Participants

Twenty-five healthy young males participated in this study. The average age, height, and mass were respectively 26.5±6.0 years, 1.79±0.10 m and 72.4±9.3 kg. No subjects had current back pain complaints or neurological disorders that could affect their gait initiation and trunk motion. The active range of motion was within normal limits [25] and the limb-length discrepancy, measured in the anatomical reference position between the anterior superior iliac spine and the medial malleoli, was less than 0.5 cm. The dominant lower limb, which was identified as the one the subject used to kick a ball [26], was the right limb, except for two of the subjects. Prior to the experimentation, all procedures were explained to each subject who signed a written consent. The experimental protocol was approved by the Ethics Committee of the University of Valenciennes.

Experimental protocol

A 14-body segment model was defined by 28 spherical reflective markers, each one with a 14-mm diameter. The International Society of Biomechanics joint coordinate standards [27] were applied. For each subject, these markers were placed on the acromions, elbows, wrists and the second and fifth knuckles to define the upper limbs. The other markers were placed on the greater trochanters, mid-thigh, tibial plateaus, mid-calf, lateral malleoli, heels and the second metatarsophalangeal joints to mark the lower limbs. The head markers were located on the tragi and the glabella, and a marker was placed on the 5^th lumbar vertebra (L5). The neck and trunk were considered as a single segment, using shoulders and hips markers.

Before initiating gait, the subjects were standing upright, looking forward, barefoot, with the upper limbs at each side of the body. Each foot was placed on a force plate (Kistler, Switzerland). The force plates were placed side-by-side and 5 mm away from each other at the beginning of a 5 m-long walkway. Ten cameras of a Vicon 612 motion capture system were located around the force plates so as to cover a calibration volume of 2 m in length, 2 m in height, and 1.5 m in width. The sampling frequency was 120 Hz for the video and the force plates.

The subjects faced a screen placed at the end of the walkway, which served to display the visual order of gait initiation [28], appearing as a green rectangle. This visual start signal was synchronized with the motion capture system in such a way to insure an appearing event was directly detectable from the video file. The visual start signal was displayed after a 5-second period of static standing upright. The precision of this appearing event was evaluated at 1/120^th of a second, which was used temporally for calculating the subjects' reaction time.

Before acquiring gait initiation data, the subjects performed practice trials to familiarize themselves with the experimental conditions. They were instructed to use a natural speed and to advance their preferred lower limb as soon as they saw the visual start signal. They had to go to the end of the walkway, looking forward, with their upper limbs at the side. The subjects performed ten trials, and only the subjects who always advanced spontaneously the right lower limb were included in this study (i.e., about 80%).

Average natural trunk inclination was measured between the appearance of the visual start signal and the right heel strike, marking the end of the gait initiation [9]. Trunk inclination was defined by the angle sustained by the line connecting the midpoint between the greater trochanters and the acromion with respect to the vertical. This option has been selected for two main reasons. First, this average trunk angle was preferred to the one measured in the orthostatic position because it better reflects the actual position of the trunk and its effect on net muscular moments during gait initiation. Second, the anatomical points that are used for defining the trunk segment are determinant for its orientation in the sagittal plane. For example, Thorstensson et al. [4] have defined the trunk segment with the 7^th cervical vertebrae and the 3^rd lumbar vertebrae and reported an overall range in trunk oscillations in the sagittal plane between 1.5° and 6° during gait. With this trunk modeling, none of their 10 subjects tilted their trunk behind the vertical. In addition, Goh et al. [29], who have defined the trunk segment with the acromion processes and the anterior superior iliac spines, observed a mean backward trunk inclination of 8.4° during walking, meaning that this trunk modeling implied that no subject maintained a forward trunk tilt with respect to the vertical. It appears then that the choice of trunk modeling is very important since, depending on that, the findings could be divergent. Furthermore, Leardini et al. [30] showed that there are different patterns and ranges of motion of the trunk movements in function of its modeling. Since our study seeks to examine the trunk inclinations around the vertical, none of the two previous trunk definitions are relevant. So our trunk modeling, using the hip markers and the acromions, was mainly motivated by functional considerations.

The trunk angle was averaged for each trial and for each subject. Afterwards, the subjects were divided into 2 groups according to their average natural trunk inclinations, either backward (BW) or forward (FW), with respect to the vertical [2]. The number of subjects in each group was determined so that a statistical power of 80% was reached [31]. The average natural trunk inclination was −1.9°±1.6° for the 13 BW subjects, whereas it was 2.6°±1.6° for the 12 FW subjects. Additionally, the subjects were retested at an interval of at least three days to ensure that their natural trunk inclinations were similar.

Data analysis

The center of mass (COM) was calculated from the video, using the anthropometric tables of Zatsiorsky-Seluyanov modified by De Leva [1]. In addition to antero-posterior instantaneous speed of the COM at the right heel strike (RHS), four other parameters are usually used to characterize gait initiation. They were calculated from the video and force plate data. The first of these parameters is the reaction time (or the beginning of the APA) that was determined by the first variation of the vertical component of the ground reaction force [28]. The three other parameters used to characterize the global kinematics of the gait initiation of our two groups are: the instant of first knee movement (1^st mvt) corresponding to the lower limb segmental movement that occurred first [32], the instant of right heel off (RHO) [33]–[35] that characterizes the beginning of the execution phase and the instant of right heel strike (RHS) that characterizes the end of the swing phase (Figure 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Illustration of the three kinematic events calculated from the appearance of visual signal (VS).

The stance lower limb (left) is traced in dashed lines; the swing lower limb (right) is traced in solid lines. RHO: right heel off; RTO: right toe off; RHS: right heel strike.

https://doi.org/10.1371/journal.pone.0055256.g001

For a better understanding of the net muscular moment curves, the right toe off (RTO) was also determined as a way to plot the beginning of the swing phase. In addition, the dependent spatio-temporal variables were also measured: the subjects' reaction times, calculated between the appearance of the visual start signal (VS) and the first variation of the vertical ground reaction force; the APA duration, calculated between the first variation of the vertical ground reaction force and RHO [34]; the gait initiation duration, calculated between the first variation of the vertical ground reaction force and RHS [9]; the amplitude of the backward displacement of the center of pressure (COP), calculated on the antero-posterior axis between the COP position at the first variation of the vertical ground reaction force and the maximum backward displacement of the COP.

The antero-posterior impulse was calculated with the antero-posterior component of the ground reaction force under the stance lower limb (left) [36]. The forces were normalized in terms of the subjects' mass and expressed between the appearance of the visual signal (VS) and the right heel strike (RHS). However, in order to not consider the subjects' reaction times, the impulse was calculated between the first variation of the vertical ground reaction force and the RHS (Figure 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Antero-posterior component of the ground reaction force under the stance limb (left) for the BW group (dashed line) and the FW group (solid line).

It was normalized temporally between the appearance of the visual signal (VS) and the right heel strike (RHS), and in amplitude in terms of the subjects' body mass. The hatched surface illustrates the antero-posterior impulse for the FW subjects, using the surface calculated under the curve between the first variation of the ground reaction force and the right heel strike. 1^st mvt: first knee movement (26% of the motor task); RHO: right heel off (55%); RTO: right toe off (68%). The average standard deviations for the BW and FW subjects are 0.1 N/kg. There is no significant difference between the two groups (p>0.05).

https://doi.org/10.1371/journal.pone.0055256.g002

The net muscular moments for each joint of the stance lower limb and the L5 vertebrae in the sagittal plane were calculated using inverse dynamics. The inverse dynamics is based on a bottom-up method, and the angular velocity and accelerations were calculated using quaternions [37]. The amplitudes of these net muscular moments were normalized with respect to the subjects' body mass. They were also normalized temporally between the appearance of the visual signal and the first right heel strike. The curves of the ankle, knee, hip and L5 vertebrae moments were characterized by their local maximum and minimum (Figure 3). The moment peaks were labeled with an alpha-numerical code, corresponding to a letter indicating the joint and a number that increases with the duration of gait initiation. These values were used for statistical analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Trunk angle and lower limb joints moments for the BW group (dashed line) and the FW group (solid line).

(a) Trunk angle, (b) L5 moment, (c) hip moment, (d) knee moment and (e) ankle moment, normalized in terms of the subjects' body mass, expressed in percentage of the gait initiation defined between the visual signal (VS) and the right heel strike (RHS). For the BW and FW groups, the average standard deviations are 0.1 Nm/kg for the ankle, knee, hip and L5 moments. The letters L, H, K & A respectively correspond to the L5 vertebrae, hip, knee, and ankle. 1^st mvt: first knee movement (26% of the motor task); RHO: right heel off (55%); RTO: right toe off (68%); DF: dorsal flexor. **: p<0.01; ***: p<0.001.

https://doi.org/10.1371/journal.pone.0055256.g003

Statistical analysis

In order to verify the homogeneity of the subject age and the anthropometric characteristics of the two groups, a one-way ANOVA was performed on these parameters. The significance threshold was set at 5%. No significant difference was observed for these parameters between the groups (p>0.05). A one-way ANOVA was performed to determine the way that the trunk's inclination affects 1) the instantaneous speed of the body's COM at the RHS, 2) the antero-posterior impulse of the stance limb, and 3) the stance limb's joint moments and the thoraco-lumbar moments during gait initiation. The ANOVAs were followed by Bonferroni post-hoc tests for p<0.05. Finally, Cohen's d coefficient was calculated for statistically significant parameters to estimate size effect. The threshold was set at 0.8 [31], [38].

Kinematic and spatio-temporal parameters

In general, as indicated by the patterns of the curves in Figure 3A, the natural inclination of the trunk in the sagittal plane varies similarly for both groups during gait initiation, but with an off-set of nearly 4°. In the period that goes from the 1^st knee movement to the right heel strike (i.e., from around 26% to 100% of gait initiation), all subjects lean forward, around 7° on average.

First, the 1^st knee movement (1^st mvt) occurs, at an average of 26%±3% of the motor task (i.e., 332 ms±73 ms after the appearance of the visual signal (VS)). In 90% of cases, it is the left knee that moves first. The interval between the 1^st movements of the left and right knee is about 17 ms. The right heel off (RHO) occurs at 55%±4% of the gait initiation, (i.e., 692 ms±120 ms after VS). Finally, the right heel strike (RHS), which characterizes the end of gait initiation, occurs at an average of 1290 ms±110 ms after VS. Considering the values expressed in percents of the motor task duration, the difference in the mean values of both BW and FW groups do not exceed 1 point. In terms of the instantaneous speed of the body COM according to the antero-posterior axis at the RHS, there is no significant difference between the BW group and the FW group (F(1.23) = 0.57; P>0.05) (Table 1). Table 1 provides the four other parameters (1^st mvt, RHO, RHS, duration of APA, reaction time), which were calculated to characterize gait initiation of the BW and FW groups. Both groups do not have significant differences for these parameters (F(4.20) = 1.15; P>0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Age, height, mass, kinematic and spatio-temporal parameters for the subjects in the Backward (BW) and Forward (FW) groups (mean±standard deviation).

https://doi.org/10.1371/journal.pone.0055256.t001

Impulse

Figure 2 shows the antero-posterior component of the ground reaction force for the stance limb (left). This component was used to calculate the antero-posterior impulse. There is no significant difference between the BW subjects (−0.83 N.s/kg±0.06 N.s/kg) and the FW subjects (−0.82 N.s/kg±0.06 N.s/kg) (F(1.23) = 0.03; P>0.05). These values are lower than the antero-posterior final COM velocity (Table 1) because only the antero-posterior component of ground reaction force of the stance lower-limb (left) was used to calculate the impulse. Mechanically, the value of the impulse per unit of mass should be the same as the one of the final COM velocity. When using the antero-posterior component of ground reaction force of the two lower limbs, it appears that the antero-posterior impulse (1.00 N.s/kg±0.08 N.s/kg for the whole population) is indeed similar (p = 0.86, df = 48) to the antero-posterior final COM velocity (0.99 m/s±0.07 m/s for the whole population), as expected from the classical mechanics.

Muscular moments

In general, the net muscular moments of the FW group show a pattern similar to the BW group, as shown in Figure 3. There is no significant difference for the two groups in terms of the ankle moments calculated at peaks A1 (P>0.05; df = 23) and A2 (P>0.05; df = 23) (Figure 3E). For the knee, there is also no significant difference between the BW and FW subjects in terms of flexor peak K1 (P>0.05; df = 23) (Figure 3D). However, just before the RHS, the knee's extensor moment K2 (Figure 3D) for the BW group (0.47 Nm/kg±0.06 Nm/kg) is 1.4 times higher (P<0.001; df = 23) than for the FW group (0.34 Nm/kg±0.10 Nm/kg). At the hip, before the RHO, the flexor moment H1 of the BW group (−0.16 Nm/kg±0.04 Nm/kg) is around 2.4 times higher (P<0.001; df = 23) than the one for the FW group (−0.07 Nm/kg±0.07 Nm/kg) (Figure 3C). The extensor moment H2, which occurs a bit after the right toe off, is around 3.1 times higher (P<0.01; df = 23) for the FW group (0.26 Nm/kg±0.16 Nm/kg) than for the BW group (0.08 Nm/kg±0.15 Nm/kg).

For the 5^th lumbar vertebrae, at the peak L1, the extensor moment for the FW group (0.36 Nm/kg±0.07 Nm/kg) is around 1.7 times higher (P<0.001; df = 23) than the one for the BW group (0.21 Nm/kg±0.09 Nm/kg) (Figure 3B). At the peak L2, the extensor moment for the FW group (0.49 Nm/kg±0.08 Nm/kg) is around 1.4 times higher (P<0.001, df = 23) than the one for the BW group (0.36 Nm/kg±0.04 Nm/kg). At the peak L3, the extensor moment for the FW group (0.34 Nm/kg±0.09 Nm/kg) is around 1.7 times higher (P<0.01; df = 23) than the one for the BW group (0.20 Nm/kg±0.11 Nm/kg). The average Cohen's d coefficient was 1.54 for all statistically significant lower limb joint moments.