Is base of support greater in unsteady gait? (Research Report).

A wide base of support (BOS) has long been believed to be a hallmark of unsteady gait. (1-5) Although static standing stability increases with increases in BOS, such as in tandem versus 30-cm BOS standing, (6) no reports to date have described BOS during gait and its relationship to dynamic stability, as determined by 3-dimensional kinematics during gait analysis.

Base of support in previous research has been defined as horizontal stride width during the double-support phase when both feet are in contact with the ground and the whole-body center of gravity (CG) remains within the BOS. Weller et al (7) found that decreased stride width was closely correlated with fall frequency and dynamic stability. Fried et al (5) and others, (6,7) however, found increased stride width in elderly people who were prone to falls. The majority of the gait cycle is spent in single-limb support (SLS), during which BOS is minimized to the width of the supporting foot. Therefore, during most of the gait cycle, the whole-body CG is outside this narrowed BOS. (8) The term "interfoot distance" (IFD), therefore, is a more accurate term to describe stride width throughout the dynamics of gait. Interfoot distance is the mediolateral distance between the center of mass of the left foot and the center of mass of the right foot, which in turn is located along the longitudinal midline of the feet (Fig. 1). Therefore, the IFD represents the purely frontal-plane interfoot separation in global coordinates.

[FIGURE 1 OMITTED]

Vestibulopathy impairs one of the 3 primary sensory modalities (vision, proprioception, and the labyrinth structures) responsible for balance information. Vestibulopathic symptoms can be caused by the vestibular system reacting too little, asymmetrically, or inappropriately to stimuli. (9) Even if the visual and proprioceptive sensory systems are intact, many people with vestibulopathy are unsteady as they walk, climb stairs, and transfer from a bed or chair. Common complaints of people with vestibulopathy include sensations of abnormal movements, dizziness, or vertigo and visual disturbances such as nystagmus or oscillopsia, (10,11) Instability worsens as other sensory inputs are removed (eg, while walking at night, standing with eyes closed). (2,12)

Most investigations of balance or instability among people with vestibulopathy have studied static standing activities alone. However, many people with vestibulopathy exhibit dynamic instabilities during gait such as ataxia, decreased trunk and head rotation, decreased gait speed, and "a widened base of support." (13) When gait has been studied in other people, most often only lower-extremity kinematics or time-distance variables have been examined. Although the measurements obtained in such studies reveal details of human locomotion, they cannot disclose global balance impairment indicators such as whole-body CG control or IFD throughout the gait cycle.

The purpose of our study was to assess IFD throughout the gait cycle in people with vestibulopathy as compared with gait in people without known neuromuscular pathology. We hypothesized that (1) subjects with vestibulopathy would have greater mediolateral IFD than matched subjects without neuromuscular pathology during gait and (2) mediolateral IFD would be correlated with other indicators of stability (ie, walking speed, mediolateral CG excursion, mediolateral CG/center of pressure [CG/CP] difference, and whole-body CG/CP moment arm) (7) (Fig. 2). In short, if IFD at heel strike, the usual measure of BOS, is a valuable indicator of gait stability, IFD could be used to differentiate subjects with vestibulopathy and unsteady gait from comparison subjects without neuromuscular pathology, and IFD should correlate with other balance indicators during gait.

[FIGURE 2 OMITTED]

Method

Subjects

Twenty-two subjects with vestibulopathy (mean age=55 years, range=25-90) and 22 comparison subjects (mean age=53 years, range=26-88) were studied. The comparison subjects had no known neuromusculoskeletal dysfunction and exercised vigorously at least 3 times per week. The groups were matched by age, height, weight, and body mass index (14) (Tab. 1). Subjects with vestibulopathy reported complaints of unsteadiness, for which they requested vestibular rehabilitation, and were diagnosed with vestibulo-ocular reflex abnormalities. These subjects' symptoms were stable for at least 2 months prior to the study. Onsets of symptoms ranged from 6 months to more than 10 years prior to entry to the study. Etiologies of vestibulopathy included ototoxicity, bilateral inner ear infections, and idiopathic inner ear damage. Subjects were excluded if evidence of other disease processes that might impair balance existed, such as a prior stroke, Meniere disease, perilymph fistula, other central nervous system pathology, peripheral neuropathy, severe arthropathy, orthopedic deformities, or severe visual impairment. (15) All subjects agreed to be studied by written consent and no monetary compensation was offered.

Sixteen of the subjects with vestibulopathy had bilateral vestibular hypofunction, 6 with bilaterally undetectable ("zero") function ([less than or equal to] 0.1 vestibulo-ocular reflex gains during 0.05-Hz sinusoidal vertical axis rotation [SVAR]) and 10 with bilateral dysfunction (0.1-0.25 vestibulo-ocular reflex gains during SVAR). Six subjects had unilateral vestibular dysfunction (unilaterally abnormal SVAR findings or absent caloric responses). (16) No subject had received vestibular rehabilitation prior to testing.

Instrumentation

The kinematic system, an 11-segment whole-body model, that we used to assess posture and balance is described in detail elsewhere (17-19) and briefly recounted here. Simultaneous bilateral whole-body kinematic data were collected with a motion analysis system using Selspot II hardware * and were analyzed with TRACK (Telemetered Rapid Acquisition of Kinematics) kinematic software [dagger] and software developed in the Massachusetts General Hospital Biomotion Laboratory (Boston, Mass). Light-emitting diode arrays were mounted on 1 body segments: right and left feet, shanks, thighs, and arms as well as the pelvis, trunk, and head. The system calculated the 6-degree-of-freedom position of each body segment within a 2x2x2-[m.sup.3] viewing volume. Segment masses and mass centers were based on regression equations from 2 stereophotogrammetric studies and were expressed as displacements from the initial position in the laboratory global coordinate system. (20-23) The anthropometric data incorporated into the TRACK software package were acquired from adult male and female subjects without known neuromuscular pathology. (21,23) Using the segment masses and center-of-mass kinematics, the kinematics of the whole-body center of mass was estimated with regression equations. (17) System precision was within 1 mm in linear displacement, and orientation was within 1 degree in angular displacement.

Kinetic data were recorded with 2 Kistler force platforms [double dagger] together with kinematic data at a sampling rate of 150 Hz. The force platforms were located in the approximate center of our viewing volume, level with the walking surface of the laboratory and hidden from the subjects' view by carpeting. The CP was calculated from the individual force-plate CPs and the known force-plate locations and orientations. Center-of-pressure displacements also were measured in the laboratory global coordinate system. Another program (SuperPlot2, [section] patent pending) determined individual foot and combined CPs with an accuracy of <3 mm.

Procedure

All measurements for a subject were obtained in a single session. All subjects were barefoot and walked without assistive devices. At least 1.5 minutes of rest was given between trials. Several practice trials of each activity were performed before data collection. Two trials each of each subject's gait at preferred speed and paced gait (120 steps/min) over a 10-m walkway were collected where one foot landed completely on one of the force platforms. Gait at preferred speed should reflect movement strategies under self-selected optimal neuromotor control; paced gait at 120 steps/min permits valid cadence-controlled within- and between-subject comparisons. (24) A 120-step/min cadence was chosen according to previous studies and pilot studies to match the average preferred gait speed among the comparison subjects, as well as to include all subjects with functional limitations. (25,26) Gait at preferred speed was performed first to prevent the preferred cadence from being influenced by the paced cadence.

Prior to gait trials at preferred speed, subjects were instructed as follows: "Move forward in as straight a line as possible, walking at your normal pace, as if you were taking a brisk walk in the park." For paced gait trials, subjects were first asked to walk in place to the beat set by a metronome set to 120 beats/min. All subjects walked at the same pace so that the different gait variables could be compared without potentially confounding cadence differences. They were then instructed as in the previous gait trials at preferred speed, "but walk to this pace." Dependent variables measured included the following: IFD, mediolateral CG excursion, maximum mediolateral CG/CP difference, whole-body CG-CP moment arm, and CG forward velocity. Within a gait cycle, walking speed is the CG forward velocity (in laboratory/global coordinates). Mediolateral CG excursion is the difference between maximum and minimum CG positions in the frontal plane. Maximum mediolateral CG/CP difference occurs when CG and CP are farthest apart in the frontal plane. Whole-body CG-CP moment arm is defined as the combined anteroposterior and mediolateral vector root mean square difference between the whole-body CG and CP during SLS. (17) When the CG and CP are most separated, equilibrium is least, and therefore gait stability is most challenged (Fig. 2). (16) The IFD was measured throughout the gait cycle, enabling us to report values during SLS as well as during the more traditionally reported double-support phase. Three IFD values were taken for statistical analysis: IFD range, IFD at heel-strike, and IFD during SLS. The IFD range is calculated as the maximum IFD value minus the minimum IFD value occurring in succession within a stride. The IFD at heel-strike, traditionally termed "base of support," is thought to be important because it represents the initiation of the double-support phase of gait when both feet are not flat on the ground and the body is in an inherent state of instability. (1) The value is obtained by measuring the IFD at heel-strike, as determined by force-plate initial deflection. The value for IFD during SLS is obtained by measuring the IFD when the whole-body CG is at its maximum vertical height where traditional BOS values considered only the single supporting foot. The maximum and average values of mediolateral CG excursion, mediolateral CG/CP difference, whole-body CG-CP moment arm, and CG forward velocity were obtained from the same individual gait cycles.

Data Analysis

A computer program (SuperPlot2) displays kinematic variables throughout each gait trial. Specific data points for analysis were manually selected by using the computer mouse to point a cursor toward the point of interest on the screen and recorded by reading them from the screen. Values recorded for 2 trials were averaged for statistical analysis. In subjects where only one trial of data was obtained (eg, they missed the force plate on one trial), that valid trial's values were used. Because observations with missing values were not included in the statistical analysis, 40 observations (out of 44 possible) were used for statistical analysis of variables for gait at preferred speed and 42 (out of 44 possible) were used for statistical analysis of variables for paced gait. To quantify variability in IFD during gait, the absolute difference between the IFD values of 2 trials was determined for both gait at preferred speed and paced gait.

All quantitative data were analyzed descriptively and with inferential statistics. Graphic results for raw IFD data were studied for shape, timing, and reproducibility (Fig. 3). Descriptive statistics were calculated for all variables during gait at preferred speed and paced gait. The main hypothesis, that subjects with vestibulopathy would have different IFD values from those of the comparison subjects, was tested with repeated-measures multivariate analysis of variance. To test the hypothesis predicting correlations between IFD and kinematic variables, Pearson correlation coefficients were calculated for all subjects combined and for each group separately. All statistical analyses were performed using the Statistical Analysis System 6.04 statistical analysis program. [parallel] An alpha (significance) level of .05 was used.

[FIGURE 3 OMITTED]

Results

During gait at preferred speed, IFD did not differ between groups, varying from 0.86 to 35.85 cm, but the subjects with vestibulopathy used smaller whole-body CG-CP moment arms than the comparison subjects. Only IFD range and IFD in SLS during paced gait were greater (P<.01) in the subjects with vestibulopathy than in the comparison subjects; IFD was not different at heel-strike in paced gait (Fig. 4). Therefore, hypothesis 1 was not supported: BOS did not differentiate subjects who were unsteady from comparison subjects. During paced gait, subjects with VSP increased their IFD range compared with free gait, whereas comparison subjects' IFD range decreased slightly (mean IFD [[+ or -] SD] for subjects with vestibulopathy=8.11 [+ or -] 4.48 cm; mean IFD for comparison subjects=5.98 [+ or -] 3.21 cm; Fig. 4). We found that timing of maximum IFD was more variable among the subjects with vestibulopathy than among the comparison subjects, occurring sometimes during SLS and sometimes during the double-support phase of gait (Fig. 2).

[FIGURE 4 OMITTED]

Descriptive statistics of whole-body kinematic variables are presented in Table 2. As a whole, the subjects with vestibulopathy had slower average and peak CG forward walking velocities during both gait conditions (P<.05). Table 3 shows correlations between IFD at heel-strike and CG stability measures; the absolute difference in IFD range between 2 trials of gait at preferred speed was different from that for paced gait when combined across subjects (P=.026). The IFD at heel-strike was inversely related to forward velocity for all subjects. The IFD at heel-strike was directly related to both medioateral CG excursion and mediolateral CG/CP difference for the subjects with vestibulopathy during gait at preferred speed, for the comparison subjects during paced gait, and for all subjects combined during both gait conditions (Tab. 3).

Discussion

In contrast to conventional clinical wisdom, our main hypothesis, that unsteady subjects with vestibulopathy would use a greater BOS than subjects without neuromuscular pathology during gait, was not supported. The IFD value at heel-strike corresponded to the "usual" measurement of BOS during gait and did not differ for gait at preferred speed or paced gait. No IFD group differences were found during gait at preferred speed. The IFD range and IFD during SLS, however, were different between groups during paced gait. Apparently, during gait at preferred speed, subjects with vestibulopathy were able to compensate for dynamic instability through adjustment of other kinematic variables such as diminished walking speed (decreased CG forward velocity) and decreased whole-body CG-CP moment arm. Paced gait, however, encouraged subjects with vestibulopathy to walk faster than their preferred speed, apparently challenging the limits of their dynamic stability. The differences in IFD values reported here may be indicative of individuals' attempts to compensate for dynamic instability.

Previous Studies

Several authors have described BOS data during gait; however, none have studied IFD during the swing phase of gait. Murray et al (27) reported stride width measurements on a group of 30 men without known pathology, aged 20 to 65 years, using interrupted light photography. Mean stride width was 7.7 cm (SD=3.5) during free-speed gait (120 steps/min) and 9.1 cm (SD=4.1) during fast-speed gait (138 steps/min). Murray et al, (28) in a later study, found similar values. Gabell and Nayak (29) measured stride width as determined by a metallic marker on the subject's shoe landing on a metal runway. Results varied from 5.05 to 17.5 cm (median=9.6 cm) for a group of young subjects and from 3.72 to 15.2 cm (median=7.48 cm) for a group of older subjects. Decreased heel width in the older subjects could have been due to increased toe-out or foot angle, which moves the heels closer together. (12,30) These investigators defined stride width as the mediolateral distance between either the medial or lateral malleoli during full floor contact.

Two research reports (31,32) support the use of BOS to predict falls among elderly subjects. In both studies, stride width values were obtained by measuring mediolateral heel distance during the double-support phase of gait. Heitmann et al (31) reported average stride width values of 7.44 cm (SD=2.68) for a group of subjects who were prone to falling and 6.54 cm (SD=3.11) for a group of subjects who were not prone to falling. Gehlsen and Whaley (32) reported heel-width measurements obtained for elderly subjects walking on a treadmill at slow and fast speeds. The average heel-width measurements were 7.19 cm (SD=1.58) at 4 km/h and 6.41 cm (SD=1.54) at 6 km/h for subjects without a history of falls and 7.77 cm (SD=1.47) at 4 km/h and 7.39 (SD=1.06) at 6 km/h for subjects with a history of falls. Our comparison subjects' IFD at initial heel contact was 13 cm, as compared with approximately 7 to 8 cm found in other studies. We believe that the greater IFD in our subjects reflects normal toe-out and toe-in/eversion at the ankle and foot, which affects foot CG location, but may not be reflected in mediolateral heel or malleolar distance measures.

The measurement technique used in our study differs from the techniques of previous studies. Previous researchers used 2-dimensional analysis, but we used a 3-dimensional kinematic analysis system. In previous research of IFD during gait, measurements were taken between the malleoli (medial or lateral) or between the heels. Our system determines IFD by calculating the medioateral distance between the longitudinal foot mid-lines. Because feet are usually turned somewhat outward during stance, the part of the foot used as a target will influence the actual IFD measurement. At initial heel contact, the foot is rotated slightly outward and inverted due to supination at the subtalar and midtarsal joints. By mid-SLS, the foot has rotated inward and everted, with the subtalar and midtarsal joints falling into pronation. By toe-off, the foot has reached maximum inversion and is toeing out. (33) Murray (28) noted foot angle (toe-in/toe-out) to be an influence on stride width. An increase in toe-out represents a slightly broader mediolateral stride width. We believe our technique more accurately reflects functional BOS during gait.

Our data are consistent with those of previous investigators in revealing substantial variability among individuals without neuromuscular pathology. Perry (34) reported that the BOS of subjects without neuromuscular pathology varied from 20 cm to crossing one foot over the other. Our IFD measurements ranged from 0.86 to 35.85 cm during gait at preferred speed. Our data do not support the findings of slight increases in BOS with increases in walking speed noted by Murray et al (27) and Inman et al, (33) but rather that BOS varied inversely with measurements of preferred walking velocity (r=-.39, Tab. 3). In cadence-controlled paced gait, however, IFD at heel-strike did not correlate with walking speed (Tab. 3).

Heel-strike and mid-SLS represent 2 clearly different events during the gait cycle. Differences in IFD between these 2 gait events may suggest that a different degree of dynamic instability occurs at different times during the gait cycle. The values for IFD at heel-strike and SLS, however, were not statistically different; heel-strike usually occurs as the feet draw closer together. In many instances across all subjects, the feet separated as SLS occurred, resulting in similar values for these 2 measures. Thus, heel-strike and SLS measurements may not reflect actual maximum and minimum IFD values, respectively, and we therefore obtained "IFD range" values.

IFD Inconsistency

Between-subject inconsistency (high standard deviations) was evident in both groups of subjects for all IFD measures (Fig. 4). Heitmann et al (31) also found high stride-width inconsistency. Winter et al (35) and Gabell et al (29) reported high between-subject IFD variability.

Our data suggest which method best measures IFD variability. The IFD range was intended to indicate an individual subject's DIFD inconsistency. The maximum and minimum values were determined by selecting a curve reversal on the IFD plot (Fig. 3). This measure was intended to isolate a single IFD adjustment. Typically, however, subjects with vestibulopathy show more frequent curve reversals than subjects without pathology, limiting the size of the IFD variable as currently defined. Perhaps selecting the largest and smallest actual values obtained within a gait cycle may reveal a better measure of variability.

IFD as Indicator of Stability

Stability during gait and locomotion can be quantified in a variety of ways. Cycle time, double-support time, and stance duration have been shown to vary inversely with balance control. (17) Our data indicate that IFD at heel-strike decreases as walking speed increases, irrespective of subject group (ie, with or without vestibulopathy). The IFD at heel-strike, similar to usual measures

of BOS, did not differentiate subjects with unsteady gait from comparison subjects without pathology. Therefore, our data suggest that conventional BOS measures may not be as helpful as simpler measures such as walking speed. Other researchers have identified altered head kinematics (36) and CG control (24) in subjects with vestibulopathy as compared with subjects without pathology. In our study, weak correlations were noted between IFD values at heel-strike and mediolateral CG measurements (mediolateral CG excursion and mediolateral CG/CP difference) (Tab. 3). Other investigators (37) have suggested the existence of multiple dynamic stability determinants. Our results support that contention, unsubstantiated in our second hypothesis, because we found differences between groups for maximum whole-body CG-CP moment arm during gait at preferred speed, IFD range, and IFD at SLS during paced gait, as well as maximum CG forward and average CG velocities and CG mediolateral velocity during both gait conditions (P<.05).

Clinical Implications

Research suggests that vestibular rehabilitation does not "cure" the vestibulopathy, but rather enhances an individual's compensation during locomotor activities of daily living, permitting greater adaptation to the dynamically changing challenges to gait stability. (16,38) Our results suggest that patients with vestibulopathy walk with more variable intersubject IFDs and that changes in IFD were correlated with measurements of walking speed, mediolateral CG excursion, mediolateral CG/CP difference, and whole body CG-CP moment arm. We suggest that changes in all of these whole-body kinematic variables are mechanical compensations of vestibulopathic instability. Because all subjects with vestibulopathy were tested prior to a course of vestibular physical therapy, it is not known whether these indexes of global stability improve after treatment. Development of dynamic active control of these variables, as well as head and eye control, should be investigated in future studies of vestibular rehabilitation efficacy.

The motion analysis system used in our study is not practical for or accessible to most physical therapists. Perhaps the primary utility of our data is to enable us to begin to quantify dynamic instability during gait. These data indicate that although lower-extremity gait kinematic studies including IFD variables are useful, full-body CG and CP variables during gait were more informative indexes of global stability.

Further comparative studies between subjects with vestibulopathy and subjects with other balance impairments are warranted to investigate differences in IFD and full-body CG and CP variables. We studied all subjects with vestibulopathy as a single pathologic group. One suggestion for future study is to compare other subjects with vestibular dysfunction, cerebellar pathology, and other impairments that produce gait unsteadiness.

Conclusions

Subjects with vestibulopathy exhibited IFD displacement and timing patterns that were more variable than IFD values of comparison subjects. Subjects with vestibulopathy used more variable IFDs when required to walk (during paced gait) at a faster pace; during free gait, IFD values were correlated with measurements of walking speed, mediolateral CG excursion, mediolateral CG/CP difference, and whole-body maximum moment arm.

When required to walk at a paced gait, subjects with vestibulopathy walked faster than their preferred pace. During preferred gait speed, we found no difference between groups for IFD. Our findings, along with the correlations between IFD measures and other whole-body kinematic variables, suggest that foot trajectory compensation for vestibulopathic instability occurs throughout the gait cycle.

Table 1.
Subject Characteristics

                         Comparison Subjects (a)        Subjects With
                         (n=22)                         Vestibulopathy
                                                       (n=22)

Male                     12                            18
Female                   10                             4

                         X       SD      Range         X       SD

Age (y)                  52.62   21.50   26.00-88.00   55.21   21.02
Height (m)                1.70    0.11    1.52-1.86     1.66    0.10
Weight (kg)              66.45   10.61   50.00-85.45   65.48   13.94
BMI (b) (kg/[m.sup.2])   22.86   2.11    19.72-27.03   23.70    4.11

                         Range

Age (y)                  24.90-90.30
Height (m)                1.52-1.87
Weight (kg)              47.73-102.27
BMI (b) (kg/[m.sup.2])   17.79-36.39

(a) Subjects without known neuromuscular pathology.

(b)  BMI=body mass index.
Table 2.
Means and Standard Deviations for Whole-Body Kinematic Variables (a)

                                       Comparison       Subjects With
                                       Subjects (b)     Vestibulopathy
                      All Subjects     (n=22)           (n--22)

                      X        SD      X        SD      X        SD

Preferred gait
  CG M-L EX (cra)       5.09    2.35     5.05    2.73     5.14    2.00
  CG/CP DIFF (cm)       7.62    2.48     7.39    2.30     7.84    2.70
  CG AVG VEL (cm/s)   112.99   19.85   122.11   14.01   104.33   20.98
  CG MAX VEL (cm/s)   122.93   24.27   135.34   15.87   112.21   25.42

Paced gait
  CG M-L EX (cm)        5.08    2.54     5.19    2.91    4.96     2.07
  CG/CP DIFF (cm)       7.93    2.78     7.64    2.59     8.30    3.04
  CG AVG VEL (cm/s)   120.61   19.15   127.04   18.08   113.16   18.01
  CG MAX VEL (cm/s)   132.02   23.22   141.11   20.42   122.50   22.54

(a) CG M-L EX=center-of-gravity mediolateral excursion during a single
gait cycle, CG/CP DIFF=maximum mediolateral center of
gravity - center-of pressure difference, CG AVG VEL=center-of-gravity
average forward velocity, CG MAX VEL=center-of-gravity maximum forward
velocity.

(b) Subjects without known neuromuscular pathology.
Table 3.
Multiple Determinants of Dynamic Stability: Pearson Correlations
Between Interfoot Distance (IFD) at Heel-strike and Center-of-Gravity
Measures During Gait (a)

                               Preferred Gait   Paced Gait
                               IFD at           IFD at
                               Heel-strike      Heel-strike

All subjects
  CG M-L EX                    r=.32            r=.37
                               P=.04            P=.02
  CG/CP DIFF                   r-.55            r=.50
                               P=.0004          P=.002
  CG AVG VEL                   r=-.39
                               P=.014
  CG MAX VEL                   r=-.46
                               P=.003

Comparison subjects
  (n=22)b
  CG M-L EX                                     r=.51
                                                P=.02
  CG/CP DIFF                                    r=.65
                                                P=.002

Subjects with vestibulopathy
  (n=221
  CG M-L EX                    r=.46
                               P=.04
  CG/CP DIFF                   r=.65
                               P=.003

(a) CG M-L EX=center-of-gravity mediolateral excursion, CG/CP
DIFF=maximum mediolateral center of gravity - center of pressure
difference, CG AVG VEL=center of gravity average forward velocity, CG
MAX VEL=center-of-gravity maximum forward velocity. Blanks represent
nonsignificant findings during a single gait cycle.

(b) Subjects without known neuromuscular pathology.

This study was approved by the Institutional Review Board of the Massachusetts General Hospital. This research was supported, in part, by National Institute for Disability and Rehabilitation Research grant H133G30041 and by National Institutes of Health grant R01AG11255.

This article was submitted February 10, 2000, and was accepted August 31, 2001.

* Selspot AB, Flojelbergsgatan 14, S-431 37 Molndal, Sweden: [dagger] Developed at the Massachusetts Institute of Technology, Cambridge, Mass.

[double dagger] Kistler Instruments AG, Winterthur, Switzerland. [section] Massachusetts General Hospital Biomotion Laboratory, Boston, MA 02114.

[parallel] SAS Institute Inc, PO Box 8000, Cary, NC 27511.

References

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

(2) Denny-Brown D. The nature of apraxia. J Nerv Ment Dis. 1958;126: 9-31.

(3) Sudarsky L. Geriatrics: gait disorders in the elderly. N Engl J Med. 1990;332:1441-1446.

(4) Shinner HB, Barrack RL, Cook SD, Haddad RJ Jr. Joint position sense in total knee arthroplasty. J Orthop Res. 1984;1:276-283.

(5) Fried AV, Cwikel J, Ring H, Galinsky D. ELGAM--extra laboratory gait assessment method: identification of risk factors for falls among the elderly at home. Int Disabil Stud. 1990;12:161-164.

(6) Riley PO, Benda BJ, Krebs DE. Phase plane analysis of stability in quiet standing. J Rehabil Res Dev. 1995;32:227-235.

(7) Weller C, O'Neill CJ, Charlett A, et al. Defining small differences in efficacy between anti-parkinsonian agents using gait analysis: a comparison of two controlled release formulations of levodopa/decarboxylase inhibitor. Br J Clin Pharmacol. 1993;35:379-385.

(8) Winter DA, Patla AE, Frank JS, Walt SE. Biomechanical walking pattern changes in the fit and healthy elderly. Phys Ther. 1990;70:340-347.

(9) Baloh RH, Hornrubia V. Clinical Neurophysiology of the Vestibular System. Philadelphia, Pa: FA Davis Co; 1990:130-152.

(10) Cawthorne T. Vestibular injuries. Proc R Soc Med. 1946;39:270-273.

(11) Baloh RW. Dizziness, hearing loss and tinnitus. In: Baloh RW, Hornrubia V, eds. The Essentials of Neuro-otology. Philadelphia, Pa: FA Davis Co; 1984:182-185.

(12) Herdman SJ. Assessment and treatment of balance disorders in the vestibular-deficient patient. In: Duncan PW, ed. Balance: Proceedings of the American Physical Therapy Association Forum, Nashville, Tennessee, June 13-15, 1989. Alexandria, Va: American Physical Therapy Association; 1990:87-94.

(13) Borello-France DF, Whitney SL, Herdman SJ. Assessment of vestibular hypofunction. In: Herdman SJ, ed. Vestibular Rehabilitation. Philadelphia, Pa: FA Davis Co; 1994:265.

(14) Bray GA. Obesity: definition, diagnosis, and disadvantages. Med J Aust. 1985;142:S2-S8.

(15) Herdman SJ. Exercise strategies in vestibular disorders. Ear Nose Throat J. 1989;68:961-964.

(16) Krebs DE, Gill-Body KM, Riley PO, Parker SW. Double-blind, placebo-controlled trial of rehabilitation for bilateral vestibular hypofunction: preliminary report. Otolaryngol Head Neck Surg. 1993;109:735-741.

(17) Riley PO, Mann RW, Hodge WA. Modelling the biomechanics of posture and balance. J Biomech. 1990;23:502-506.

(18) Krebs DE, Wong DK, Riley PO, Hodge WA. Trunk kinematics during locomotor activities. Phys Ther. 1992;72:505-514.

(19) Azen R, Krebs DE, Gill-Body KM. Head Angular Displacements and Accelerations During Standing, Chair Rise, and Gait [MS thesis]. Boston, Mass: MGH Institute of Health Professions; 1992.

(20) Hutchinson EB, Riley PO, Krebs DE. A dynamic analysis of the joint forces and torques during rising from a chair. IEEE. 1994;2(2):49-56.

(21) McConville JT, Clauser CE, Churchill TD, et al. Anthropometric Relationship of Body And Body Segments of Inertia. Technical Report, US Air Force Aerospace Laboratory. Washington, DC: Government Printing Office; 1980.

(22) Patla AE, Frank JS, Winter DA, et al. Age-related changes in balance control systems: initiation of stepping. Clin Biomech. 1993;8:179-184.

(23) Young J, Chandler RF, Snow LL, et al. Anthropometric and Mass Distribution Characteristics of the Adult Female. Technical Report, FAA Civil Aeromedical Institute. Washington, DC: Government Printing Office; 1983.

(24) Tucker CA, Ramirez J, Krebs DE, Riley PO. Center of gravity dynamic stability in normal and vestibulopathic gait. Gait & Posture. 1998;8:117-123.

(25) Murray MP, Drought AB, Kory RC. Walking patterns of normal men. J Bone Joint Surg Am. 1964;46:335-360.

(26) Murray MP. Gait as a total pattern of movement. Am J Phys Med. 1967;46:290-333.

(27) Murray MP, Kory RC, Clarkson BH, Sepic SB. Comparison of free and fast walking patterns of normal men. AmJPhys Med. 1966;45:8-23.

(28) Murray MP, Mollinger LA, Gardner GM, Sepic SB. Kinematic and EMG patterns during slow, free, and fast walking. J Orthop Res. 1984;2:272-280.

(29) Gabell A, Nayak BA. The effect of age on variability in gait. J Gerontol. 1984;39:662-666.

(30) Smidt GL. Aging and gait. In: Smidt GL, ed. Gait in Rehabilitation. Clinics in Physical Therapy. New York, NY: Churchill Livingstone; 1990:185-198.

(31) Heitmann DK, Gossman MR, Shaddeau SA, Jackson JR. Balance performance and step width in noninstitutionalized, elderly, female failers and non failers. Phys Ther. 1989;69:923-931.

(32) Gehlsen GM, Whaley MH. Falls in the elderly, part I: gait. Arch Phys Med Rehabil. 1990;71:735-738.

(33) Inman VT, Ralston HJ, Todd F. Human Walking. Baltimore, Md: Williams & Wilkins; 1981:1-61.

(34) Perry J. The mechanics of walking: a clinical interpretation. Phys Ther. 1967;47:778-801.

(35) Winter DA. The Biomechanics and Motor Control of Human Gait: Normal, Elderly, and Pathological. 2nd ed. Waterloo, Ontario, Canada: University of Waterloo Press; 1991:1-26.

(36) Pozzo T, Berthoz A, Lefort L, Vitte E. Head stabilization during various locomotor tasks in humans, II: patients with bilateral peripheral vestibular deficits. Exp Brain Res. 1991;85:208-217.

(37) Crutchfield CA, Shumway-Cook AS, Horak FB. Balance and coordination training. In: Scully RM, Barnes MR, eds. Physical Therapy. Philadelphia, Pa:JB Lippincott Co; 1989:825-843.

(38) Horak FB,Jones-Rycewicz C, Black FO, Shumway-Cook A. Effects of vestibular rehabilitation on dizziness and imbalance. Otolaryngol Head Neck Surg. 1992;106:175-180.

DE Krebs, PT, PhD, is Professor, MGH Institute of Health Professions, 101 Merrimac St, Boston, MA 02114-4719 (USA) ([e-mail]), and Director, Massachusetts General Hospital Biomotion Laboratory, Boston, Mass. Address all correspondence to Dr Krebs at the first address.

D Goldvasser, MScE, is Research Associate, Massachusetts General Hospital Biomotion Laboratory.

JD Lockert, PT, MS, is in private practice at Advantage Sports Medicine, Woburn, Mass. She was a master's degree student at MGH Institute of Health Professions at the time of this study.

LG Portney, PT, PhD, is Associate Professor, MGH Institute of Health Professions.

KM Gill-Body, PT, MS, NCS, is Assistant Professor, MGH Institute of Health Professions, and Clinical Associate, Physical Therapy Services, Massachusetts General Hospital.

Dr Krebs, Ms Lockert, and Dr Portney provided concept/research design. Dr Krebs, Mr Goldvasser, Ms Lockert, Dr Portney, and Ms Gill-Body provided writing. Dr Krebs, Mr Goldvasser, and Ms Lockert provided data collection, and Dr Krebs, Mr Goldvasser, Ms Lockert, and Dr Portney provided data analysis. Dr Krebs, Mr Goldvasser, and Ms Gill-Body provided subjects. Dr Krebs provided project management, fund procurement, facilities/equipment, institutional liaisons, and clerical support, and Mr Goldvasser provided facilities/equipment and institutional liaisons. Dr Krebs, Ms Lockert, Dr Portney, and Ms Gill-Body provided consultation (including review of manuscript before submission). The authors acknowledge the assistance of Peter Velyvis, Jose V Ramirez, MD, Scott A Banks, PhD, Patrick O Riley, PhD, and Richard Kirkpatrick, Jr, MD, in data acquisition and processing.