Introduction

The gait control process begins when appropriate afferent information is filtered from different sources. This information determines the organization of muscle synergies. Gait movements are then moderated to actual needs1 through the interaction of this information with central programs. Different combinations of afferent inputs as well as the subsequent interaction of reflex mechanisms determine the correct programmed pattern2 for locomotion. Correct pattern formation may also be determined by the instruction for a particular locomotion condition.1,2

A central pattern generator may drive the rhythmic lower extremity movement patterns shown during gait. This central generator is made up of a network of neurons intrinsic to the lumbosacral spinal cord,3,4,5 and are regulated by supraspinal motor centers in the brain.6,7,8 Trauma to the spinal cord and interruption of the spinal interneuronal circuits connecting the brainstem and the supraspinal motor center interfere with several aspects of normal gait.1,2,6 Owing to resultant physical barriers, such as spasticity, muscle weakness, and abnormal proprioception, spinal patients are unable to bear their weight sufficiently on the affected lower limbs at the weight-bearing stance phase of the gait cycle, which leads to loss of balance and falls.1,2,6,9,10 In recent studies,6,8,10,11,12,13 it has been demonstrated that central pattern generator can be triggered in patients with complete or incomplete spinal cord injury (SCI) when the body is partially unloaded. Evidences include increased stride length and relative time span of the single-stance phase, and a reduction in the relative time span of the double-stance phase during the gait cycle.

Owing to their reduced reserves to support balance and gait, patients with incomplete SCI might adopt varying temporal distance strategies to achieve a safer, more stable gait pattern for an efficient control of dynamic balance during walking.6,10

Gait initiation challenges dynamic balance control because it is the transitional phase between static balance in an upright position and the start of steady-state walking. The analysis of temporal-distance strategies during gait initiation has provided important information about the coordination of posture and intentional movement in normal adults.14,15,16 It has been found that the first step length, reaction time, the relative time span of single- and double-stance phases, propulsive forces, gait initiation speed, covariation of muscle activities, and sensory feedback are all vital components of generation of the appropriate gait initiation pattern.9,16,17,18,19,20,21,22,23,24, 25

Problems with gait initiation performance may exaggerate locomotive deficiencies in the elderly26,27,28 as well as neurologically impaired subjects, such as those suffering from Parkinson's disease29,30,31 and stroke32,33,34 and could be a sensitive indicator of balance dysfunction.7,26,27,32,35,36 No studies investigate gait initiation deficiencies in the SCI population. Also, patients with lumbosacral spinal cord lesion may initiate gait differently from patients with high-level SCI. The differences may not only come from the varied areas of physical barriers between different levels of SCI but also from the influence of possible damaged central pattern generator3,4,5,8 in patients with lumbosacral spinal cord lesion.

The purposes of this study are: (1) to compare temporal-distance strategies adopted during gait initiation among three groups: patients with cervical SCI, patients with lumbosacral injury, and unimpaired control adults; (2) to identify modification of temporal-distance strategies adopted during gait initiation in order to help us determine how the multivariable control is adjusted to impairments caused by SCI; and (3) to understand the modifications that SCI patients adopt to reduce the risk of falls. Gaining this understanding may thus enable further improvements to be made in rehabilitation design by working toward specific goals common to these patients.

Methods

Subjects

In all, 19 subjects were recruited for this study, including five patients with cervical SCI (Group 1, aged 20–47 years), five patients with lumbosacral spinal lesions (Group 2, aged 31–59 years), and nine unimpaired control adults (Group 3, aged 20–25 years). Sex, age, height, weight, body mass index (BMI), leg length, years after injury (chronicity), preferred leg, lower extremity ASIA motor, and sensory scores37 were recorded for each subject (Table 1). Inclusion criteria for Groups 1 and 2 were: (1) class D of the ASIA impairment scale,37 and a score of less than or equal to 1 in the lower extremity muscles as measured by the modified Ashworth scale was required for Group 1;38 (2) a history of traumatic SCIS of more than 6 months; (3) the ability to walk more than 10 steps without any assistance or devices. Exclusion criteria for all three groups included any significant comorbid disease that would interfere with ambulation function. The experimental procedures were approved by the local institutional review board for human research and adhered to the Occupational Health and Safety administration regulations. Each subject gave their informed consent.

Table 1 Characteristics of Groups 1–3
Full size table

Instruments

A computerized, high-resolution three-dimensional motion analysis system (Vicon 370 suppliers: Vicon 370, Vicon motion system, 14, Minus Business Park, West Way, Oxford, OX2OJB, UK – six cameras, 60 Hz) was employed to obtain the temporal-distance parameters and the kinematic data during gait initiation in all subjects.

Procedures

Subjects stood barefoot, with 13 reflective markers placed on the sacrum and both sides of lower limbs at the anterior superior iliac spine (ASIS), lateral thigh, knee, lateral tibia, lateral malleolus, and the second metatarsal head. For the first trial, each subject was asked to step with their preferred leg first at a self-selected speed in response to an auditory cue. For the second trial, the SCI patients (Groups 1 and 2) were then asked to step with the nonpreferred leg first. Leg preference was defined as the leg chosen to kick an imaginary ball.39,40 This was repeated twice with both trials yielding the same results.41 Data collection began with the participants standing steady for at least 5 s prior to applying the auditory cue. Subjects were allowed several practices before three formal data collection trials.

Data analysis

The total gait initiation period (TGIP) was defined as the interval from the start of gait initiation as the auditory cue began, to heel-strike of the first swing leg. The TGIP was divided into three subphases: Phase I (preparatory phase, start of the auditory cue to heel-off of the first swing leg) – the span of Phase I was recognized as the reaction time of the gait initiation;24 Phase II (push-up phase, heel-off to toe-off of the first swing leg); and Phase III (single-stance phase, toe-off to heel-strike of the first swing leg). TGIP and Phases I, II, and III were identified by using the kinematic measurement system. Phases I, II, and III were normalized by the TGIP as relative phasing of gait initiation of individual subjects.25 The first step length during TGIP was defined as the distance between the heels of the first stance leg and the first swing leg at the moment when the first swing leg contacted the ground. The first step length was also normalized by the lower-extremity length of each subject to exclude individual differences due to various lower-extremity lengths.

According to Table 1, leg preference is not related to lower-extremity ASIA motor and sensory scores across all three groups. Problems in motor control may be indicated by leg preference, with the preferred leg tending to be better than the nonpreferred leg in functional use.40,41,42 Following neurological injury, leg sensitivity is quite considerable and leg preference is therefore very important for clinical rehabilitation purposes.40 Therefore, in Groups 1 and 2, both the preferred leg and the nonpreferred leg were selected separately as the first swing leg, since the SCI pathology might be functional asymmetrically implicated in both legs,41,42 and thus interfere with the entire gait initiation process. Only the preferred leg was used as the first swing leg in Group 3 given that unimpaired control adults initiate gait symmetrically, no matter which leg starts first.9,32,40,42,43

The Kruskal–Wallis one-way ANOVA by ranks test (K–W ANOVA) was carried out to compare the differences among the three groups with regard to TGIP, phase duration, and the first step length. A critical α-level of P<0.05 was used throughout the study.

Results

The median and interquartile range (IQR) of the temporal distance variables with the preferred leg stepping first, and with the nonpreferred leg stepping first, is shown in Figures 1, 2, 3, 4, 5 and 6. The TGIP with the preferred leg leading was significantly longer in Groups 1 and 2 than in Group 3 (P<0.05; Figure 1a). TGIP was 24.41% greater in Group 1 than in Group 2, 41.61% greater in Group 1 than in Group 3, and 13.83% greater in Group 2 than in Group 3. The TGIP with the nonpreferred leg leading was markedly more delayed in Groups 1 and 2 than in Group 3 (P<0.05; Figure 1b). TGIP was 38.22% greater in Group 1 than in Group 2, 59.28% greater in Group 1 than in Group 3, and 15.24% greater in Group 2 than in Group 3 (Figure 1b). The reaction times (Figure 2) and the durations of Phase I (Figure 3), Phase II (Figure 4), and Phase III (Figure 5) did not reach significance (P>0.05), no matter which leg stepped first. There was no significant difference in the first step length, with the preferred leg stepping first among the groups (P>0.05; Figure 6a). However, the IQR value of the first step length in Group 2 (16.183% leg length) was higher than those in Group 1 (11.679% leg length) and Group 3 (5.119% leg length; Figure 6a). The first step length with the nonpreferred leg leading was significantly greater in Group 3 than in Groups 1 and 2 (P<0.05; Figure 6b). The first step length was 36.65% greater in Group 1 than in Group 2, 10.74% greater in Group 3 than in Group 1, and 51.34% greater in Group 3 than in Group 2. The IQR value of the first step length in Group 2 (32.776% leg length) was higher than those in Group 1 (9.123% leg length) and Group 3 (5.119% leg length; Figure 6b).

Figure 1
figure 1

Median (Q2) and IQR (the first quartile Q1 to the third quartile Q3) of total gait initiation period (TGIP) while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all three groups. The TGIP was the interval from the outset of gait initiation as the auditory cue began to the heel-strike of the first swing leg. TGIP unit: second. Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults. *P-value shows significant statistical differences across the three groups (P<0.05)

Full size image
Figure 2
figure 2

Median (Q2) and IQR (the first quartile Q1 to the third quartile Q3) of reaction time while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all 3 groups. Reaction time is the time interval between the auditory cue and the heel-off of the first swing leg. Reaction time unit: second. Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults

Full size image
Figure 3
figure 3

Median (Q2) and IQR (the first quartile Q1 to the third quartile Q3) of Phase I while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all three groups. Phase I is the preparatory phase, one of the relative phasings of gait initiation. Phase I is the duration from the start of gait initiation as the auditory cue began to heel-off of the first swing leg normalized by TGIP. Phase I unit: percentage of TGIP (%TGIP). Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults

Full size image
Figure 4
figure 4

Median (Q2) and IQR (the first quartile Q1 to the third quartile Q3) of Phase II while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all three groups. Phase II is the push-up phase, one of the relative phasings of gait initiation. Phase II is the duration from heel-off to toe-off of the first swing leg normalized by TGIP. Phase II unit: percentage of TGIP (%TGIP). Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults

Full size image
Figure 5
figure 5

Median (Q2) and IQR (the first quartile Q1to the third quartile Q3) of Phase III while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all three groups. Phase III is the single-stance phase, one of the relative phasings of gait initiation. Phase III is the duration from toe-off to heel-strike of the first swing leg normalized by TGIP. Phase III unit: percentage of TGIP (%TGIP). Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults

Full size image
Figure 6
figure 6

Median (Q2) and IQR (the first quartile Q1 to the third quartile Q3) of the first step length while the preferred leg steps first (a) and while the nonpreferred leg steps first (b) in all three groups. The first step length is the distance between the heels of the first stance leg and the first swing leg at the moment when the first swing leg contacted the ground normalized by the lower-extremity length. The First Step Length unit: Percentage of the lower extremity length (%LE length). Group 1: patients with cervical SCI; Group 2: patients with lumbosacral spinal lesion; Group 3: normal adults. *P-value shows significant statistical differences across the three groups (P<0.05)

Full size image

Discussion

Total gait initiation period (TGIP)

Our results showed that patients with incomplete SCI took more time in gait initiation than unimpaired control adults (Figure 1). The results were in line with the findings of various previous researches.9,14,29,30,32 Proprioceptive disorders, muscle weakness, joint stiffness, pain, and fear of falling are all possible contributors to this phenomenon.9,14,29,30,32 It revealed that impairment to the sensory and/or motor systems could delay the execution of gait initiation. Therefore, most of these neurogenic victims might adopt a less aggressive strategy for gait initiation. In addition, we found that lumbosacral spinal patients start their gait faster than cervical victims. Trunk instability, spasticity, and poor coordination of muscle activities may be related to the delay of gait initiation in high-level SCI patients.

Three subphases of gait initiation

The time-invariant central motor program includes characteristics such as invariant relative phasing limb movement and invariant submovement activation sequence.25 The central programming in gait initiation is time invariant in many neurogenic disorders and normal control subjects.25,26,29,30,39 This present study supports this conclusion. There was no statistical difference in the relative time span of the preparatory, push-up, and single-stance phases in gait initiation. Patients with cervical or lumbosacral SCI might execute a similar central motor programming for voluntary gait initiation, as did the unimpaired control adults.32,36,44 Although the segmental spinal cord control in the patients of the two study groups was impaired, the central motor programming of gait initiation seemed to be mostly preserved in the ambulatory ASIA-D SCI cases. The unchanged relative time span of the preparatory, push-up, and single-stance phases, regardless of which leg stepped first, provide evident that during gait initiation, lower extremity movement pattern is alternative and rhythmic. It suggests that a central pattern generator may be activated because of gait initiation in the ambulatory ASIA-D SCI cases. The central pattern generator may also be preserved in ambulatory patients with lumbosacral spinal cord lesions. As a result of these studies, it is therefore possible that the central pattern generator in human beings is not necessarily located in the lumbosacral spinal cord area.3,4,5

Reaction time

Reaction time in gait initiation is the time interval between the auditory cue and the heel-off of the first swing leg.24 In previous research, the reaction time in elders, children, and patients with Parkinson's disease has shown to increase significantly as compared with normal adults.17,28,29,44 A slow gait initiation response may be attributed to degenerative changes and immaturity or pathologies of CNS. Thus, gait initiation might carry a high risk of falling when rapid stepping is carried out to increase the base of support and avoid hazard events.28 However, the reaction time between the study and control groups did not reach significance, and our data suggest that SCI pathology in ambulatory ASIA-D SCI patients did not interfere with reaction time during gait initiation.

The first step length

Forward momentum generated by the first stance leg in the preparatory phase during gait initiation can determine the length of the first step.9,19 The first step length in the hemiparetic subjects was longer when they stepped with the affected leg first instead of the nonaffected leg. The affected leg of hemiparetic patients had less weight-bearing ability than the normal leg. In this study, when the nonpreferred leg stepped first, the step length was significantly shorter in the study groups than in the unimpaired control groups. The preferred leg, as the first stance leg in the study groups, may have less weight-bearing ability and generated less forward momentum as compared to unimpaired control adults. On the other hand, despite the trend that the IQR of the first step length in the unimpaired controlled group (62.934–68.053% leg length) was higher than those in the cervical SCI patients (45.284–56.963% leg length) and the lumbosacral SCI patients (47.372–63.555% leg length), the step length did not reach significance when the preferred leg stepped first (Figure 6a). The small case numbers in our study might account for the nonstatistical difference. However, both legs were affected due to SCI pathology, and leg preference might also lead to the discrepancy in the tolerance of body weight in term of dynamic stability.9,32 This conclusion needs further investigation on the ground reaction force of the first stance leg during gait initiation, especially in the preparatory phase, to find out if leg preference influences the tolerance of body weight in the SCI patient population.

A high IQR value of the first step length was also found in lumbosacral spinal cases (Figure 6a and b). From a clinical point of view, patients with lumbosacral implications had high variations in the muscle strength of both lower limbs. This might result in asymmetry in the propulsive force generated in the preparatory phase and foot clearance during the single-stance phase of gait initiation. Therefore, their first step length varied because of the need for compensated hip and knee movements.

Gait versus gait initiation analysis

The characteristics of gait patterns in neurogenic patients include shortened step length, a lengthened gait cycle, and a shortened relative time span of single-stance phase.10,29 The shortened relative time span of single-stance phase during the gait cycle is due to decreasing weight-bearing ability in the affected leg as the other leg takes a step forward. Prior studies have shown that gait training with partial body weight support may increase the step length and the relative time span of the single-stance phase in SCI patients with various levels implicated. In this study, we found that SCI and ASIS-D patients had lengthened gait initiation period, without a shortened relative time span of the single-stance phase of gait initiation despite which leg stepped first. However, a shortened first step length while the nonpreferred leg stepped first was noted. The results indicate that during gait initiation, the nonpreferred leg in our study subjects may have no difficulty in weight bearing in the single-stance phase, instead of a lengthened gait initiation period. A shortened first step length while the nonpreferred leg stepped first is, however, common in SCI and ASIS-D patients. The forward momentum generated by the first stance leg during the preparatory phase of gait initiation can determine the first step length.9,19 Therefore, these patients may have difficulty in weight bearing and generating forward momentum of the preferred leg in the preparatory phase. From this point of view, we suggest that relearning the preferred leg's weight-bearing ability, especially in the preparatory phase of gait initiation, should be added to ambulatory strategies. Thus, those who cannot tolerate a few steps of walking or who tend to be easily fatigued after walking a certain distance, might benefit from this task retraining.

Limitations of the study

Our patients who had ASIA-D SCI could walk independently more than 10 steps. The data in our study could not be applied to patients with SCI other than ASIA-D who rely on assistive devices to walk. Also, maximum hip and ankle angles and force measurements before toe-off of the swing limb during gait initiation were different between male and female subjects.14,45 In our study, most of the study group patients were male (M:F=9:1) and most of our normal control group adults were female (M:F=2:7). This may partly explain the high variations of task performance in our subjects.

Conclusion

All three unchanged relative phasing of the gait initiation process indicated that centrally programmed gait initiation was preserved in the SCI ASIA-D patients. This study also demonstrated dissimilarity in first step length, which may be due to asymmetry in spinal implications. Given that the SCI patients had difficulty in generating forward momentum in the preparatory phase, there should be a greater emphasis on relearning weight bearing in the preparatory phase of gait initiation, to enhance/strengthen walking ability after SCI.