The Relation Between Limb Loading and Control Parameters of Gait Initiation in Persons With Stroke

627 The Relation Between Limb Loading and Control Parameters of Gait Initiation in Persons With Stroke Denis Brunt, EdD, PT, Darl W. Vander Linden, PhD, PT, Andrea L. Behrman, MSc, PT ABSTRACT. Brunt D, Vander Linden DW, Behrman AL. The relation between limb loading and control parameters of gait initiation in persons with stroke. Arch Phys Med Rehabil 1995;76:627-34. Objective: This study investigated the relation between limb loading and selected characteristics of gait initiation in patients after stroke. Subjects and Setting: Thirteen patients attending a rehabilitation clinic volunteered for the study. Design: For the description of clinical features, patients were divided into two groups dependent on the amount of body weight shared by the involved limb during stance before gait initiation. Main Outcome Measures: Patients performed six trials of gait initiation with either their involved or noninvolved limb on a force platform. Peak ground reaction forces and bilateral tibialis anterior and gastrocnemius electromyographic (EMG) activity were used for group comparison. Results: All patients showed the correct loading and unloading characteristics in the vertical and medial-lateral direction during gait initiation. Strong correlations were noted between initial limb loading and ground reaction forces during gait initiation (r =.79 to.95). Changes in ground reaction forces were significantly less (p <.001) for those patients who demonstrated decreased weight bearing on their involved limb before gait initiation. These patients were also unable to generate forward momentum, as evidenced by the fore-aft ground reaction force, with the involved limb. For all patients, increased gastrocnemius activity was noted in the stance (noninvolved) limb. The data are further discussed in regard to the relationship of the interaction of bilateral EMG activity and ground reaction forces. Conclusion: This study suggests that there is a correlation between symmetrical weight bearing and the ability to provide those forces that generate forward momentum in the initiation of gait. 1995 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation There is evidence to support a positive relationship between limb loading and muscle activity during perturbation of upright posture and gait. For example, when upright posture was displaced during assymetrical stance, electromyographic (EMG) activity was shown to be absent t or reduced 2 in the unloaded limb. Similar data have been reported for gait where the EMG response was greater to perturbation during early stance as compared with late stance when that limb became relatively unloaded. 3 In perturbation studies, individual limb responses are peripherally driven to modulate the selected combination of neural networks that are a result of the centrally determined set. 4 In the initiation of voluntary movement, however, prior knowledge of limb loading may be used in determining the central set as opposed to modification of the intended set that follows the disturbance of the center of mass. Persons with unequal limb loading owing to pathology therefore should show diminished EMG activity in the involved limb during voluntary movement that could effect the dynamics of the task. This may well be the case for persons with stroke where unequal limb loading exists during standing 5 and has been reported to effect gait parameters 6 and other mobility tasks] More recent data have shown that persons with stroke failed to From the Department of Physical Therapy (Dr. Brunt), University of Florida, Gainesvine; Department of Physical Therapy (Dr. Vander Linden), Eastern Washington University, Spokane, WA; and Departments of Physical Therapy and Exercise Science (Dr. Behrman), University of Florida, Gainesville. Submitted for publication November 1, 1994. Accepted in revised form February 6, 1995. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. 1995 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003-9993/95/7607-328653.00/0 proportion their weight appropriately during a simple leg flexion task in standing and were unable to provide appropriate limb loading and propulsive forces. 8 This limb flexion task is similar in many respects to the process of gait initiation except that it does not require the generation of forward momentum. In healthy subjects gait initiation is a well-programmed task, 9,t where tibialis anterior (TA) activity is responsible for the decoupling of the center of pressure and center of mass lltz such that the center of mass moves forward over the stance foot. Gastrocnemius activity serves to control this forward acceleration of the center of mass. Normal gait velocity is reached by the end of the first step and is approximately 1.2 m/s. t3 Because it is well documented that the interaction of the ankle musculature is abnormal in persons with stroke during gait TM and after postural perturbation, t5 gait initiation is a task that may help develop a better understanding of the relation between limb loading, muscle activity, and the dynamics of motor control in this patient population. The purpose of this study was twofold: (1) to explore the relation between asymmetry in limb loading of persons with stroke and their ability to generate the appropriate forces to initiate gait; (2) to describe the relation between EMG activity of the involved and noninvolved limb and the resultant ground reaction forces. It was predicted that asymmetrical limb loading during stance would be related to patient-generated forces during gait initiation and that these forces would be related to the interaction of the amplitude and timing of anterior and posterior ankle muscle activity. METHODS Subjects Thirteen patients with hemiplegia secondary to a cerebral vascular accident participated in the study. One patient had

628 STROKE GAIT INITIATION, Brunt Table 1: Patient Characteristics Mean Involved Limb Patient Age* Sex Duration * Limb Loading* 1 81 M 5 R 52 2 42 F 21 R 36 3 76 M 7 L 42 4 47 M 6 R 43 5 68 M 5 L 43 6 80 M 4 R 45 7 81 F 15 R 43 8 42 F 23 R 29 9 60 F 13 R 18 10 78 M 21 R 27 11 76 F 4 L 24 12 77 M 11 L 32 13 78 F 8 R 21 * Patients age in years. t Time since stroke in weeks. * Mean limb loading (% BW) on the involved limb during stance before gait initiation. suffered a previous ipsilateral stroke. All patients were capable of following simple instructions and walking at least 5m at their preferred speed with guarded assistance. Patients were asked to walk barefoot. No patients were diagnosed as having a subcortical or brain stem lesion. Patient characteristics are listed in table 1. Equipment Surface electrodes were applied bilaterally to the center of the muscle belly of the medial gastrocnemius (G) and TA muscles. Each recording electrode a consisted of two silversilver chloride 1-cm diameter electrodes embedded in an epoxy-mounted preamplifier system (x35) and whose centers were spaced 2cm apart. A ground electrode was attached to the medial aspect of the lower left limb. Electric switches were built into mats b that were strategically placed to monitor foot-ground contact. The EMG signal was high-pass filtered (20Hz), full-wave rectified, and then root mean square processed with a 2.5-ms time constant. The amplifier gain was set at either 5k or 10k to reproduce a signal suitable for visual interpretation. 2 A level platform (1.22m wide and 10m in length) served as a walkway during data collection. Incorporated into the walkway was a force platform c that simultaneously measured three force components along the x, y, and z axis. Processed EMG, foot switch, and amplified force platform signals were sampled on-line at a rate of 1,000Hz for 4 seconds, d A light signal cued the patient to begin walking while simultaneously initiating data sampling. Procedures Patients were first asked to stand with their involved limb on the force platform and were instructed to begin walking at the onset of the light signal. The limb that first entered swing phase (swing limb) was always the involved limb. Rarely did patients begin walking with their uninvolved limb, and if so, the trial was repeated. Patients completed a total of six satisfactory trials in which the first three trials began with their involved limb on the force platform and the last three trials with their uninvolved limb on the force platform. Data Analysis For each trial, the baseline vertical force (Fz) before gait initiation was determined. Based on these values, patients were arbitrarily placed into a symmetrical limb loading group (SLL, patients 1-7) or an asymmetrical limb loading group (ALL, patients 8-13) (table 2). For further analysis, force platform data for peak Fz for both the stance and swing limb (see the results for an explanation of Sw Pkl and St Pkl to St Pk 3) and the peak fore/aft ground reaction force (Fx) for the swing limb are reported. Analysis of Variance (ANOVA) was used to determine group (SLL vs ALL) differences for the above peak ground reaction forces. Mean trial data were obtained for the medial/lateral (Fy) ground reaction force. Correlation coefficients were determined between initial limb loading and Fz and Fx peak ground reaction forces. All ground reaction forces were expressed as a percent of the patients' body weight (% BW). Also noted was the presence of TA activity in the swing and stance limb occurring with the onset of force plate activity and the presence of stance G activity as it served to control forward progression. RESULTS Normal Gait Initiation Figure 1A and B shows the typical Fx, Fy, and Fz ground reaction forces during gait initiation for healthy individuals. The sequence of events for gait initiation have been fully explained elsewhere 9tl and will receive brief mention here. The ground reaction forces generated by the limb that first enters swing phase (swing limb) are shown in figure 1A and that of the contralateral limb (stance limb) in figure lb. The Fy and Fx ground reaction forces have the effect of propelling the center of mass towards the stance limb and forwards. For this to occur there must be an increase in Fy of the swing limb, a simultaneous decrease in Fy for the stance limb, and accelerating forces (positive Fx) generated by both the stance and swing limb. 9 There is a resultant loading (increase in Fz) of the swing limb (Sw Pkl) and an unloading of the stance limb (St Pkl). The timing of limb loading and unloading is very symmetrical in healthy subjects with the swing limb loading to approximately 63% BW. After toeoff of the swing limb, the Fz of the stance limb resembles that of the stance phase of normal gait with peaks of 105 and 116% BW (St Pk2 and St Pk3). Gait initiation ends with toe-off of the stance limb. Bilateral TA activity precedes the onset of force platform activity and functions to unload the four feet. Small bursts of G activity may occur before toeoff of both the swing and stance limb that in the stance limb functions to control forward momentum. Initial Limb Loading Table 2 shows the patient's distribution of BW (Fz) for individual trials attributed to the involved and noninvolved limb before and during gait initiation. The data was used before gait initiation to arbitrarily group the patients according to symmetry of limb loading. Before gait initiation, patients 1 to 7 showed relatively symmetrical limb loading (SLL group) when compared with patients 8 to 13 (ALL

STROKE GAIT INITIATION, Brunt 629 Table 2: Patient Limb Loading Data (% BW) for Quiet Standing and During Gait Initiation Limb Loading* Peak Vertical Ground Reaction Force * Patient Inv Noninv Sw PK1 St Pkl St Pk2 St Pk3 Peak F/A* 6 48, 51, 56 42, 56, 56 68, 62, 75 24, 39, 43 99 97 11, 15, 11 2 33, 37, 39 40, 57, 60 51, 50, 59 14, 26, 39 99 107 02, 01, 06 3 34, 45, 46 51, 51, 59 43, 58, 61 31, 35, 44 101 103 02, 07, 09 4 32, 43, 55 39, 40, 51 46, 65, 61 23, 17, 37 97 98-6, 03, 00 5 39, 41, 50 49, 61, 70 51, 59, 71 26, 53, 47 102 97 05, 03, 11 6 44, 44, 48 40, 45, 54 62, 64, 69 18, 18, 31 101 100 01, 01, 11 7 41, 43, 45 52, 54, 57 52, 59, 62 29, 23, 43 99 101-2, 01, 00 8 27, 28, 31 66, 72, 72 32, 34, 38 64, 64, 63 99 97-4, -6, -1 9 16, 18, 19 68, 70, 75 22, 19, 30 62, 60, 62 101 103-3, -4, -11 10 24, 26, 31 76, 76, 82 34, 30, 42 65, 67, 73 103 97-1, -5, -7 11 14, 24, 35 73, 80, 89 15, 30, 41 65, 78, 88 t02 101-10, -3, -2 12 26, 31, 42 57, 61, 66 51, 37, 51 48, 51, 51 102 100-7, -6, -4 13 18, 20, 24 78, 78, 79 25, 29, 34 63, 67, 64 102 102-7, -77-8 Total Mean s 35 62 47 47 101 100 -.03 Patient 1-7 44 52 59 31 100 100 4.39 Patient 8-13 25 73 32 64 102 100-5.17 * Individual trial data for baseline vertical ground reaction force before the initiation of gait for the involved (lnv) and noninvolved (Noninv) limb. t See text and figure 1 for explanation of Sw P and St P1 to St P3. Individual trial data are shown for Sw P and St P1. For St P2 and St P3, only 6 of 78 trials were outside the 95% to 105% BW range. Individual trial data for the fore/aft ground reaction force for the swing limb. Mean data for all patients that are further divided into mean data for patients 1 to 7 who showed more symmetrical limb loading and patients 8 to 13 who showed more asymmetry in limb loading before gait initiation. group). For the SLL group, their mean weight bearing during stance before gait initiation was 44% BW on the involved limb compared with 25% BW for the ALL group. The AN- OVA indicated a significant group difference for both the involved (F = 41.11, df[ 1,11 ], p <.001) and noninvolved (F = 36.62, df[1,11], p <.001) limb for initial limb loading before gait initiation. As only one force platform was used in this study then the intertrial variability of initial limb loading would snggest that loading conditions were not identical for the testing of both the involved and noninvolved limb. However, an ANOVA indicated no significant difference (p >.05) between loading of the involved limb (mean of 35% BW, table 2) and the inferred loading of the noninvolved limb (100% BW minus mean of 62% BW = mean of 38% BW for the noninvolved limb, table 2). Ground Reaction Forces The SLL group clearly showed an initial loading of the swing limb and an unloading of the stance limb that approximated those proportions observed in healthy elderly subjects. ~3 That is, the swing or involved limb was loaded more than 50% BW before toe-off at Sw PK1. The ALL group also showed a pattern of initial swing limb loading, but this was less than 50% BW. Sw PK1 was significantly greater for SLL (F = 51.87, df[1,11], p <.001). Simultaneous with loading of the swing involved limb both groups unloaded the stance limb. The ALL group unloaded their stance limb to a lesser extent than the SLL group (F = 54.84, df[1,11], p <.001). Once the swing limb was unloaded (in swing phase), similar recordings of Fz for St Pk2 and St Pk3 were noted for all subjects. There were no significant group differences for these stance peaks (p >.05). Individual patient mean data from table 2 are shown in figure 2 where the limb loading of the involved limb (SW PK1) is plotted. All patients clearly loaded the involved limb (swing limb) before toe-off. The increase in the amount loaded was dependent on initial loading of the swing limb before gait initiation. That is, the increase in limb loading was greater for the SLL group than the ALL group. Figure 3 shows similar data for the uninvolved stance limb in which the initial unloading of the stance limb (ST PK1) was proportionally greater for the SLL group. This relationship of the amount of swing limb loading and the amount of stance limb unloading to initial limb loading before gait initiation is reflected in high correlations of.95 (fig 4A and B). Table 2 also shows a clear difference between the SLL group and the ALL group for the direction of Fx. That is, for the SLL group, the mean Fx for the swing limb was 4.39% BW (forward ground reaction force) compared with -5.17% BW for the ALL group. This difference was significant (F = 26.33, df[1,11], p <.001). Figure 5 shows the relationship for all trials between the amount of initial limb loading before gait initiation and swing limb peak Fx (r =.80). Both groups of patients showed an appropriate increase in swing limb Fy. The mean peak Fy for the SLL group was 14% BW (range of patient means 9 to 21) and 6% BW for the ALL (range of patient means 0 to 10). The group difference was significant (F = 11.76, df[1,11], p <.01). Force Platform Data and EMG--The Involved Swing Limb Figure 6A and B shows the force platform and EMG data of patient 1. In figure 6A, the swing limb force platform data before gait initiation shows a backward ground reaction force (negative Fx), an adductor torque (negative Fy), and a vertical ground reaction force at approximately 50% BW. After movement onset, a forward ground reaction force is generated reflected by a positive Fx, Fy increases, and Fz increases to approximately 65% BW. The resultant ground reaction forces have the effect of propelling the center of mass of the patient in a forwards direction and towards the stance limb. In comparison, for patient 13 (fig 7A), a positive

630 STROKE GAIT INITIATION, Brunt Fx of the swing limb is never attained and, although the 70 swing limb is loaded, peak Fz is only 30% BW. However, similar to patient 1, the adductor torque increased (negative.~ Fy). During gait initiation bilateral TA activity is responsible.~ 60 for the backwards movement of the center of pressure and then pulling the lower limbs over the base of support. From ~: 50 the comparisons of the EMG for the two patients, the absence of swing limb TA (involved right limb) at movement onset for patient 13 (see figs 7A and B) is noted. TA is not active d~ 40 until the limb is being unloaded in preparation for toe-off A F~t it, ~, 10% IIW 15% IIW aa 30 t_. ~a 20 10 I,A s qs# ~o o Limb Loading - -..e,--- Swing Peak l I I I I I I I 1 2 3 4 5 6 7 8 Subject It II I I I I I I I I 0 10111213 Fig 2--Loading characteristics of the swing limb. Individual patient data showing the relationship between peak swing limb loading (Sw PK) during gait initiation and swing limb loading before gait initiation. F,, 209 ms 40% U~V and the swing phase. Gastrocnemius activity in the involved swing limb is also tonic creating some cocontraction at the ankle joint prior to toe-off that will provide a plantar flexion torque that opposes the force generated by the TA. For the ALL group four patients (8, 11, 12, 13) showed G activity in the absence of TA activity at the beginning of gait initiation. In figure 6 (patient 1), there is bilateral TA activity before movement onset that will contribute to forward progression of the center of mass. Only patients 1, 3, 5, and 6 showed consistent swing limb TA activity before toe-off that preceded force platform activity (18 of the 24 trials). Fx and swing limb TA activity also is shown in figure 8 for patients it- #y Fz 14% DW 204.,J Fig 1--(A) Healthy subject swing limb force platform data. Channels from top to bottom are fore-aft (Fx), horizontal (Fy), and vertical (Fz) ground reaction forces. The vertical lines from left to right are movement onset and swing toe-off. (B) Healthy subject stance Hmb force platform data. Channels are the same as for Fig 1A. The vertical lines from left to right are movement onset, swing toe-off, swing heel-strike, and stance toe-off, ie, the end of gait initiation.,j= om ~a >., 0 == ~a L. 1 =., 9O 80 - - -e- - 70 60 50 40 30 20 Limb Loading Stance Peak i I ~ 4, I I I I I I I I 1 2 3 4 5 6 7 8 Subject t I tl I I I I I 9 10111213 Fig 3--Unloading characteristics of the stance limb. Individual patient data showing the relationship between peak stance limb unloading (St PK1)during gait initiation and stance limb loading before gait initiation.

STROKE GAIT INITIATION, Brunt 631,80 t rloo r-=.95 r:.95 m 1.~ 60 ~- ~. 40 B ~ B IBB n= 20 o 10 20 30 40 50 60 0, I I i I, 3( 40 50 60 70 80 90 Initial Swing limb Loading (% BW) Initial Stance Limb Loading (% BW) Fig 4mCorrelations between initial limb loading and peak vertical ground reaction forces (Fz) before swing toe-off. A. Swing limb peak Fz (Sw PK) to swing limb loading before gait initiation and B. Stance limb peak Fz (St PK1) to stance limb unloading before gait initiation. 5 and 12 for three trials. For patient 5, the onset of TA activity is before force platform activity (vertical line), and a forward ground reaction force is generated (positive Fx). Whereas for patient 12, TA activity occurs just before toeoff (zero Fx), and Fx remains negative. The negative component of Fx for patient 5 corresponds to swing limb G activity, excessive in this trial (not shown), and prevents the tibia from advancing over the fixed foot before toe-off. EMG and Force Platform Data-- The Noninvolved Stance Limb Stance limb fi3rce platform data and bilateral EMG data are shown in figures 6B and 7B. The onset of stance TA activity was normal for all patients and preceded force platform activity. In normal gait initiation, TA activity will continue until just before stance toe-off where there will be a short burst of G activity. Typically, the duration of G activity in the noninvolved stance limb was excessive for all patients. For example, note in figure 6B there is a cocontraction of G and TA throughout most of stance before heel strike of the swing limb and in figure 7B a predominance of G activity over TA activity. DISCUSSION Strong relationships were presented between initial swing fimb loading and peak medial/lateral (Fx) and peak vertical (Fz) ground reaction forces. That is, the greater the swing fimb was loaded before movement onset the more these forces approached those values observed in healthy adults. The high correlations imply proportional changes. This makes sense because the greater the stance limb is loaded before gait initiation (ALL group), then less force is required to shift the center of mass in the frontal plane in preparation for single limb stance. All patients therefore showed an increase in Fz and Fy of the involved limb that was proportional to initial limb loading. A good correlation (r =.80) was also reported between initial limb loading and Fx. In addition, a positive peak Fx was related to the presence of TA activity. These data indicate that increased loading on the involved limb may be coincident with more normal ankle flexor activity. It appears therefore that for those patients who demonstrate more asymmetry in limb loading (ALL group), the beginning of gait initiation may well be simply lifting the paretic limb off the ground, a process that is very similar to the leg lift task as described Rogers and colleagues. 8"16 With the leg lift task, the time of the peak medial/lateral ground reaction force (Fy) coincided with the peak vertical ground reaction force (Fz). This was the case for all of their subjects regardless of the amount of force, although some timing differences with some subjects were reported. Before onset of gait initiation, both the swing and stance limb apply small adductor torques (Fy). During gait initiation Fy increases in the swing limb and decreases in the stance limb. 1~ This same pattern for Fy was found in the present study and is demonstrated in figures 6 and 7. However, the leg lift task employed by Rogers 8"~6 involved movement in only the frontal plane, whereas with gait initiation forward momentum also must be generated. When monitoring the swing limb only in subjects 1 through 7 (SLL group) was a positive Fx noted that would contribute to a movement of the center of mass in a forward direction, whereas those patients in the ALL group showed a negative Fx that is attributed to a plantar flexion torque. An example of the relationship between negative Fx and a plantar flexion moment is demonstrated in figure 6A. In this trial, the timing of swing G activity coincides with

632 STROKE GAIT INITIATION, Brunt 2O ~I0 the stance limb (uninvolved limb) for both the ALL and SLL groups did generate a forward ground reaction force that is consistent with expected normal data (figs 6B r =.79. and 7B). In healthy individuals, it has been shown that the center of pressure under both the swing and stance limb does in B B ~ fact move backwards before swing toe-off in gait initiation, A I 10% llw IN)..~ lo, c~ [] ~/" [] [] 14 %! y. " Fz I 30% i] r B B Swing G I -20 I " I " I ' I 0 2O 3O 40 50 6O Initial Swing Limb Loading (% BW) Fig 5--Correlation between initial swing limb loading and peak fore-aft ground reaction force (Fx) before swing toe-off. a decrease in Fx (ie, the return of Fx towards baseline). Although G activity before swing toe-off during gait initiation is normal in healthy subjects, 9 the amplitude may be excessive in this instance. The same effect of G activity on Fx is also demonstrated in figure 8 for three consecutive trials for patient 5. The inability of the ALL group to generate a dorsiflexion torque to initiate gait cannot be attributed to solely G activity in the absence of TA activity as two patients 9'1 demonstrated a negative Fx with no apparent G activity. Negative Fx could also be a result of posterior tightness or an extension synergy. Unlike the swing limb, B I,'x Fy Fz,,... [.l' nlv 2. ms I] l ~J "/~--~,I'~':".w_.._..-~---"~ I% Bw ~ \. j \ \ Fig 6--EMG and force platform data for single trials for patient 1. (A) Swing (involved) limb. Channels from top to bottom are fore-aft (Fx), horizontal (Fy) and vertical ground reaction forces (Fz), swing limb tibialis anterior (TA) and gastrocnemius (G) activity, and stance limb tibialis anterior and gastrocnemius activity. The vertical lines from left to right are time to movement onset, peak vertical ground reaction force, and swing toe-off. (B) Stance (noninvolved limb). Channels are the same as for figure 6A. The vertical lines from left to right are time to movement onset, minimum peak vertical ground reaction force, swing limb toe-off, swing limb heel-strike, and stance toe-off. Swing TA Swing G Slance TA Stance G nlv

STROKE GAIT INITIATION, Brunt 633 A Fx 6, BW / ~ ~_.~.~,/ 6 o II w /" '7 B Fx I 30% Fy Fy Fz 20 $ BW Fz.15 my Swing TA..... Swing TA.i my Swing G Stance TA I J'" 2oo,ns ~,N~,~ Swing G Stance TA 250 ms i[ ].3 my Stance G Stance G Fig 7mEMG and force platform data for single trials for patient 13. (A) Swing (involved) limb. Channels and vertical lines are the same as in figure 6A. (B) Stance (noninvolved) limb. Channels and vertical lines are the same as for figure 6B. indicating that both limbs should contribute to forward momentum. 9'm The interaction of the TA and G indicates that the latter muscle acts to decelerate or control forward motion of the tibia over the fixed foot (fig 8, patient 5), whereas the TA is responsible for forward motion and presumably the backwards movement of the center of pressure. In the present study, two apparent characteristics in phasic EMG activity of persons with stroke during gait initiation can be reported. Subject 5 Subject 12 Fig 8mlndividual trial tibialis anterior (TA) and fore-aft force (Fx) data for the swing (noninvolved) limb for patients 5 and 12. First, the direction of Fx was clearly related to the TA muscle activity before force platform activity. Decreased TA muscle activity clearly resulted in a decrease in the contribution of the swing limb to forward momentum before toe-off of that limb. Crenna 17 has recently reported a similar strong relationship between the amount of backwards movement of the center of pressure and TA activity. Second, there was often an increased G activity of the noninvolved stance limb that at times resulted in a TA and G cocontraction. The increased G activity in the stance limb would tend to control or slow forward momentum until the swing, or paretic limb, had safely completed swing phase and was ready to support partial body weight. CONCLUSION The inability of persons with stroke to generate forces in the swing (involved) limb that contribute to the forward progression of the center of mass appear to be related to the absence of TA activity. Gait initiation in these patients therefore involves lifting that leg off the ground, whereas the forward momentum forces must be generated solely by the noninvolved stance limb. The pattern of the vertical (Fz) and medial-lateral (Fy) ground reaction forces are similar to that reported for persons with stroke during a simple leg flexion task. 8 The fore-aft (Fx) ground reaction force of those patients with more symmetrical limb loading was positive and contributed to the forward momentum of the center of mass and is consistent with increased TA activity. An increase in the noninvolved stance limb G activity was noted in both groups of patients that would presumeably control forward momentum in preparation for swing limb heel strike. This report provides evidence as to the relationship of limb

634 STROKE GAIT INITIATION, Brunt loading ability to gait initiation. It appears that therapy aimed towards symmetrical limb loading must be task specific. This has been shown during balance retraining where improvement has not transferred to functional tasks such as gait. ~8'19 It seems important therefore that symmetry in limb loading during quiet stance, or limb loading characteristics that are task specific (as in gait initiation), should be an integral component of the rehabilitation of persons with stroke. References 1. Dietz V, Berger W. Spinal coordination of bilateral leg muscle activity during balancing. Exp Brain Res 1982;47:172-6. 2. Brunt D, Anderson JC, Huntsman B, Reinhert LB, ThoreU AC, Sterling JC: Postural responses to lateral perturbation in healthy subjects and ankle sprain patients. Med Sci Sports Ex 1992;24:171-6. 3. Nashner LM. Balance adjustments of humans perturbed while walking. J Neurophysiol 1980;44:650-64. 4. Diener HC, Horak FB, Nashner LM. Influence of stimulus parameters on postural responses. J Neurophysiol 1988;59:1888-1905. 5. Wall CJ, Turnbull GI: Gait asymmetries in residual hemiplegia. Arch Phys Med Rehabil 1986;67:550-3. 6. Dettmann MA, Linder MT, Sepic SB: Relationships among walking performance, postural stability, and functional assessments of the hemiplegic patient. Am J Phys Med 1987;66:77-90. 7. Car~ JH, Shepherd RB, Gordan J, Gentile AM, Held MJ. Movement science: foundations for physical therapy in rehabilitation. Rockville, MD: Aspen, 1987. 8. Rogers MW, Hedman LD, Pai YC. Kinetic analysis of dynamic transitions in stance support accompanying voluntary leg flexion movements in hemiparetic adults. Arch Phys Med Rehabil 1993; 74:19-25. 9. Brunt D, Lafferty MJ, Mulhausen C, Mckeon A, Goode B, Polk P. Invariant characteristics of gait initiation. Am J Phys Med Rehabil 1991;70:206-12. 10. Mann RA, Hagy JL, White V, Liddell D. The initiation of gait. J Bone Joint Surg [Am] 1979;61:232-9. 11. Nissan M, Whittle MW. Initiation of gait in normal subjects: A preliminary study. J Biomed Eng 1990; 12:165-71. 12. Breniere Y, Do MC. When and how does steady gait movement induced from uptight posture begin. J Biomech 1986; 19:1035-40. 13. Breniere Y, Do MC, Bouisset S. Are dynamic phenomena prior to stepping essential to walking. J Motor Behav 1987; 19:62-76. 14. Dimitrijevic MR, Faganel J, Sherwood AM, McKay WB. Activation of paralyzed leg flexors and extensors during gait in patients after stroke. Scand J Rehabil Med 1981; 13:109-15. 15. Dietz V, Berger W. Interlimb coordination of posture in patients with spastic paresis: impaired function of spinal reflexes. Brain 1984; 107:965-78. 16. Rogers MW, Pai YC. Dynamic transitions in stance support accompanying leg flexion movements in man. Exp Brain Res 1990; 81:398-402. 17. Crenna P, Frigo C. A motor programme for the initiation of forwardoriented movements in humans. J Physiol 1991;437:635-53. 18. Hocherman S, Dickstein R, Pillar T. Platform training and postural stability in hemiplegia. Arch Phys Med Rehabil 1984;65:588-92. 19. Winstein CJ, Gardner ER, McNeal DR, Barto PS, Nicholson DE. Standing balance training: Effect on balance and locomotion in hemiparetic adults. Arch Phys Med Rehabil 1989;70:755-62. Suppliers a. Therapeutics Unlimited, 2835 Friendship Street, Iowa City, IA 55240. b. Lafayette Instruments, 3700 Sagamore Parkway N, Lafayette, Indiana 47903. c. AMTI, Incorporated, 141 California Street, Newton, MA 02158. d. BIOPAC Systems, INC., 275 South Orange Avenue, Ste E., Goleta CA 93117.