Pathophysiological and motor factors associated with collision avoidance behavior in individuals with stroke

2.1 Participants

Patients were recruited from two rehabilitation hospitals in Japan. All the participants provided written informed consent. This study was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committee.

The participants were patients in or discharged from a subacute hospital (more than 1 month after stroke onset). The inclusion criteria were unilateral stroke and ability to independently walk for more than 100 m with or without an assistive device. The exclusion criteria were as follows: (a) neurological, orthopedic, or other disorders that could affect walking; (b) a history of visual deficits; (c) visual field deficits and visual spatial neglect; or (d) a score of less than 24 on the Mini-Mental State Examination (MMSE) (Holsinger et al., 2007). The following four assessments were conducted in the order of assessment of patient characteristics, perceptual judgment test, assessment of walking-through-an-opening test, and measurement of relative minimum passable width.

2.2 Assessment of patient characteristics

Participant characteristics and pathophysiological documentation of individuals with stroke were obtained from their medical records. All participants with stroke were asked whether they had fallen during the past 12 months. A fall was defined as an event that resulted in a person coming to rest unintentionally on the ground or another lower level (Clark et al., 1993). Participants’ cognitive function was assessed using the MMSE (Holsinger et al., 2007). Functional mobility was assessed using the Timed Up and Go (TUG) test (Podsiadlo and Richardson, 1991): participants were instructed to stand up from a standard chair, walk a distance of 3 m at maximum speed, turn, walk back to the chair, and sit down. To describe the impairment in stroke participants, the upper and lower extremity Brunnstrom recovery stages (BRS) (Brunnstrom, 1966), which assess the degree of recovery and muscle tone of upper and lower extremity mobility, were determined. Muscle tone of the upper and lower limbs was classified into 5 categories (normal, increased mildly, increased moderately, decreased, and increased) by assessing the passive resistance of the elbow and knee joint flexion and extension using the Stroke Impairment Assessment Set (SIAS) criteria (Tsuji et al., 2000; Liu et al., 2002). These measurements were performed in a random order.

2.3 Perceptual judgment test

Participants stood 4 m in front of an opening. They observed openings of various widths and reported whether they believed they would be able to walk through the opening without colliding. A series of opening widths was presented in either an ascending or descending order with 2-cm intervals (staircase method). The presentation of the series was terminated when the participants alternated their response (i.e., from passable to impassable, or from impassable to passable) six times. The perceived minimum passable width was averaged from 6 times response, and each ascending and descending order was averaged. Perceptual judgment passability is the perceived minimum passable width relative to the minimum passable width. We interpreted that if the relative value was close to 1.0 times their shoulder width, the participants’ judgment was correct; however, if it was less than 1.0, they underestimated the space (Cowie et al., 2010; Muroi, Hiroi, et al., 2017).

2.4 Assessment of walking-through-an-opening test

Participants were instructed to walk through an opening without colliding with a screen (Fig. 1a). Four different opening widths were used: 0.9-, 1.0-, 1.1-, and 1.2- times the width of the participants’ shoulders. The opening widths were randomly selected. To ensure the participants’ safety, two therapists stood behind the door opening while the participants performed the task. The participants were required to start walking and were permitted to rotate their bodies when necessary to avoid collisions. Each participant performed a total of 12 trials (three trials for each of the four opening widths). Three therapists observed the participants’ spontaneous body rotation and determined the collision with the opening and which body side entered first.

OPEN IN VIEWER

Fig. 1

(a) The walking-through-an-opening test. Participants were permitted to rotate their bodies when necessary to avoid collisions. (b) Assessment of relative minimum passable width. Participants walked through the opening without body rotation.

At the end of the assessment, participants’ minimum passable widths were measured (Fig. 1b). The minimum passable width was defined as the minimum width that a participant could walk through the opening twice consecutively, without body rotation. The relative minimum passable width is the ratio between the minimum passable width and the participant’s shoulder width. A large relative value indicates that the left-right (frontal plane) sway during the task is large, because they collide with the door due to left-right sway while walking.

2.6 Statistical analysis

We described the characteristics for all participants and two outcomes during the walking-through-an-opening test: (1) the number of collisions and (2) the direction of entry. The number of collisions was categorized as “high” if there were two or more collisions, or “low” if less than two. We classified the direction of entering when a subject walked through the opening as the “paretic” or “non-paretic” side. We calculated the mean±standard deviation (SD) for continuous variables.

Multiple regression analyses were conducted with pathophysiology, motor function, and judgment ability as predictor variables, and number of collisions and the frequency of entering the opening from the non-paretic side as outcome variables. Logistic regression analysis for each outcome was conducted using the factors selected based on the hypothesis. At first, we estimated the association of number of collisions with relative minimum passable widths and the TUG test. In the model, by adding variables of (i) upper and lower limb muscle tone, (ii) falls in the past year, and (iii) perceptual judgment passability, we verified changes in the association. The relative minimum passable width to the shoulder width was separated at a cut-off of 1.1. The muscle tone in the upper or lower limb, measured using the SIAS, was grouped as normal and others (Brunnstrom, 1966; Tsuji et al., 2000): 3 corresponds to normal, and 2 or less corresponds to increased or decreased. Perceptual judgment passability was separated using a cut-off of 1.0. We also estimated the association of the direction of entry with pathophysiology factors: (i) lesion on thalamus, (ii) hemiplegic side, and (iii) motor factor Brunnstrom stage. Then we added a variable of perceptual judgment passability to the model. The Brunnstrom stage was cut-off at stage 3. All analyses were performed using R software, version 4.0.5, and a p-value<0.05 was considered statistically significant.

4.1 Factors resulting in collisions during obstacle avoidance tasks

We first investigated walking ability (i.e., minimum passable width and TUG), and then factors that widen the sway (i.e., action space) to the left and right (i.e., muscle tone, fall history, and perceptual judgment). We found that in individuals with stroke, the collisions with an obstacle were related to the minimum passage and TUG speed. However, past fall history did not significantly associate with an increase in the number of collisions and remained a trend (Table 2). In this study, the minimum passable width of the opening was measured as an index of the action space, including sway during walking. When the width of the opening was larger than 1.1 times the shoulder width (cf., for a 40-cm shoulder width, the left and right space margins must be 2 cm or more each), there were many collisions. Gait in individuals with stroke has a large lateral displacement in both the stance and swing phases, particularly in the stance phase (do Carmo et al., 2015; Kao et al., 2014). This is because of weakness of the gluteus medius muscle and desensitization of the lower limbs on the paretic side (Dean and Kautz, 2015). Thus, individuals with stroke need a wider action space than healthy individuals for obstacles at the frontal plane. However, hypertonia of the upper and lower limbs did not directly affect the collision because even if it is difficult to control muscle tone voluntarily, it is possible to avoid obstacles if such individuals have good recognition of their physical condition. Therefore, it is highly possible that the left-right sway was connected to the collisions; individuals not fully aware of their own sway during walking could not measure and time their rotation in the walking-through-an-opening task.

The ability to judge passability was measured as a cognitive aspect (whether one’s own physical ability could be recognized), but it was not a factor in collisions. The reason may be that judgment ability in a static standing position and judgment ability in a dynamic scene, such as walking, may differ. Hackney and Cinelli (2013) reported that static perception was not much different between the elderly and the young, but dynamic perception showed that the elderly had a large margin in consideration of their left-right sway. Considering the findings of Muroi, Saito, Koyake, Higo et al. (2022), who reported unexpected collisions in individuals with stroke, dynamic perception may be reduced. Thus, thinking about how to avoid obstacles while walking becomes a dual task, which may have made decision making more difficult for individuals with stroke leading to reduced walking automaticity. Another possibility as to why cognitive factors were not affected is that most participants had lesions of the thalamus, putamen, and crown of radiation, with no impairment of the dorsal pathways (see Supplemental material). It has been proposed that the dorsal visual pathway is important for actions related to visual objects (such as obstacle) based on coordinate information obtained by spatial visual perception (Milner and Goodale, 1995). Cognitive factors may not have contributed to the conflict because most of the participants in this study had this pathway intact. Our findings indicate that it may be necessary to improve walking ability in individuals with stroke to avoid collisions with obstacles. This would include standing and turning movements (such as those in the TUG test) and acquiring a walking pattern with minimal sway on the frontal plane.

4.2 Factors that make it difficult to enter from the paretic side

Previous studies have shown that entering an opening from the paretic side was useful for reducing the collision rate (Muroi, Hiroi, et al., 2017; Muroi, Saito, Koyake, Higo, et al., 2022). In this study, we examined factors for the lesion (site and left or right), the function of the lower limbs on the paretic side, and the influence of perceptual judgment on passability. Three factors were found to be associated with the conditions that inhibited entering from the paretic side (conditions for entering from the non-paretic side): right hemisphere injury (left motor paralysis), thalamic lesions, and lower limb BRS stage 3.

First, right hemisphere injury that causes “Rightward Orienting Bias” may lead to entering from the non-paretic (right) side while walking-through-an-opening, even in the absence of manifested higher brain dysfunction. The right hemisphere is involved in attention, body image, and spatial cognition, and disorders of the right hemisphere are prone to falls as action errors (Rapport et al., 1993; Wada et al., 2007). Action errors are reported to occur on the left side, particularly in obstacle avoidance situations, and their detection is more sensitive than the assessment of unilateral spatial neglect (USN) on paper (Webster et al., 1989). This phenomenon is described as “Rightward Orienting Bias” (attention to the right side is biased even if there is no special neglect). This study excluded participants with USN in the assessment on paper, but some participants may have had biased attention to the right and hence, difficulty paying attention to the left. Potential USN during motion may have been detected by passing through a narrow opening. In short, if there is a network failure due to a right hemisphere injury, recognition and attention to the paretic side may be impaired, and it is possible that movements predominant on the non-paretic side repeatedly fail.

Second, thalamic lesions may also contribute to “Rightward Orienting Bias.” The thalamus aggregates sensory information and contributes to the formation of a body schema (Massion, 1992). The thalamus is also involved in the neural network of spatial attention (Corbetta and Shulman, 2011); damage to the right hemisphere causes imbalance in the left and right dorsal attention systems. Thus, spatial factors, such as eye movements and attention, shift to the right space. Although a previous study (Corbetta and Shulman, 2011) explained the USN mechanism, this result may indicate that “Rightward Orienting Bias” can be observed without the USN. Further research is needed to elucidate the mechanism of the behavioral characteristics of thalamic lesions.

Third, the degree of motor paralysis of the lower limbs influences the determination of entry direction. Lower limb BRS stage 3 is less supportive and is a segregation movement due to motor paralysis (Brunnstrom, 1966). Multiple regression analysis has previously been performed to determine whether individuals with stroke stepped on the paretic or non-paretic sides (Mansfield et al., 2012). The degree of motor paralysis in the lower limb and the amount of load on the paretic side are related to the paretic side step. Thus, decreasing the load on the paretic side tends to be predominant on the non-paretic side. In this study, it is possible that individuals with stroke who had moderate paralysis (lower limb BRS stage 3) showed predominantly obstructive avoidance strategies on the non-paretic side due to decreased supportability on the paretic side and difficulty in separation movement.

Contrary to our hypothesis, the ability to judge passability did not affect collision or direction of entry. The width that can be passed without body rotation was measured in the image; however, participants rotated and passed through most of the opening widths. Thus, image and actual behavior may have different judgment processes. In the future, it will be necessary to examine its relationship with the image ability to determine the direction of entry.

We performed a hypothesis verification-type multiple regression analysis based on the number of participants, but the variables used in this study were limited. The minimum passable width of walking-through the opening was measured as an index of sway during walking, but it is not clear whether it accurately indicated the left-right sway or load deviation during walking. Therefore, in the future, it may be necessary to investigate the load ratio and lateral sway during normal walking and include them as dependent variables.