Human Movement Science 42 (2015) 15–26
Contents lists available at ScienceDirect
Human Movement Science journal homepage: www.elsevier.com/locate/humov
Effects of visual focus and gait speed on walking balance in the frontal plane Adam Goodworth ⇑, Kathryn Perrone, Mark Pillsbury, Michelle Yargeau University of Hartford, Department of Rehabilitation Sciences, West Hartford, CT, United States
a r t i c l e
i n f o
PsycINFO classification: 2330 Keywords: Balance Visual feedback Gait speed Perturbations Head position
a b s t r a c t We investigated how head position and gait speed influenced frontal plane balance responses to external perturbations during gait. Thirteen healthy participants walked on a treadmill at three different gait speeds. Visual conditions included either focus downward on lower extremities and walking surface only or focus forward on a stationary scene with horizontal and vertical lines. The treadmill was positioned on a platform that was stationary (non-perturbed) or moving in a pattern that appeared random to the subjects (perturbed). In non-perturbed walking, medial–lateral upper body motion was very similar between visual conditions. However, in perturbed walking, there was significantly less body motion when focus was on the stationary visual scene, suggesting visual feedback of stationary vertical and horizontal cues are particularly important when balance is challenged. Sensitivity of body motion to perturbations was significantly decreased by increasing gait speed, suggesting that faster walking was less sensitive to frontal plane perturbations. Finally, our use of external perturbations supported the idea that certain differences in balance control mechanisms can only be detected in more challenging situations, which is an important consideration for approaches to investigating sensory contribution to balance during gait. Ó 2015 Elsevier B.V. All rights reserved.
⇑ Corresponding author at: 200 Bloomfield Ave, West Hartford, CT 06117, United States. Tel.: +1 860 768 5571. E-mail addresses: [email protected] (A. Goodworth), [email protected] (K. Perrone), [email protected] (M. Pillsbury), [email protected] (M. Yargeau). http://dx.doi.org/10.1016/j.humov.2015.04.004 0167-9457/Ó 2015 Elsevier B.V. All rights reserved.
16
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
1. Introduction Many studies have been completed to better understand the sensory contributions to balance in standing (e.g., Goodworth & Peterka, 2010; Horak & Macpherson, 1996; Peterka, 2002), but less is known about the contribution of sensory feedback to balance in gait. One reason may be that describing walking balance is complicated by the number of variables affecting sensory feedback. For example, changes in visual focus (Hollands & Marple-Horvat, 2001; Lackner & DiZio, 1988; Patla & Vickers, 2003) and gait speed will have an impact on how sensory feedback is used. Also, during gait, the base of support changes with every step and the type of surface walked upon (Marigold & Patla, 2008; Menz, Lord, St George, et al., 2004); and changes in base of support and surface characteristics are known to affect sensory contributions to standing balance (Day, Steiger, Thompson, & Marsden, 1993; Goodworth & Peterka, 2010). Therefore, to enhance our knowledge of sensory contributions to walking balance, the current study investigates how changes in head position and gait speed influence balance responses to perturbations. While several feedback systems may contribute to walking balance, it has been shown that visual feedback is particularly strong. Visual stimulation can dominate over proprioceptive feedback and is critical to adjust specific gait mechanics (Graci, Elliot, & Buckley, 2009; Iosa, Fusco, Morone, & Paolucci, 2012; Lackner & DiZio, 1988). Occluding different portions of the visual field, including peripheral vision, has a significant effect on gait mechanics (Graci et al., 2009; Marigold & Patla, 2008). Other studies have shown that visual stimulation can evoke body sway and increased variability during gait, especially in the frontal plane (McAndrew, Dingwell, & Wilken, 2010; O’Connor & Kuo, 2009). Although natural visual scanning can include looking forward at stable environmental cues or looking down at the surface and feet, the relative importance of each is not well known. In particular, when balance is perturbed with a moving surface, a person may be more stable when focusing their vision downward because they can see where to place their feet relative to the surface for optimal foot placement. Foot placement in gait forms the base of support and anchoring for balance control (Osaki, Kunin, Cohen, & Raphan, 2007; Pijnappels, Bobbert & van Dieen, 2005). Also, individuals tend to look down more when walking over uneven terrain (Rietdyk & Drifmeyer, 2010) and exhibit decreased performance when the lower visual field is occluded (Marigold & Patla, 2008). In contrast, a person may be more stable if he/she focuses forward because he/she can receive visual feedback that is congruent with gravity. Stationary visual cues aligned with gravity are known to enhance balance during gait (McAndrew et al., 2010; O’Connor & Kuo, 2009) and standing in the frontal (Goodworth & Peterka, 2010; Oie, Kiemel, & Jeka, 2002) and sagittal plane (Peterka, 2002). Another factor that can influence balance during gait is speed. There has been conflicting evidence regarding which gait speed is most stabilizing. Different relationships between gait speed and fall risk exist, with some studies showing no relationship between the two variables (Feltner, MacRae, & McNitt-Gray, 1994; Gehlsen & Whaley, 1990) and others showing that faster walking is associated with a decreased risk of falling (Lord, Lloyd, & Li, 1996; Wolfson, Whipple, Amerman, & Tobin, 1990), consistent with suggestions from Craik, Herman, and Finley (1976) and Murray (1967). In contrast, researchers have shown that gait variability is lowest (presumably most controlled) at normal walking speeds (Oberg, Karsznia, & Oberg, 1993) and others have reported that slow walking is most stable using measurements of ‘‘local dynamic stability’’ (Dingwell & Marin, 2006) and induced tripping in the sagittal plane (Bhatt, Wening, & Pai, 2005; Pavol, Owings, Foley, & Grabiner, 1999). Therefore, further investigation into the role of gait speed on balance is valuable; and the interaction with visual feedback may be especially important in the frontal plane because control in the frontal plane is more influenced by sensory feedback than the sagittal plane (Bauby & Kuo, 2000; O’Connor & Kuo, 2009). In the current study, we use three gait speeds to address the question of whether visual feedback of a stable visual scene congruent with gravity is better for frontal plane balance than visual feedback of one’s own lower extremities and surface. We apply continuous external perturbations to elicit reactive balance responses as stability is generally defined as the ability of a system to respond to perturbations (Dingwell & Marin, 2006; Reeves, Narendra, & Cholewicki, 2007). The different gait speeds also allow us to determine (a) if head position influences balance more at one gait speed compared to
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
17
others, and (b) if gait speed has a significant influence on the sensitivity to external perturbations in the frontal plane. Finally, quantitative results are interpreted in conjunction with a questionnaire assessing perception of stability in each subject. 2. Methodology 2.1. Subjects 13 healthy subjects (mean age of 28.38 years, 5 males, 8 females) participated in this study. Subjects had no known history of orthopedic impairment in the last 6 months and were free from any known balance disorders. Subjects gave written informed consent and all tests were approved by the Institutional Review Board at the University of Hartford. 2.2. General protocol Subjects performed a total of 12 balance tests under various visual and gait speed conditions using a treadmill mounted on a platform (Fig. 1). The tread on the treadmill was 0.50 m wide. The platform was able to rotate about a vertical axis to deliver perturbations to walking participants in certain test conditions. Details of platform motion are described in more detail later in Section 2.2. In each condition, subjects crossed their arms over their chest in a comfortable position and wore headphones that played a story while walking on the treadmill. The story eliminated background noise and helped normalize cognitive contributions across subjects. All subjects were instructed to walk naturally and respond naturally to any movements of the platform. Data collection lasted 60 s. Periodic breaks were provided. To minimize any possible anxiety associated with the external perturbations, subjects were given a warm up test on the treadmill with and without perturbations prior to any data collection.
(A) “Horizon”
Visual scene with vertical and horizonal cues
Upper body sensor
Perturbing platform motion Treadmill motion Platform sensor
(B)
“Head down” Upper body sensor
Perturbing platform motion Treadmill motion Platform sensor
Fig. 1. Schematic of the two visual conditions used. (A) Represents subjects walking on the treadmill while wearing goggles that occluded the lower half of their visual field while vision was focused straight ahead at a stationary visual scene with vertical and horizontal lines. (B) Represents subjects walking on the treadmill while wearing a modified head piece that forced them to look down at their lower extremities and surface, with no peripheral vision or stationary visual cues.
18
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Platform velocity (deg/s)
The 12 balance tests were randomly ordered and consisted of 2 different visual conditions, 2 perturbation conditions, and 3 gait speeds. The 2 visual conditions were denoted as ‘‘horizon’’ or ‘‘head down’’ (Fig. 1). The ‘‘horizon’’ condition consisted of subjects wearing a pair of goggles that restricted vision of the lower third of their visual field. In the horizon condition, subjects were instructed to focus on the vertical and horizontal orienting cues (congruent with gravity) of a visual scene on the wall in front of them. The ‘‘head down’’ condition utilized a custom made headpiece that was adjustable to each subject’s head. The material extended past each subject’s forehead as well as next to their eyes to the level of their nose. The construction of the headpiece allowed us to occlude peripheral vision, forcing subjects to focus on their lower extremities and the treadmill. The two perturbation conditions were either ‘‘with perturbation’’ (WP), whereby the platform rotated about a vertical axis evoking external, frontal plane perturbations that appeared random to the participants, or ‘‘no perturbation’’ (NP), whereby the platform was stationary and subjects performed standard treadmill walking. The WP tests were used to investigate balance in a more dynamic and challenging context because certain balance reactions may only be detectable in more challenging conditions (Horak, Henry, & Shumway-Cook, 1997; Horak, Wrisley, & Frank, 2009; McAndrew et al., 2010; O’Connor & Kuo, 2009). In WP tests, perturbations started 6 s after the test began and subjects perceived the platform motion as unpredictable. The perturbations consisted of a sum of five continuous sine waves at different frequencies (0.5, 1.0, 1.5, 2.0, and 2.5 Hz). Perturbations evoked continuous medial–lateral (ML) balance responses (see Fig. 2 for an example) meant to simulate walking on an unpredictable terrain (Goodworth et al., 2014). Previous studies indicate that frontal plane stability involves more active neural control than sagittal plane (Kuo & Donelan, 2010) and frontal plane motion during gait may be particularly sensitive to external perturbations (Hof, Vermerris, & Gjaltema, 2010; McAndrew et al., 2010; O’Connor & Kuo, 2009). Therefore, applying frontal plane perturbations was an ideal method to address the research question about visual focus in the current study. Each subject’s position on the treadmill was 0.85 m away from the center of the platform and this position was visually monitored by an experimenter throughout every test. At this location on the treadmill, the peak-to-peak displacement of the platform was 0.032 m and the peak velocity was 0.14 m/s in WP conditions. Platform velocity was measured using a high precision gyroscope (Watson Industries, WI, USA) at 200 samples per second.
5
... -5
start of platform motion
Upper body ML tilt (deg)
20
... -20 20
0
...
55
60
Time (sec) Fig. 2. Sample data from one subject. Medial–lateral (ML) body motion is shown at a chosen gait speed, with perturbations, and with a horizontal visual condition.
19
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Finally, the three gait speed conditions used in this study included slow, chosen, or fast. The chosen speed was self-selected by each subject, limited to a range between 0.89 and 1.01 m/s. After the selfselected speed was chosen, 0.14 m/s was added or subtracted to the chosen speed to determine each subject’s fast and slow speed, respectively. The mean subject comfortable treadmill speed was 0.95 m/ s. 4 subjects chose 1.01 m/s, 6 subjects chose 0.89 m/s, and 3 subjects chose a speed between 0.89 and 1.01 m/s. All subjects reported the chosen speed as comfortable. The chosen speed was limited to a specific range so that fast and slow speeds stayed within a normal range of walking speeds. 2.3. Kinematic variables In each test, trunk motion in the sagittal and frontal plane was measured using accelerometerbased dual-axis tilt sensors at 200 samples per second (Goodworth et al., 2014). The trunk sensor was secured between the two scapulae at the level of the T7 vertebrae (Figs. 1 and 2). Kinematics of the trunk were measured, similar to previous studies (Dingwell & Marin, 2006; Iosa et al., 2012; Winter, 1989) because its large mass located distal to the base of support is critical to control for maintenance of balance (Winter, 1989). Outputs from trunk tilt sensors were related to the amount of trunk tilt and the velocity of trunk tilt throughout each test. 12 kinematic variables were calculated (Table 1), and divided into steady-state or initial time periods because we considered the possibility that initial responses to the perturbation may differ from steady-state responses. The initial time period included only the first 2 s following the onset of the perturbation. Since perturbations were only used in half of the tests, these kinematic variables are only reported for WP test conditions. The steady-state period included all data from 6 s following the onset of the perturbation to the end of the test, resulting in 48 s of steady-state data. Standard balance-related measures were taken, zero-mean root-mean-square (RMS) and peak-topeak. We also calculated a measure of sensitivity to perturbations as the ratio of body motion in WP to NP conditions. This ratio was selected to normalize data for the observed increase in trunk motion with increasing gait speed. Sensitivity values equal to one indicated subjects’ RMS motion was unaffected by the perturbation, whereas sensitivity values greater than one indicated subjects’ RMS motion was increased during conditions with perturbations. 2.4. Questionnaire Because there is a close relationship between perception and motor control, we created a custom balance perception questionnaire which was administered to subjects following the completion of all
Table 1 List of upper body kinematic variables. ‘‘RMS’’ stands for zero-mean root-mean-square, ‘‘WP’’ for with perturbation and ‘‘NP’’ for no perturbation. Sensitivity to perturbations was measured as ratio of WP to NP. Variable Medial–lateral kinematics RMS tilt RMS tilt velocity Peak-to-peak tilt WP RMS tilt NP RMS tilt WP RMS tilt velocity NP RMS tilt velocity WP peak-to-peak tilt NP peak-to-peak tilt
Time period Steady-state Steady-state Steady-state Steady-state Steady-state Steady-state
RMS tilt RMS tilt velocity Peak-to-peak tilt
Initial Initial Initial
Anterior–posterior kinematics RMS tilt RMS tilt velocity Peak-to-peak tilt
Steady-state Steady-state Steady-state
20
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
tests. Responses to the questionnaire provided a complimentary interpretation to the kinematic data. The questionnaire included six questions, both open and closed ended. The questions included: (1) did you feel safe while walking on the treadmill? (2) Did you notice any difference in how you walk when you were looking at your feet compared to looking straight ahead? (3) How was the perturbation when you were walking slowly? (4) How was the perturbation when you were walking at your chosen speed? (5) How was the perturbation when you were walking fast? (6) Which walking speed did you feel the most stable? 2.5. Statistics To determine if visual condition, perturbation, or gait speed had a significant effect on trunk kinematics, we used separate repeated-measures ANOVAs for each kinematic variable (Table 1). For the steady-state medial–lateral (ML) and anterior–posterior (AP) variables, RMS tilt, RMS tilt velocity, and peak-to-peak tilt, model effects were visual condition (horizon vs. head down), platform perturbation (WP vs. NP), and gait speed. Model effects were visual condition and gait speed for the 3 steadystate sensitivity variables WP RMS tilt/NP RMS tilt, WP RMS tilt velocity/NP RMS tilt velocity, and WP peak-to-peak tilt/NP peak-to-peak tilt, and for the 3 initial period variables RMS tilt, RMS tilt velocity, and peak-to-peak tilt. We also tested for interaction effects in all variables. In all statistical tests, P < .05 was considered significant. 3. Results 3.1. General trunk motion behavior Fig. 2 illustrates sample ML trunk data from one subject at the chosen gait speed, WP, and with a horizon visual condition. During the first 6 s of the test, prior to perturbation onset, trunk motion was steady and cyclical with amplitude of about 8°. Following onset of the perturbation, trunk motion remained cyclical but was less consistent in amplitude, with amplitudes typically ranging from about 10–12°. 3.2. Steady-state medial–lateral trunk motion across subjects Detailed results from a selection of ML trunk motion variables across all subjects are shown in Fig. 3. In the NP condition (Fig. 3A), ML RMS tilt linearly increased with gait speed and the visual condition had essentially no impact on the trunk motion. As expected, in the WP condition, ML RMS tilt was higher than in the NP condition. More importantly, in the WP condition (Fig. 3B), there was a clear increase in ML RMS tilt in the head down condition compared to the horizon condition. Perturbation, visual condition, and speed all had significant effects on ML RMS tilt, as indicated in Table 2. Body motion was significantly increased with perturbations compared to non-perturbations, increases in gait speed, and in head down compared to horizontal conditions. In addition, there was a significant interaction effect between perturbation and visual conditions, indicating that the visual condition had a significantly larger impact on ML tilt in the WP condition compared to NP. There was also a significant interaction effect of perturbation and speed, indicating that gait speed influenced ML tilts more in NP compared to WP conditions. The ML RMS tilt velocity results (Fig. 3C and D), showed similar trends. Perturbation, visual condition, and speed each had a significant effect on ML tilt velocity, similar to tilt position results. In addition, there was a significant interaction effect between perturbation and visual conditions, indicating that the visual condition had a significantly larger impact on ML tilt velocity in the WP condition compared to NP. To characterize sensitivity to the perturbations, the ratio of WP to NP was calculated for ML RMS tilt and tilt velocity (Fig. 3E and F). In all cases, sensitivity was greater than one, indicating that subjects had greater trunk motion in the WP compared to the NP condition. Sensitivity values were larger in the head down (gray bar plots) compared to the horizon (white bar plots) condition. In both the
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
(A)
(B)
NP
WP
RMS tilt (deg)
8
“Head down” “Horizon”
4
(C)
(D)
140
WP
NP
RMS tilt velocity (deg/s)
21
40
NP (deg)
WP (deg)
st
2 “Head down”
“Horizon”
1 NP Vel (deg/s)
(F) WP Vel (deg/s)
fa
en
en
os
st
ow
ch
sl
fa
os
ow
ch
sl
(E)
2
1 st
fa
en
os
ch
ow
sl
st
fa
en
os
ch
ow
sl
Fig. 3. Summary results for several medial–lateral trunk sway kinematics. (A) Represents the RMS tilt in the ‘‘no perturbation’’ (NP) condition. (B) Represents the RMS tilt in the ‘‘with perturbation’’ (WP) condition. (C) and (D) Represent RMS tilt velocity in the NP and WP conditions, respectively. Error bars are +1 SE. (E) and (F) Represent the sensitivity to perturbations, defined as the relative increase in motion in the WP conditions with respect to the NP conditions for RMS tilt and tilt velocity, respectively. The relative decrease in tilt or tilt velocity as the subject walked faster indicates a decrease in sensitivity to the perturbations with increasing gait speed.
horizon and head down condition, there was a trend toward lower sensitivity values with increasing gait speed. Statistical analyses revealed that both visual condition and gait speed had a significant effect on WP/NP tilt and tilt velocity, indicating that subjects’ trunk motion was more sensitive to perturbations when looking down and at slow gait speeds. This finding is consistent with the significant interaction effect between gait speed and perturbation in the ML RMS tilt variable. 3.3. Statistical summary of kinematics Table 2 provides a statistical summary of P values for all kinematic variables across subjects. ML and AP dependent variables (steady state RMS tilt, RMS tilt velocity and peak-to-peak tilt) were significantly affected by at least one (if not all) of the main independent variables (vision, perturbation, and/or speed). Interaction effects between vision and perturbation were present in several variables. There were no variables where a significant interaction effect between vision and speed was found. This result indicates that the vision condition did not systematically influence trunk motion differently across gait speeds. Trunk responses during the first 2 s following the onset of the perturbation (i.e., initial) were similar to steady-state responses in that vision and speed had significant effects on trunk motion, but there was not a significant interaction effect between vision and speed.
22
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Table 2 Summary of statistical analyses for medial–lateral (ML) and anterior–posterior (AP) motion. All variables are in ‘‘steady-state’’ unless otherwise noted as ‘‘initial’’ in parentheses. ‘‘RMS’’ stands for root-mean-square, ‘‘WP’’ stands for with perturbations, ‘‘NP’’ stands for no perturbations and ‘‘pert’’ stands for perturbations. Bolded values indicate statistical significance of P < .05. Statistical summary of P values
ML trunk variable RMS tilt RMS tilt velocity Peak-to-peak tilt WP RMS tilt NP RMS tilt WP RMS tilt velocity NP RMS tilt velocity WP peak-to-peak tilt NP peak-to-peak tilt
RMS tilt (initial) RMS tilt velocity (initial) Peak-to-peak tilt (initial) AP trunk variable RMS tilt RMS tilt velocity Peak-to-peak tilt
Vision
Pert
Speed
Contents lists available at ScienceDirect
Human Movement Science journal homepage: www.elsevier.com/locate/humov
Effects of visual focus and gait speed on walking balance in the frontal plane Adam Goodworth ⇑, Kathryn Perrone, Mark Pillsbury, Michelle Yargeau University of Hartford, Department of Rehabilitation Sciences, West Hartford, CT, United States
a r t i c l e
i n f o
PsycINFO classification: 2330 Keywords: Balance Visual feedback Gait speed Perturbations Head position
a b s t r a c t We investigated how head position and gait speed influenced frontal plane balance responses to external perturbations during gait. Thirteen healthy participants walked on a treadmill at three different gait speeds. Visual conditions included either focus downward on lower extremities and walking surface only or focus forward on a stationary scene with horizontal and vertical lines. The treadmill was positioned on a platform that was stationary (non-perturbed) or moving in a pattern that appeared random to the subjects (perturbed). In non-perturbed walking, medial–lateral upper body motion was very similar between visual conditions. However, in perturbed walking, there was significantly less body motion when focus was on the stationary visual scene, suggesting visual feedback of stationary vertical and horizontal cues are particularly important when balance is challenged. Sensitivity of body motion to perturbations was significantly decreased by increasing gait speed, suggesting that faster walking was less sensitive to frontal plane perturbations. Finally, our use of external perturbations supported the idea that certain differences in balance control mechanisms can only be detected in more challenging situations, which is an important consideration for approaches to investigating sensory contribution to balance during gait. Ó 2015 Elsevier B.V. All rights reserved.
⇑ Corresponding author at: 200 Bloomfield Ave, West Hartford, CT 06117, United States. Tel.: +1 860 768 5571. E-mail addresses: [email protected] (A. Goodworth), [email protected] (K. Perrone), [email protected] (M. Pillsbury), [email protected] (M. Yargeau). http://dx.doi.org/10.1016/j.humov.2015.04.004 0167-9457/Ó 2015 Elsevier B.V. All rights reserved.
16
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
1. Introduction Many studies have been completed to better understand the sensory contributions to balance in standing (e.g., Goodworth & Peterka, 2010; Horak & Macpherson, 1996; Peterka, 2002), but less is known about the contribution of sensory feedback to balance in gait. One reason may be that describing walking balance is complicated by the number of variables affecting sensory feedback. For example, changes in visual focus (Hollands & Marple-Horvat, 2001; Lackner & DiZio, 1988; Patla & Vickers, 2003) and gait speed will have an impact on how sensory feedback is used. Also, during gait, the base of support changes with every step and the type of surface walked upon (Marigold & Patla, 2008; Menz, Lord, St George, et al., 2004); and changes in base of support and surface characteristics are known to affect sensory contributions to standing balance (Day, Steiger, Thompson, & Marsden, 1993; Goodworth & Peterka, 2010). Therefore, to enhance our knowledge of sensory contributions to walking balance, the current study investigates how changes in head position and gait speed influence balance responses to perturbations. While several feedback systems may contribute to walking balance, it has been shown that visual feedback is particularly strong. Visual stimulation can dominate over proprioceptive feedback and is critical to adjust specific gait mechanics (Graci, Elliot, & Buckley, 2009; Iosa, Fusco, Morone, & Paolucci, 2012; Lackner & DiZio, 1988). Occluding different portions of the visual field, including peripheral vision, has a significant effect on gait mechanics (Graci et al., 2009; Marigold & Patla, 2008). Other studies have shown that visual stimulation can evoke body sway and increased variability during gait, especially in the frontal plane (McAndrew, Dingwell, & Wilken, 2010; O’Connor & Kuo, 2009). Although natural visual scanning can include looking forward at stable environmental cues or looking down at the surface and feet, the relative importance of each is not well known. In particular, when balance is perturbed with a moving surface, a person may be more stable when focusing their vision downward because they can see where to place their feet relative to the surface for optimal foot placement. Foot placement in gait forms the base of support and anchoring for balance control (Osaki, Kunin, Cohen, & Raphan, 2007; Pijnappels, Bobbert & van Dieen, 2005). Also, individuals tend to look down more when walking over uneven terrain (Rietdyk & Drifmeyer, 2010) and exhibit decreased performance when the lower visual field is occluded (Marigold & Patla, 2008). In contrast, a person may be more stable if he/she focuses forward because he/she can receive visual feedback that is congruent with gravity. Stationary visual cues aligned with gravity are known to enhance balance during gait (McAndrew et al., 2010; O’Connor & Kuo, 2009) and standing in the frontal (Goodworth & Peterka, 2010; Oie, Kiemel, & Jeka, 2002) and sagittal plane (Peterka, 2002). Another factor that can influence balance during gait is speed. There has been conflicting evidence regarding which gait speed is most stabilizing. Different relationships between gait speed and fall risk exist, with some studies showing no relationship between the two variables (Feltner, MacRae, & McNitt-Gray, 1994; Gehlsen & Whaley, 1990) and others showing that faster walking is associated with a decreased risk of falling (Lord, Lloyd, & Li, 1996; Wolfson, Whipple, Amerman, & Tobin, 1990), consistent with suggestions from Craik, Herman, and Finley (1976) and Murray (1967). In contrast, researchers have shown that gait variability is lowest (presumably most controlled) at normal walking speeds (Oberg, Karsznia, & Oberg, 1993) and others have reported that slow walking is most stable using measurements of ‘‘local dynamic stability’’ (Dingwell & Marin, 2006) and induced tripping in the sagittal plane (Bhatt, Wening, & Pai, 2005; Pavol, Owings, Foley, & Grabiner, 1999). Therefore, further investigation into the role of gait speed on balance is valuable; and the interaction with visual feedback may be especially important in the frontal plane because control in the frontal plane is more influenced by sensory feedback than the sagittal plane (Bauby & Kuo, 2000; O’Connor & Kuo, 2009). In the current study, we use three gait speeds to address the question of whether visual feedback of a stable visual scene congruent with gravity is better for frontal plane balance than visual feedback of one’s own lower extremities and surface. We apply continuous external perturbations to elicit reactive balance responses as stability is generally defined as the ability of a system to respond to perturbations (Dingwell & Marin, 2006; Reeves, Narendra, & Cholewicki, 2007). The different gait speeds also allow us to determine (a) if head position influences balance more at one gait speed compared to
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
17
others, and (b) if gait speed has a significant influence on the sensitivity to external perturbations in the frontal plane. Finally, quantitative results are interpreted in conjunction with a questionnaire assessing perception of stability in each subject. 2. Methodology 2.1. Subjects 13 healthy subjects (mean age of 28.38 years, 5 males, 8 females) participated in this study. Subjects had no known history of orthopedic impairment in the last 6 months and were free from any known balance disorders. Subjects gave written informed consent and all tests were approved by the Institutional Review Board at the University of Hartford. 2.2. General protocol Subjects performed a total of 12 balance tests under various visual and gait speed conditions using a treadmill mounted on a platform (Fig. 1). The tread on the treadmill was 0.50 m wide. The platform was able to rotate about a vertical axis to deliver perturbations to walking participants in certain test conditions. Details of platform motion are described in more detail later in Section 2.2. In each condition, subjects crossed their arms over their chest in a comfortable position and wore headphones that played a story while walking on the treadmill. The story eliminated background noise and helped normalize cognitive contributions across subjects. All subjects were instructed to walk naturally and respond naturally to any movements of the platform. Data collection lasted 60 s. Periodic breaks were provided. To minimize any possible anxiety associated with the external perturbations, subjects were given a warm up test on the treadmill with and without perturbations prior to any data collection.
(A) “Horizon”
Visual scene with vertical and horizonal cues
Upper body sensor
Perturbing platform motion Treadmill motion Platform sensor
(B)
“Head down” Upper body sensor
Perturbing platform motion Treadmill motion Platform sensor
Fig. 1. Schematic of the two visual conditions used. (A) Represents subjects walking on the treadmill while wearing goggles that occluded the lower half of their visual field while vision was focused straight ahead at a stationary visual scene with vertical and horizontal lines. (B) Represents subjects walking on the treadmill while wearing a modified head piece that forced them to look down at their lower extremities and surface, with no peripheral vision or stationary visual cues.
18
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Platform velocity (deg/s)
The 12 balance tests were randomly ordered and consisted of 2 different visual conditions, 2 perturbation conditions, and 3 gait speeds. The 2 visual conditions were denoted as ‘‘horizon’’ or ‘‘head down’’ (Fig. 1). The ‘‘horizon’’ condition consisted of subjects wearing a pair of goggles that restricted vision of the lower third of their visual field. In the horizon condition, subjects were instructed to focus on the vertical and horizontal orienting cues (congruent with gravity) of a visual scene on the wall in front of them. The ‘‘head down’’ condition utilized a custom made headpiece that was adjustable to each subject’s head. The material extended past each subject’s forehead as well as next to their eyes to the level of their nose. The construction of the headpiece allowed us to occlude peripheral vision, forcing subjects to focus on their lower extremities and the treadmill. The two perturbation conditions were either ‘‘with perturbation’’ (WP), whereby the platform rotated about a vertical axis evoking external, frontal plane perturbations that appeared random to the participants, or ‘‘no perturbation’’ (NP), whereby the platform was stationary and subjects performed standard treadmill walking. The WP tests were used to investigate balance in a more dynamic and challenging context because certain balance reactions may only be detectable in more challenging conditions (Horak, Henry, & Shumway-Cook, 1997; Horak, Wrisley, & Frank, 2009; McAndrew et al., 2010; O’Connor & Kuo, 2009). In WP tests, perturbations started 6 s after the test began and subjects perceived the platform motion as unpredictable. The perturbations consisted of a sum of five continuous sine waves at different frequencies (0.5, 1.0, 1.5, 2.0, and 2.5 Hz). Perturbations evoked continuous medial–lateral (ML) balance responses (see Fig. 2 for an example) meant to simulate walking on an unpredictable terrain (Goodworth et al., 2014). Previous studies indicate that frontal plane stability involves more active neural control than sagittal plane (Kuo & Donelan, 2010) and frontal plane motion during gait may be particularly sensitive to external perturbations (Hof, Vermerris, & Gjaltema, 2010; McAndrew et al., 2010; O’Connor & Kuo, 2009). Therefore, applying frontal plane perturbations was an ideal method to address the research question about visual focus in the current study. Each subject’s position on the treadmill was 0.85 m away from the center of the platform and this position was visually monitored by an experimenter throughout every test. At this location on the treadmill, the peak-to-peak displacement of the platform was 0.032 m and the peak velocity was 0.14 m/s in WP conditions. Platform velocity was measured using a high precision gyroscope (Watson Industries, WI, USA) at 200 samples per second.
5
... -5
start of platform motion
Upper body ML tilt (deg)
20
... -20 20
0
...
55
60
Time (sec) Fig. 2. Sample data from one subject. Medial–lateral (ML) body motion is shown at a chosen gait speed, with perturbations, and with a horizontal visual condition.
19
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Finally, the three gait speed conditions used in this study included slow, chosen, or fast. The chosen speed was self-selected by each subject, limited to a range between 0.89 and 1.01 m/s. After the selfselected speed was chosen, 0.14 m/s was added or subtracted to the chosen speed to determine each subject’s fast and slow speed, respectively. The mean subject comfortable treadmill speed was 0.95 m/ s. 4 subjects chose 1.01 m/s, 6 subjects chose 0.89 m/s, and 3 subjects chose a speed between 0.89 and 1.01 m/s. All subjects reported the chosen speed as comfortable. The chosen speed was limited to a specific range so that fast and slow speeds stayed within a normal range of walking speeds. 2.3. Kinematic variables In each test, trunk motion in the sagittal and frontal plane was measured using accelerometerbased dual-axis tilt sensors at 200 samples per second (Goodworth et al., 2014). The trunk sensor was secured between the two scapulae at the level of the T7 vertebrae (Figs. 1 and 2). Kinematics of the trunk were measured, similar to previous studies (Dingwell & Marin, 2006; Iosa et al., 2012; Winter, 1989) because its large mass located distal to the base of support is critical to control for maintenance of balance (Winter, 1989). Outputs from trunk tilt sensors were related to the amount of trunk tilt and the velocity of trunk tilt throughout each test. 12 kinematic variables were calculated (Table 1), and divided into steady-state or initial time periods because we considered the possibility that initial responses to the perturbation may differ from steady-state responses. The initial time period included only the first 2 s following the onset of the perturbation. Since perturbations were only used in half of the tests, these kinematic variables are only reported for WP test conditions. The steady-state period included all data from 6 s following the onset of the perturbation to the end of the test, resulting in 48 s of steady-state data. Standard balance-related measures were taken, zero-mean root-mean-square (RMS) and peak-topeak. We also calculated a measure of sensitivity to perturbations as the ratio of body motion in WP to NP conditions. This ratio was selected to normalize data for the observed increase in trunk motion with increasing gait speed. Sensitivity values equal to one indicated subjects’ RMS motion was unaffected by the perturbation, whereas sensitivity values greater than one indicated subjects’ RMS motion was increased during conditions with perturbations. 2.4. Questionnaire Because there is a close relationship between perception and motor control, we created a custom balance perception questionnaire which was administered to subjects following the completion of all
Table 1 List of upper body kinematic variables. ‘‘RMS’’ stands for zero-mean root-mean-square, ‘‘WP’’ for with perturbation and ‘‘NP’’ for no perturbation. Sensitivity to perturbations was measured as ratio of WP to NP. Variable Medial–lateral kinematics RMS tilt RMS tilt velocity Peak-to-peak tilt WP RMS tilt NP RMS tilt WP RMS tilt velocity NP RMS tilt velocity WP peak-to-peak tilt NP peak-to-peak tilt
Time period Steady-state Steady-state Steady-state Steady-state Steady-state Steady-state
RMS tilt RMS tilt velocity Peak-to-peak tilt
Initial Initial Initial
Anterior–posterior kinematics RMS tilt RMS tilt velocity Peak-to-peak tilt
Steady-state Steady-state Steady-state
20
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
tests. Responses to the questionnaire provided a complimentary interpretation to the kinematic data. The questionnaire included six questions, both open and closed ended. The questions included: (1) did you feel safe while walking on the treadmill? (2) Did you notice any difference in how you walk when you were looking at your feet compared to looking straight ahead? (3) How was the perturbation when you were walking slowly? (4) How was the perturbation when you were walking at your chosen speed? (5) How was the perturbation when you were walking fast? (6) Which walking speed did you feel the most stable? 2.5. Statistics To determine if visual condition, perturbation, or gait speed had a significant effect on trunk kinematics, we used separate repeated-measures ANOVAs for each kinematic variable (Table 1). For the steady-state medial–lateral (ML) and anterior–posterior (AP) variables, RMS tilt, RMS tilt velocity, and peak-to-peak tilt, model effects were visual condition (horizon vs. head down), platform perturbation (WP vs. NP), and gait speed. Model effects were visual condition and gait speed for the 3 steadystate sensitivity variables WP RMS tilt/NP RMS tilt, WP RMS tilt velocity/NP RMS tilt velocity, and WP peak-to-peak tilt/NP peak-to-peak tilt, and for the 3 initial period variables RMS tilt, RMS tilt velocity, and peak-to-peak tilt. We also tested for interaction effects in all variables. In all statistical tests, P < .05 was considered significant. 3. Results 3.1. General trunk motion behavior Fig. 2 illustrates sample ML trunk data from one subject at the chosen gait speed, WP, and with a horizon visual condition. During the first 6 s of the test, prior to perturbation onset, trunk motion was steady and cyclical with amplitude of about 8°. Following onset of the perturbation, trunk motion remained cyclical but was less consistent in amplitude, with amplitudes typically ranging from about 10–12°. 3.2. Steady-state medial–lateral trunk motion across subjects Detailed results from a selection of ML trunk motion variables across all subjects are shown in Fig. 3. In the NP condition (Fig. 3A), ML RMS tilt linearly increased with gait speed and the visual condition had essentially no impact on the trunk motion. As expected, in the WP condition, ML RMS tilt was higher than in the NP condition. More importantly, in the WP condition (Fig. 3B), there was a clear increase in ML RMS tilt in the head down condition compared to the horizon condition. Perturbation, visual condition, and speed all had significant effects on ML RMS tilt, as indicated in Table 2. Body motion was significantly increased with perturbations compared to non-perturbations, increases in gait speed, and in head down compared to horizontal conditions. In addition, there was a significant interaction effect between perturbation and visual conditions, indicating that the visual condition had a significantly larger impact on ML tilt in the WP condition compared to NP. There was also a significant interaction effect of perturbation and speed, indicating that gait speed influenced ML tilts more in NP compared to WP conditions. The ML RMS tilt velocity results (Fig. 3C and D), showed similar trends. Perturbation, visual condition, and speed each had a significant effect on ML tilt velocity, similar to tilt position results. In addition, there was a significant interaction effect between perturbation and visual conditions, indicating that the visual condition had a significantly larger impact on ML tilt velocity in the WP condition compared to NP. To characterize sensitivity to the perturbations, the ratio of WP to NP was calculated for ML RMS tilt and tilt velocity (Fig. 3E and F). In all cases, sensitivity was greater than one, indicating that subjects had greater trunk motion in the WP compared to the NP condition. Sensitivity values were larger in the head down (gray bar plots) compared to the horizon (white bar plots) condition. In both the
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
(A)
(B)
NP
WP
RMS tilt (deg)
8
“Head down” “Horizon”
4
(C)
(D)
140
WP
NP
RMS tilt velocity (deg/s)
21
40
NP (deg)
WP (deg)
st
2 “Head down”
“Horizon”
1 NP Vel (deg/s)
(F) WP Vel (deg/s)
fa
en
en
os
st
ow
ch
sl
fa
os
ow
ch
sl
(E)
2
1 st
fa
en
os
ch
ow
sl
st
fa
en
os
ch
ow
sl
Fig. 3. Summary results for several medial–lateral trunk sway kinematics. (A) Represents the RMS tilt in the ‘‘no perturbation’’ (NP) condition. (B) Represents the RMS tilt in the ‘‘with perturbation’’ (WP) condition. (C) and (D) Represent RMS tilt velocity in the NP and WP conditions, respectively. Error bars are +1 SE. (E) and (F) Represent the sensitivity to perturbations, defined as the relative increase in motion in the WP conditions with respect to the NP conditions for RMS tilt and tilt velocity, respectively. The relative decrease in tilt or tilt velocity as the subject walked faster indicates a decrease in sensitivity to the perturbations with increasing gait speed.
horizon and head down condition, there was a trend toward lower sensitivity values with increasing gait speed. Statistical analyses revealed that both visual condition and gait speed had a significant effect on WP/NP tilt and tilt velocity, indicating that subjects’ trunk motion was more sensitive to perturbations when looking down and at slow gait speeds. This finding is consistent with the significant interaction effect between gait speed and perturbation in the ML RMS tilt variable. 3.3. Statistical summary of kinematics Table 2 provides a statistical summary of P values for all kinematic variables across subjects. ML and AP dependent variables (steady state RMS tilt, RMS tilt velocity and peak-to-peak tilt) were significantly affected by at least one (if not all) of the main independent variables (vision, perturbation, and/or speed). Interaction effects between vision and perturbation were present in several variables. There were no variables where a significant interaction effect between vision and speed was found. This result indicates that the vision condition did not systematically influence trunk motion differently across gait speeds. Trunk responses during the first 2 s following the onset of the perturbation (i.e., initial) were similar to steady-state responses in that vision and speed had significant effects on trunk motion, but there was not a significant interaction effect between vision and speed.
22
A. Goodworth et al. / Human Movement Science 42 (2015) 15–26
Table 2 Summary of statistical analyses for medial–lateral (ML) and anterior–posterior (AP) motion. All variables are in ‘‘steady-state’’ unless otherwise noted as ‘‘initial’’ in parentheses. ‘‘RMS’’ stands for root-mean-square, ‘‘WP’’ stands for with perturbations, ‘‘NP’’ stands for no perturbations and ‘‘pert’’ stands for perturbations. Bolded values indicate statistical significance of P < .05. Statistical summary of P values
ML trunk variable RMS tilt RMS tilt velocity Peak-to-peak tilt WP RMS tilt NP RMS tilt WP RMS tilt velocity NP RMS tilt velocity WP peak-to-peak tilt NP peak-to-peak tilt
RMS tilt (initial) RMS tilt velocity (initial) Peak-to-peak tilt (initial) AP trunk variable RMS tilt RMS tilt velocity Peak-to-peak tilt
Vision
Pert
Speed
