: Stelmach GE, Homberg V, eds. Sensorimotor Impairment in the Elderly. Amsterdam, the Netherlands: Kluwer Academic Publishers; 1993:231-252.

[19] Patla AK. Neurobiomechanical bases for the control of human locomotion. In: Bronstein A, Brandt TH, Woollacott MH, eds. Clinical Aspects of Balance and Gait Disorders. London, England: Edward Arnold (Publishers) Ltd; 1996:19-40.

[20] Shik ML, Orlovsky GN. Neurophysiology of locomotor automatism. Physiol Rev. 1976;56:465-501.

[21] Mori S, Sakamoto T, Ohta Y, et al. Site-specific postural and locomotor changes evoked in awake, freely moving intact cats by stimulating the brainstem. Brain Res. 1989;505:66-74.

[22] Patla AE. Visual control of human locomotion. In: Patla AE, ed. Adaptability of Human Gait. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1991:55-95.

[23] Burleigh AL, Horak FB, Malouin F. Modification of postural responses and step initiation: evidence for goal-directed postural interactions. J Neurophysiol. 1994;72:2892-2902.

[24] Burleigh AL, Horak FB. Influence of instruction, prediction, and afferent sensory information on postural organization of step initiation. J Neurophysiol. 1996;75:1619-1628.

[25] Jian Y, Winter DA, Gilchrist IL. Trajectory of the body COG and COP during initiation and termination of gait. Gait and Posture. 1993;1:9-22.

[26] Frank JS, Winter DA, Craik RL. Gait disorders and falls in the elderly. In: Bronstein AM, Brandt T, Woollacott MH, eds. Clinical Disorders of Balance, Posture, and Gait. London, England: Edward Arnold (Publishers) Ltd; 1996:287-300.

[27] Sutherland DA, Kauman KR, Moitoza JR. Kinematics of normal human walking. In: Rose J, Gamble JG, eds. Human Walking. Baltimore, Md: Williams & Wilkins; 1995:23-44.

[28] Shimba T. An estimation of center of gravity from force platform data. J Biomech. 1984;17:53-60.

[29] MacKinnon CD, Winter DA. Control of whole body balance in the frontal plane during human walking. J Biomech. 1993;26:633-644.

[30] Dofferhof ASM, Vink P. The stabilising function of the mm iliocostales and the mm multifidi during walking. J Anat. 1985;140:329-336.

[31] Winter DA, Ruder CK, MacKinnon CD. Control of balance of upper body during gait. In: Winters JM, Woo S-LY, eds. Multiple Muscle Systems: Biomechanical and Movement Organization. New York, NY: Springer-Verlag New York Inc; 1990:534-541.

[32] Blanke DJ, Hageman PA. Comparison of gait of young and elderly men. Phys Ther. 1989;69:144-148.

[33] Guimaraes RM, Issacs B. Characteristics of the gait in old people who fall. Int Rehabil Med. 1980;2:177-180.

[34] Gabell A, Nakay USL. The effect of age on variability in gait. J Gerontol. 1984;39:662-666

[35] Winter DA. Biomechanics and Motor Control of Human Movement. New York, NY: John Wiley & Sons Inc; 1991.

[36] Murray MP, Kory RC, Clarkson BH. Walking patterns in healthy old men. J Bone Joint Surg [Am]. 1969;24:169-178.

[37] Crowninshield RD, Brand RA, Johnston RC. The effects of walking velocity and age on hip kinematics and kinetics. Clin Orthop. 1978;132: 140-144.

[38] Pozzo T, Berthoz A, Lefort L. Head stabilization during various locomotor tasks in humans. Exp Brain Res. 1990;82:97-106.

[39] Prince F, Winter DA, Stergiou P, Walt SE. Anticipatory control of upper body balance during human locomotion. Gait and Posture. 1994;2:19-25.

[40] Winter DA. The Biomechanics and Motor Control of Human Gait: Normal, Elderly and Pathological. Waterloo, Ontario, Canada: University of Waterloo Press; 1991.

[41] Shumway-Cook A, Woollacott MH. Motor Control: Theory and Practical Applications. Baltimore, Md: Williams & Wilkins; 1995.

[42] Alexander NB. Gait disorders in older adults. J Am Geriatr Soc. 1996;44:434-451.

[43] Nutt JG, Marsden CD, Thompson PD. Human walking and higher-level gait disorders, particularly in the elderly. Neurology. 1993;43:268-279.

[44] Wolfson LI, Whipple R, Amerman P, et al. Gait and balance in the elderly. Clin Geriatr Med. 1985;1:649-659.

[45] Patla AE, Prentice S, Robinson C, Neufeld J. Visual control of locomotion: strategies for changing direction and for going over obstacles. J Exp Psychol Hum Percept Perform. 1991;17:603-634.

[46] Patla AE, Rietdyk S. Visual control of limb trajectory over obstacles during locomotion: effect of obstacle height and width. Gait and Posture. 1993;1:45-60.

[47] Patla AE, Robinson C, Samways M, Armstrong CJ. Visual control of step length during overground locomotion: task-specific modulation of the locomotor synergy. J Exp Psychol Hum Percept Perform. 1989;15: 603-617.

[48] Chen H-C, Ashton-Miller JA, Alexander NB, Schultz AB. Stepping over obstacles: gait patterns of healthy young and old adults. J Gerontol 1991;46:M196-M203.

[49] Chen H-C, Ashton-Miller JA, Alexander NB, Schultz AB. Effects of age and variable response time on ability to step over an obstacle. J Gerontol. 1994;49:M227-M233.

[50] Chen H-C, Schultz AB, Ashton-Miller JA, et al. Stepping over obstacles: dividing attention impairs performance of old more than young adults. J Gerontol. 1996;51:M116-M122.

[51] Teasdale N, Lajoie Y, Bard Y, et al. Cognitive processes involved for maintaining postural stability while standing and walking. In: Stelmach CE, Homberg V, eds. Sensorimotor Impairment in the Elderly. Amsterdam, the Netherlands: Kluwer Academic Publishers; 1993:157-168.

[52] Dietz V, Quintern J, Berger W. Corrective reactions to stumbling in man: functional significance of spinal and transcortical reflexes. Neurosci Lett. 1984;44:131-135.

[53] Nashner LM. Balance adjustments of humans perturbed while walking. J Neurophysiol. 1980;44:650-664.

[54] Nashner LM. Fixed patterns of rapid postural responses among leg muscles during stance. Exp Brain Res. 1977;30:13-24.

[55] Nashner LM, Woollacott M, Tuma G. Organization of rapid responses to postural and locomotor-like perturbations of standing man. Exp Brain Res. 1979;36:463-476.

[56] Dietz V, Quintern J, Sillem M. Stumbling reactions in man: significance of proprioceptive and pre-programmed mechanisms. J Physiol (Lond). 1987;386:149-163.

[57] Forssberg H, Grillner S, Rossignol S. Phase-dependent reflex reversal during walking in chronic spinal cats. Brain Res. 1975;85:103-107.

[58] Stein RB. Reflex modulation during locomotion: functional significance. In: Patla AK, ed. Adaptability of Human Gait. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1991:21-36.

[59] Yang JF, Stein RB. Phase-dependent reflex reversal in human leg muscles during walking. J Neurophysiol. 1990;63:1109-1117.

[60] Dietz V, Quintern J, Boos G, Berger W. Obstruction of the swing phase during gait: phase-dependent bilateral leg muscle coordination. Brain Res. 1986;384:166-169.

[61] Eng JJ, Winter DA, Patla AK. Strategies for recovery from a trip in early and late swing during human walking. Exp Brain Res. 1994;102: 339-349.

[62] Tang P-F. Balance Adjustments in Perturbed Human Walking: Neuromuscular Control Mechanisms and Effects of Aging. Eugene, Ore: University of Oregon; 1997. Doctoral dissertation.

[63] Wolfson L, Whipple R, Amerman P, Tobin JN. Gait assessment in the elderly: a gait abnormality rating scale and its relationship to falls. J Gerontol. 1990;45:M12-M19.

[64] VanSwearingen JM, Paschal KA, Bonino P, Yang J-F. The modified Gait Abnormality Rating Scale for recognizing the risk of recurrent falls in community-dwelling elderly adults. Phys Ther. 1996;76:994-1002.

[65] Thorbahn LDB, Newton RA. Use of Berg balance test to predict falls in elderly persons. Phys Ther. 1996;76:576-585.

[66] Winstein AJ. Standing balance training: effects on balance and locomotion in hemiparetic adults. Arch Phys Med Rehabil. 1989;70:755-762.

[67] Wolf SL, Barnhart HX, Kutner NG, et al. Reducing frailty and falls in older persons: an investigation of tai chi and computerized balance training. J Am Geriatr Soc 1996;44:489-497.

MH Woollacott, PhD, is Professor, Department of Exercise and Movement Science and Institute of Neuroscience, University of Oregon, Eugene OR, 97403 (USA) ([e-mail]). Address all correspondence to Dr Woollacott.

P-F Tang, PhD, PT, is Adjunct Assistant Professor, Department of Exercise and Movement Science and Institute of Neuroscience, University of Oregon.

This work was supported by NIH grant AG05317-06 to Dr Woollacott and a research fellowship from the Geriatrics Section of American Physical Therapy Association to Dr Tang.