Effects of uni- and multimodal cueing on handrail grasping and associated gaze behavior in older adults

Accident Analysis and Prevention 59 (2013) 407–414

Contents lists available at ScienceDirect

Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap

Effects of uni- and multimodal cueing on handrail grasping and associated gaze behavior in older adults夽,夽夽 Sandra M. McKay a,b , Julia E. Fraser a,b , Brian E. Maki a,b,c,d,e,∗ a

Toronto Rehabilitation Institute (University Health Network), Canada Centre for Studies in Aging, Sunnybrook Health Sciences Centre, Canada Institute of Medical Science, University of Toronto, Canada d Institute of Biomaterials and Biomedical Engineering, University of Toronto, Canada e Department of Surgery, University of Toronto, Canada b c

a r t i c l e

i n f o

Article history: Received 1 May 2013 Received in revised form 21 June 2013 Accepted 25 June 2013 Keywords: Aging Falls prevention Postural balance Reach-to-grasp reactions Stair safety Visual attention

a b s t r a c t Introduction: It appears that age-related changes in visual attention may impair ability to acquire the visuospatial information needed to grasp a handrail effectively in response to sudden loss of balance. This, in turn, may increase risk of falling. To counter this problem, we developed a proximity-triggered cueing system that provides a visual cue (flashing lights) and/or verbal cue (“attention use the handrail”) to attract attention to the handrail. This study examined the effect of handrail cueing on grasping of the rail and associated gaze behavior in a large cohort (n = 160) of independent and ambulatory older adults (age 64–80). Methods: The handrail and cueing system was mounted on a large (2 m × 6 m) motion platform configured to simulate a real-life environment. Subjects performed a daily-life task that required walking to the end of the platform, which was triggered to perturb balance by moving suddenly when they were adjacent to the rail. To prevent adaptation, each subject performed only one trial, and a deception was used to ensure that the perturbation was truly unexpected. Each subject was assigned to one of four cue conditions: visual, verbal, multimodal (visual-plus-verbal) or no cue. Results: Verbal cueing attracted overt visual attention to the handrail and markedly increased proactive grasping (prior to the onset of the balance perturbation) particularly when delivered unimodally. Subjects were otherwise much more likely to grasp the rail in reaction to the perturbation. A possible trend for visual cueing to improve the accuracy of these reactions was offset by adverse effects on reaction speed and on frequency of proactive grasping. Conclusions: The results support the viability of using unimodal verbal cueing to reduce fall risk by increasing proactive handrail use. Conversely, they do not strongly support use of visual cueing (either alone or in combination with verbal cueing) and suggest that it may even have adverse effects. Further study is needed to evaluate effects of handrail cueing in a wide range of populations and real-life settings. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction Balance-recovery reactions that involve rapid reaching movements to touch or grasp a handrail for support can play a critical role in preventing falls, particularly in older adults who tend to be more dependent than younger persons in using the arms to respond to

夽 Data collected at Centre for Studies in Aging, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5. 夽夽 Presented in preliminary form at the 20th International Conference on Posture and Gait Research (Trondheim, Norway; June, 2012). ∗ Corresponding author at: Toronto Rehabilitation Institute, Rm 12-121, 550 University Avenue, Toronto, Ontario, Canada M5G 2A2. Tel.: +1 416 597 3422x7808. E-mail address: [email protected] (B.E. Maki). 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.06.031

sudden loss of balance (Maki and McIlroy, 2005). Volitional reaching movements to grasp or touch an object such as a handrail are often guided by eye movements that lead to fixation of the target in the central visual field (Abrams, 1992; Carnahan and Marteniuk, 1991; Land, 2006). However, for reaching reactions that are triggered by sudden unexpected or unpredictable loss of balance, it appears that the urgent need to react rapidly imposes temporal constraints that preclude use of eye movements to guide the reach, forcing instead a reliance on peripheral vision and/or visuospatial information that has been previously tracked and stored in working memory (Cheng et al., 2012a,b; Ghafouri et al., 2004; King et al., 2009, 2010, 2011). The need to monitor one’s surroundings so as to track the location of objects such as handrails suggests a critical role for the acquisition, processing and storage of visual information, involving

408

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

various aspects of visual attention, gaze control and spatial working memory. All of these aspects of visual processing are known to decline with aging (Kemps and Newson, 2006; Munoz et al., 1998; Salthouse, 1992) and there is evidence that such deficits can impair motor behavior in situations that require visual monitoring of the surroundings. For example, age-related decline in the ability to rapidly extract information from the peripheral visual field predicts increased risk of driving accidents (Owsley et al., 1998), and also correlates with reduced mobility (Owsley and Mcgwin, 2004). In another study, older adults at high risk of falling were found to exhibit different gaze behavior in comparison to low-risk subjects (Chapman and Hollands, 2006), suggesting that the strategy used to gather visuospatial information during walking may affect risk of falling. With regard to handrail grasping behavior, we recently found that older adults were highly dependent on using a handrail to recover balance (in reacting to an unexpected balance perturbation while ambulating in an unfamiliar environment), but commonly failed to direct overt visual attention to the rail (King et al., 2009). In contrast, the majority of young adults fixated on the rail one or more times upon entering the unfamiliar environment (King et al., 2009, 2011). To help counter age-related problems in acquiring the visuospatial information (VSI) needed to guide effective handrail grasping behavior, we developed a proximity-triggered handrail cueing system that provides a visual cue (flashing lights) and/or verbal cue (“attention, use the handrail”) so as to attract attention to the handrail as the person approaches (Scovil et al., 2007). The cueing is intended to automatically draw attention to the handrail, and thereby compensate for age-related deficits in visual attention that might otherwise cause a failure to detect the presence of the handrail or to map its location accurately. In doing so, we anticipated that the cueing would improve ability to rapidly and accurately reach to grasp the handrail for support in response to sudden loss of balance. We also anticipated that the cueing might help to avoid age-related problems in executing effective reach-to-grasp reactions by increasing the tendency to hold the rail proactively, before loss of balance occurs. To evaluate the effects of the cueing, the handrail system was mounted on a large (2 m × 6 m) motion platform configured to simulate a “real-life” environment, and older-adult subjects performed an activity that required walking to the end of the platform, which was triggered to perturb balance by moving suddenly and

Fig. 1. Conceptual drawing (A) of the handrail cueing system and photographs of the railing portion of the system (B and C). During visual cueing, green light-emitting diodes (LEDs) mounted within the translucent black railing are controlled to suddenly start to flash rapidly on (B) and off (C). During verbal cueing, audio speakers deliver a verbal prompt (“attention, use the handrail”). The proximity sensors (photocells) that trigger the cueing are mounted on the wall near floor level, ∼1.5 m from the end of the railing. Adapted from Scovil et al. (2007).

unexpectedly when they were adjacent to the rail. A deception was used to ensure that the perturbation was truly unexpected. To prevent adaptation, subjects performed only one trial, which was their very first exposure to the perturbation and environment. Each of 160 subjects was assigned to one of four cueing conditions: (1) no cue, (2) visual cue; (3) verbal cue; or (4) multimodal (visual-plusverbal) cue. We hypothesized that the primary effect of the visual cueing would be improvement in the speed, accuracy and effectiveness of the perturbation-evoked reach-to-grasp reactions, whereas the primary effect of the verbal cueing would be an increased tendency to hold the rail proactively. In view of evidence that multimodal cueing is more effective than unimodal cueing (Laurienti et al., 2006), we hypothesized that the combination of visual and verbal cueing would enhance both of these benefits. Data from a small subset of the current sample (12 of the no-cue subjects) have been reported previously in a study of age-related differences in handrail grasping behavior (King et al., 2009). 2. Methods 2.1. Handrail cueing system The handrail cueing system was developed in accordance with established principles of attentional control and optimal design of warning systems, and has been described in detail elsewhere (Scovil et al., 2007). Briefly, the system comprises: (1) a translucent plastic black railing; (2) a series of green light-emitting diodes (LEDs) mounted inside the railing (along the longitudinal axis); (3) an audio speaker mounted in close proximity to the railing; and (4) a photocell that triggers onset of visual and/or verbal cueing when a person approaches (∼2 s before the body is adjacent to the rail); see Fig. 1. For the visual cueing, the photocell triggers the LEDs to suddenly begin to flash at a frequency of 3 Hz, and these continue to flash for an interval of 3 s. For the verbal cueing, the photocell triggers immediate playback of a 1.5 s recorded message (“attention, use the handrail”) that is delivered twice in rapid succession in an urgent tone (within an interval of ∼3 s) by a female voice (sound level >15 dB above background noise). 2.2. Participants A cohort of 160 healthy older adults (41 males, 119 females) aged 64–80 (mean age 70, SD 4.6) participated in the study. None had participated in previous balance studies, and all were naïve to the present protocol. Volunteers were recruited via advertisements (placed in local newspapers), posters (placed in stores, churches, apartment buildings and community centers) and word of mouth, and were asked to respond (over the telephone) to questions about their medical history, mobility level and handedness. The recruitment advertisements specified only that we were looking for volunteers potentially interested in participating in a research study and that they should be “65+ years of age, right-handed and generally healthy”. In addition to being right-handed, subjects were required to be able to stand and walk without aid and to understand English instructions. They were excluded from the study if they reported any: (1) neurological disorders; (2) eye disease or visual disorders; (3) vestibular or somatosensory disorders; (4) recurrent dizziness or unsteadiness; (5) use of medications that may affect balance; (6) musculoskeletal disorders or other medical conditions interfering significantly with daily activities; or (7) functional limitations of limb use. Visual acuity was tested in our laboratory, prior to starting the experiment. Subjects were required to have a minimum corrected Snellen visual acuity of 20/40 and were permitted to wear corrective lenses during the experiment. Only three

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

409

volunteers were excluded (one failed the vision test, one had a recent injury that compromised arm movements, and one had difficulty understanding English instructions). The protocol was approved by the institutional ethics review board and each subject provided written informed consent. The subjects were active and independent community-dwellers. All but ten subjects had Activities-specific Balance Confidence (ABC) scores of 80% or better (range 66–100%, data unavailable for one subject), and such scores are generally associated with a high level of function and activity (Myers et al., 1998). In addition, 133 of 156 subjects (85%) reported that they engaged in some form of physical activity for two hours or more per week (data unavailable for 4 subjects). These and other characteristics of the cohort are summarized in Table 1. 2.3. Assignment of subjects to cueing conditions The 160 subjects were each assigned to one of four handrail-cue conditions: (1) no cue; (2) visual cue; (3) verbal cue; (4) multimodal (visual-plus-verbal) cue. The original study design included only the no-cue, visual-cue and multimodal-cue conditions, and 40 subjects were assigned to each of these three conditions using a constrained randomization procedure (Scott et al., 2002) that minimized inter-group differences in key characteristics that could potentially confound the results. The characteristics that were controlled using the “minimization” procedure were: age, sex, recent history of falls (in the last year), Activities-specific Balance Confidence, working memory and attention, wearing of corrective lenses, amount of time per week engaged in physical exercise, and self-reported frequency of handrail use in daily life (“rarely”, “often” or “always”). The decision to add the fourth cueing condition (unimodal verbal cue) to the study was made after completing the testing of the other three cueing conditions. Recruitment to this group involved the first 40 consecutive new volunteers who met the study inclusion/exclusion criteria. Despite the fact that the minimization procedure could not be applied for the verbal-cue subjects, statistical comparison of the four groups of subjects indicated that there were no significant between-group differences in any of the minimization variables (p’s ≥ 0.15; Table 1). 2.4. Testing protocol The testing protocol used to evaluate the effects of the handrail cueing was identical to that used in previous young- and olderadult studies involving a conventional handrail, and has been described in detail previously (King et al., 2009, 2011). Balance-recovery reactions were evoked by sudden, unpredictable horizontal movement of a large (2 m × 6 m), computercontrolled, motor-driven motion platform (Maki et al., 2008; Maki and McIlroy, 2007; Scovil et al., 2007), which was configured to simulate a realistic living environment, including a door, a singlestep stair (leading up to a simulated office area), a handrail (located to the right of the stair) and various visual distracters (Fig. 2). A wall and door prevented the subject from viewing the environment prior to the start of the trial. A standardized script informed the subjects that there was a room behind the door, with an office area located at the far end of the room, and instructed them to open the door, enter the room, walk to the end at a normal pace and make a telephone call. This task thus required a visual search for the telephone while walking to the end of the platform. For safety, all subjects wore a harness attached (via a load cell) to a lowfriction overhead track that moved smoothly and did not impede the subject’s movements. The handrail and stair were mounted near the middle of the platform (near-end of rail 1.8 m from doorway, 1.5 m in front of stair riser). Sudden forward translation of the platform (square-wave

Fig. 2. Experimental set-up. Panel A is a schematic drawing of the large (2 m × 6 m) motion platform which was configured to simulate a “real-life” office environment. Subjects were instructed to enter the room and make a phone call, which required them to perform a visual search to find the phone while walking to the end of the platform in order to use the phone. The platform was triggered to suddenly and unexpectedly move forward when the subject stepped on the pressure mat adjacent to the handrail. Panel B is a photograph of the view that subjects saw upon opening the door at the start of the trial. Panel C displays an example output image from the head-mounted eye-tracker, with the point-of-gaze location (black dot) and associated gaze ellipses (corresponding to visual angles of 5◦ , 10◦ , etc.) superimposed on the video recorded by the head-mounted scene camera. In the displayed video frame, the point of gaze was directed at the computer monitor on the desk and a portion of the handrail was visible within a visual angle of 15◦ . Since no portion of the handrail was within a visual angle of 5◦ , this gaze fixation would not have been classified as a central-field handrail fixation.

acceleration/deceleration profile: amplitude 3.5 m/s2 , peak velocity 1.1 m/s, displacement 0.43 m, duration 0.6 s) was triggered to occur when the subject stepped on a pressure-sensitive mat adjacent to the handrail, thereby inducing a backward falling motion (similar to the effect of a slip). Objects mounted on the platform forced subjects to walk within a relatively narrow corridor (0.74 m wide) when approaching the stair, and thereby ensured that the handrail was well within reach when the perturbation was delivered. The rail was cylindrical, with a diameter (38 mm) and height (0.88 m above leading edge of stair tread) previously shown to allow effective grasping by persons encompassing a wide range of body heights and hand sizes (Maki et al., 1984, 1998, 2006, 2011). To avoid confounding effects of learning and adaptation, analysis was restricted to one trial per subject, which was the subject’s first exposure to the platform motion and to the simulated office environment. A deception was used to ensure that the perturbation was truly unexpected: subjects were told that the first trial

410

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

Table 1 Cohort descriptors. Values shown are means, with standard deviations in parentheses (unless otherwise indicated). The p-values are for the comparison between the four cue-condition groups (based on Fisher’s Exact Test for categorical data and Kruskal–Wallis non-parametric test for continuous measures). No cue Number of subjects Demographics and anthropometricsa Age (years)b Sex (men:women)b Height (m) Weight (kg) Falls (# of falls in last year)b , c Visual and cognitive function Visual acuity (Snellen)d Contrast sensitivity (dB)e Eyeglasses worn (# of subjects)b , f None Single focal Multifocal Contacts Visual processing speed (ms)g Visual divided attention (ms)g Visual selective attention (ms)g Mini mental state exam Working memory (#errors)b , i Simple hand reaction time (ms)h Activity and handrail use Balance confidence (% score)b , j Physical activity (minutes/week)b , k Handrail use (# of subjects)b , l Rarely Often Always

Visual cue

Verbal cue

Multimodal cue

p-value

40

40

40

40

70.0 (4.9) 9:31 1.65 (0.09) 68.9 (15.8) 0.18 (0.5)

70.2 (4.5) 3:37 1.66 (.01) 71.9 (14.7) 0.23 (0.5)

69.4 (4.5) 9:31 1.65 (0.08) 68.8 (12.2) 0.50 (0.9)

69.7 (4.5) 10:30 1.64 (0.09) 68.1 (15.1) 0.28 (0.6)

0.82 0.72 0.82 0.50 0.15

0.85 (0.27) 21 (1.8)

0.80 (0.25) 22 (1.7)

0.85 (0.28) 22 (1.7)

0.87 (0.25) 22 (1.4)

0.51 0.11 0.50

1 17 19 3 29 (28) 145 (121) 286 (127) 28.7 (1.3) 1.8 (2.0) 236 (39)

2 19 19 0 35 (36) 151 (135) 256 (114) 28.6 (1.5) 2.0 (2.1) 234 (40)

2 14 24 0 25 (16) 109 (107) 247 (94) 28.8 (1.2) 2.4 (2.9) 236 (32)

4 17 18 1 27 (22) 134 (128) 263 (119) 28.1 (1.8) 1.7 (1.8) 241 (52)

93.0 (7.5) 360 (244)

93.4 (8.0) 418 (297)

94.4 (6.4) 396 (350)

95.0 (6.3) 502 (434)

12 17 11

9 14 17

11 13 16

0.87 0.37 0.63 0.42 0.96 0.98 0.26 0.13 0.41

5 18 17

a

Missing data points: height and weight (n = 4); contrast sensitivity (n = 9); simple reaction time (n = 12); balance confidence (n = 1); physical activity (n = 6). Variables used to minimize inter-group differences during the randomized assignment of subjects to the no-cue, visual-cue and multimodal-cue groups. c A fall was defined as an event where the subject came to rest unintentionally on the ground, floor, or other lower level. d Expressed in Snellen decimal, for example 20/40 = 0.5. e Melbourne Edge Test, dB: decibels (ranging from 1 = poor vision to 24 = good vision) (Lord and Dayhew, 2001). f The number reported is the number of subjects who reported wearing ‘no glasses’, ‘single focal’ (reading or distance lenses), multifocal lenses (bifocal and trifocal lenses), or ‘contacts’. g Useful field of view, subtests 1–3 (processing speed, divided attention, selective attention) administered (Ball and Owsley, 1993). h Participant instructed to respond as quickly as possible, with a button press, to the onset of a red light (Lord and Dayhew, 2001). i Backward digit recall, participant asked to repeat a sequence of 5 digits in reverse order (Maylor and Wing, 1996). j Activity-specific Balance Confidence scale (Powell and Myers, 1995). k Self-reported total number of minutes engaged in mild, moderate or strenuous physical activity in a typical week. l Self reported frequency of handrail use in daily life. The number reported is the number of subjects who responded as ‘rarely’, ‘often’ or ‘always’. b

was a “practice trial” to help them become accustomed to the testing procedure and that the platform would not move during this trial. The effectiveness of the deception was confirmed by querying the subjects after the trial. 2.5. Data collection and analysis Four high-resolution 60-Hz video cameras (mounted overhead and above the handrail) were used, in conjunction with reflective markers placed on the arms and hands, to determine whether the handrail was grasped or touched, and if so, the timing of reach onset and initial rail contact. We also used the video to characterize whether an overt grasping error occurred (collision between the back of the forearm, wrist or hand and the medial surface of the handrail), whether a full “power” grip was achieved (thumb and fingers all wrapped around the rail), and whether balance was recovered successfully (i.e. without falling into the safety harness). The onset-timing of the perturbation-evoked reach-to-grasp reactions was derived from surface electromyographic (EMG) recordings from the right medial-deltoid and biceps-brachii muscles (band-pass filtered, 10–500 Hz; sampled at 1000 Hz). EMG onset was determined by a computer algorithm (McIlroy and Maki, 1993) and confirmed by visual inspection. A binocular head-mounted eye-tracker recorded eye movements and gaze direction (ASL model 501; Bedford, MA, USA). The eye tracker uses infrared corneal reflections to determine gaze

direction, relative to the head, and superimposes the point-ofgaze on 60-Hz video images recorded by a forward-facing “scene camera” mounted rigidly on the head. Custom-designed software (Scovil et al., 2009) was then used to superimpose “gaze ellipses” (corresponding to visual angles of 5◦ , 10◦ , etc., in relation to the point-of-gaze) on each frame of the scene-camera video (Fig. 2c). Based on these images, central-field fixation of the handrail was deemed to have occurred if: (1) some portion of the rail was within a 5◦ visual angle of the point-of-gaze, and (2) this gaze location was maintained for at least 100 ms [as per previous studies, e.g. King et al. (2011), Patla and Vickers (1997) and Vickers (1996)]. Timing of all arm-movement, EMG and gaze-fixation data was defined relative to onset of platform acceleration (>0.1 m/s2 , as measured by an accelerometer mounted on the platform), so as to distinguish events occurring before or after perturbation onset. Analyses included any grasping behavior or gaze fixations occurring over a time interval of approximately 5 s, beginning when the subject triggered a photocell by opening the door (at the entrance to the motion platform) and ending 1.5 s after the onset of the platform acceleration. The primary statistical analyses involved using the nonparametric Fisher’s Exact Test to evaluate associations between cueing condition and: (1) proactive grasping (number of subjects who reached to grasp or touch the rail prior to the onset of the balance perturbation); (2) reactive grasping (number of subjects who grasped or touched the rail in reaction to the balance perturbation);

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

Fig. 3. Effect of cueing on grasping behavior. Percentage of subjects who grasped or touched the handrail either: (A) proactively (prior to onset of the balance perturbation) or (B) reactively (after perturbation-onset). All statistically significant differences (p < 0.05) between cueing conditions are indicated in each panel (along with one near-significant difference, p = 0.055; the only other non-significant pairwise comparison involved post-perturbation grasping, no-cue vs visual-cue, p > 0.99). Note the much higher frequency of proactive grasping in the verbal-cue subjects, and the tendency for this to be somewhat less pronounced in the multimodal (combined verbal-plus-visual) subjects.

and (3) gaze behavior (number of subjects who fixated on the rail one or more times before or after perturbation onset). In addition, for trials in which the subject grasped or touched the rail in reaction to the balance perturbation, we used Fisher’s Exact Test to evaluate associations between cueing condition and the number of trials in which a full power-grip was achieved or a collision error occurred, and we used the Kruskal–Wallis non-parametric test to evaluate cueing-related differences in the timing of the perturbation-evoked reach reactions (EMG onset latency and time to rail contact). All of the above analyses were performed using SAS statistical software (Version 9.2; SAS Institute, Inc.; Cary, NC, USA). 3. Results The frequency data used in the analyses of grasping and gaze behavior are shown in Table 2. As detailed in this table, the vast majority of subjects (89%, 142/160) reached to grasp or touch the handrail, initiating the reach either before (41%, 65/160) or after (48%, 77/160) the onset of the balance perturbation, and were almost always able to recover balance successfully (an overt fall into the safety harness occurred in only two trials: one verbal-cue and one multimodal-cue). 3.1. Grasping behavior Verbal cueing markedly increased the tendency to grasp the rail proactively (prior to perturbation onset): 93% (37/40) of verbalcue and 58% (23/40) of multimodal-cue subjects, vs 0% (0/40) of visual-cue and 13% (5/40) of no-cue subjects; p’s < 0.0001; Fig. 3a. Surprisingly, the combination of verbal and visual cues in the multimodal-cue trials was actually less effective than verbal cueing alone in promoting proactive handrail use [93% (37/40) vs 58% (23/40); p = 0.0005]. Consequently, multimodal-cue subjects were instead more likely to rely on reactive balance control, grasping in reaction to the balance perturbation in 30% (12/40) of cases vs 5% (2/40) of verbal-cue subjects (p = 0.0064; Fig. 3b). In the absence of verbal cueing, subjects were even more likely to rely on reactive balance control, grasping the rail in reaction to the perturbation in 78% (31/40) of visual-cue trials and 80% (32/40) of no-cue trials, vs 5% (2/40) of verbal-cue trials and 30% (12/40) of

411

Fig. 4. Effect of cueing on accuracy of perturbation-evoked reach-to-grasp reactions. Percentage of reactions exhibiting: (A) collision errors, or (B) attainment of a full “power” grip. Note the apparent (but not statistically significant, p’s ≥ 0.31) trends for less frequent collisions and more frequent power grips in the visual-cue subjects (data are not shown for verbal- and multimodal-cue subjects, due to the low frequency of perturbation-evoked grasp reactions).

multimodal-cue trials (p’s < 0.0001; Fig. 3b). Such reactive grasping was equally frequent in visual-cue and no-cue subjects [78% (31/40) vs 80% (32/40); p = 1.00; Fig. 3b]. A small number of nocue subjects (n = 5) did instead grasp proactively, but there was a surprising near-significant tendency for unimodal visual cueing to inhibit this type of proactive behavior [0% (0/40) of subjects vs 13% (5/40) for no-cue; p = 0.055; Fig. 3a]. 3.2. Perturbation-evoked grasping reactions Due to the infrequency of reactive responses in the verbal-cue (n = 2) and multimodal-cue (n = 12) subjects, the analyses of the timing and accuracy of the perturbation-evoked reach-to-grasp reactions were limited to a comparison of the visual-cue (n = 31) and no-cue (n = 32) conditions. These analyses suggest that visual cueing had mixed effects. In terms of timing, the visual-cueing actually appeared to have some adverse effects, leading to a small but significant increase in reaction onset latency (earliest onset in biceps or deltoid: 229 ms vs 213 ms for no-cue; p = 0.015; Fig. 5a). There was also a similar delay in rail contact; however, this was not statistically significant (636 ms vs 618 ms for no-cue; p = 0.50; Fig. 5b). Conversely, there appeared to be a possible trend for visual cueing to improve ability to grasp the handrail accurately. Collision errors (where the initial rail contact involved the back of the hand, wrist or forearm), occurred less frequently in visual-cue subjects, in comparison to no-cue subjects [10% (3/31) of subjects vs 19% (6/32) for no-cue; p = 0.47; Fig. 4a], and the visual-cue subjects were instead more likely to achieve a full power grip during the initial rail contact [48% (15/31) vs 34% (11/32); p = 0.31; Fig. 4b]; however, neither of these trends was statistically significant. 3.3. Gaze behavior Gaze data were available for 105 of the 160 subjects (technical problems with the eye tracker precluded analysis of gaze in 55 cases). As detailed in Table 2, 14 subjects only fixated on the handrail (one or more times) prior to the onset of the balance perturbation and 15 only fixated on the rail after perturbation onset. In 10 cases, the subject fixated one or more times on the rail before perturbation onset and also initiated one or more new rail fixations after perturbation onset. Thus, a total of 24 subjects (14 + 10)

412

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

Table 2 Frequency of grasping and gaze fixation of the handrail. Numbers of subjects who used the handrail before or after perturbation onset (PO). Reached pre-PO (proactive)

Reached post-PO (reactive)

No-cue Visual-only Verbal-only Multi-modal

5 0 37 23

32 31 2 12

Did not reach 3 9 1 5

Total 40 40 40 40

Total

65

77

18

160

Numbers of subjects who visually fixated on the handrail (one or more times) before and/or after perturbation onset (PO), shown separately for: all available trials (A) and for subsets of trials involving proactive grasping (P), reactive grasping (R) or no grasping (N).a Fixated pre-PO only A No-cue Visual-only Verbal-only Multi-modal Total a

P

Fixated post-PO only A

Fixated both pre- and post-PONo handrail fixations A

P

Total

R

N

P

R

N

P

R

N

A

R

N

P

R

N

1 1 5 7

0 0 5 6

1 1 0 1

0 0 0 0

4 2 4 5

1 0 3 3

3 2 1 2

0 0 0 0

0 1 1 8

0 0 1 5

0 1 0 3

0 0 0 0

27 19 8 12

3 0 8 5

23 16 0 4

1 3 0 3

A 32 23 18 32

4 0 17 19

27 20 1 10

1 3 0 3

14

11

3

0

15

7

8

0

10

6

4

0

66

16

43

7

105a

40

58

7

Note: 105 of 160 subjects had useable gaze data (55 subjects missing due to technical problems with the eye tracker).

fixated prior to perturbation onset and 25 (15 + 10) fixated after perturbation onset. Analysis of the available gaze data suggested that the verbal cueing was effective in attracting overt visual attention to the handrail, particularly when combined with the visual cueing (Fig. 6a). In these multimodal trials, 47% (15/32) of subjects fixated on the rail one or more times after entering the unfamiliar environment, prior to onset of the balance perturbation, in comparison to only 3% (1/32) of no-cue subjects (p < 0.0001) and 9% (2/23) of visual-cue subjects (p = 0.003). Unimodal verbal cueing also increased the tendency to look directly at the rail prior to the perturbation, although the effect was not quite as large [33% (6/18) vs 3% (1/32) for no-cue (p = 0.006) and 9% (2/23) for visual-cue (p = 0.11)]. In contrast, unimodal visual cueing had only a small effect, if any [9% (2/23) vs 3% (1/32) for no-cue (p = 0.57)]. Analysis of the visual fixations that occurred after the onset of the balance perturbation showed similar trends (Fig. 6b). Again, rail fixation was most likely to occur when there was multimodal cueing [41% (13/32) vs 13% (3/23) of visual-cue subjects (p = 0.036) and 13% (4/32) of no-cue subjects (p = 0.022)], unimodal verbal cueing showed a similar but less pronounced trend [28% (5/18) vs 13%

(3/23) for visual-cue (p = 0.27) and 13% (4/32) for no-cue (p = 0.25)], and unimodal visual cueing had no effect [13% (3/23) vs 13% (4/32) for no-cue; p = 1.00]. Visual handrail fixation occurred in the majority of the trials that involved proactive grasping [53% (9/17) for verbal-cue, 74% (14/19) for multimodal-cue], and these fixations often began prior to rail contact [67% (6/9) for verbal-cue, 79% (11/14) for multimodal-cue]. In contrast, for the trials that involved perturbation-evoked grasp reactions, visual fixation of the handrail occurred infrequently, either prior to perturbation onset [4% (1/27) of no-cue subjects, 10% (2/20) of visual-cue subjects] or after perturbation onset [11% (3/27) of no-cue subjects, 15% (3/20) of visual-cue subjects]. Furthermore, in all but one (no-cue) subject, the post-perturbation visual handrail fixations began subsequent to handrail contact, and hence occurred too late to aid in guiding the initial reaching movement.

Fig. 5. Effect of cueing on timing of perturbation-evoked reach-to-grasp reactions: (A) reaction time, and (B) handrail-contact time (means and standard deviations are shown). Note the statistically significant slowing in reaction-time in the visual-cue subjects, and the similar (but not statistically significant, p = 0.50) trend in contacttime (data are not shown for verbal- and multimodal-cue subjects, due to the low frequency of perturbation-evoked grasp reactions).

Fig. 6. Effect of cueing on gaze behavior. Percentage of subjects who fixated on the handrail (one or more times): (A) before the onset of the balance perturbation, and/or (B) after perturbation-onset. All statistically significant differences (p < 0.05) between cueing conditions are indicated in each panel (p’s ≥ 0.11 in all other pairwise comparisons). Note the higher frequency of handrail fixations in the verbal-cue and multimodal-cue (combined verbal-plus-visual) subjects.

4. Discussion The gaze data from the present study indicate that the verbal cueing was effective in attracting overt visual attention to the

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

handrail. Moreover, the grasping data indicate an even more pronounced behavioral effect, strongly supporting our hypothesis that verbal cueing would increase proactive handrail use. Indeed, over 90% of the subjects who were tested with the unimodal verbalcue initiated an arm movement to grasp the rail proactively (prior to the onset of the motion-platform balance perturbation), even though they were not expecting the platform to move. Such proactive grasping occurred in only 13% of the no-cue subjects and never occurred in the visual-cue subjects. For these latter two cueing conditions, the predominant grasping behavior (79% of subjects) was instead reactive in nature, i.e. a rapid reach-to-grasp balancerecovery reaction evoked in response to the sudden unexpected platform motion. The collision-error and power-grip data suggest a tendency for visual cueing to increase the accuracy of perturbation-evoked reach-to-grasp reactions; however, these trends were not statistically significant. The failure to show more pronounced effects could be related to the lack of an effect on gaze behavior. Previously, we showed that older adults were much less likely than young adults to look at a handrail upon entering an unfamiliar environment (King et al., 2009, 2011). The visual cueing was intended to counter this failure to direct overt visual attention to the handrail and thereby provide high-acuity (central-field) spatial information that would enhance the accuracy of any subsequent grasp reaction; however, the present results showed no effect of visual cueing on overt handrail fixation. The apparent (non-significant) improvements in grasp accuracy that were observed in visual-cue trials could instead reflect an improved ability to locate the handrail in the peripheral visual field. The potential benefits of visual cueing in improving graspreaction accuracy were offset by two adverse effects: (1) reduction in the speed of the perturbation-evoked reactions (time required for initiation and completion), and (2) reduction in frequency of proactive grasping. The slowing of the reaction initiation could potentially be attributed to a distraction effect (Quant et al., 2000), whereas the reduction in proactive handrail use could possibly reflect confusion regarding the interpretation of the flashing lights. Indeed, in previous pilot tests, a small number of subjects reported that they interpreted the flashing lights as a warning not to touch the handrail (Scovil et al., 2007). Those pilot tests involved yellow lights, which may have “overlearned” associations with warning systems, and it was for this reason that we elected to use green lights in the present study; however, it is still possible that the novelty of the flashing lights was confusing to some of the participants. In fact, it is even possible that the lights caused some participants to perceive that the object was not a handrail. Such confusion might also explain why the multimodal verbalplus-visual cueing was less effective in eliciting proactive handrail use, in comparison to unimodal verbal cueing. In contrast to the present findings, previous studies have suggested that congruent multimodal stimuli are more effective in influencing behavior of older adults, compared to stimuli that involve a single sensory modality (Laurienti et al., 2006). The present gaze data do, in fact, suggest a trend for multimodal cueing to be more effective than unimodal visual cueing in attracting visual attention to the handrail, as would be expected. Hence, the failure of multimodal cueing to enhance the beneficial effect of unimodal verbal cueing on proactive grasping is likely attributable to the lack of clear congruency in the meaning of the verbal and visual cues, rather than a failure of the multimodal cueing to facilitate attention capture. It is noteworthy that the vast majority (∼90%) of the perturbation-evoking grasping reactions, as well as a sizeable proportion (∼50%), of the proactive grasps, were executed without any prior visual fixation of the handrail. This indicates an important role for peripheral vision, both in identifying the presence of handrails in one’s immediate surroundings and in helping to guide the motion

413

of the hand to the rail. Although our subjects were typically able to recover balance successfully, a tendency to rely on peripheral vision in responding to sudden loss of balance could potentially lead to falls in older adults who are less able to rapidly process peripheral visual information. We are currently investigating the use of computer-based training of rapid visual processing as one possible way to address this problem (Maki et al., 2011). The present results suggest that use of unimodal verbal cueing to increase proactive handrail use may be a viable approach to reducing risk of falls. By enhancing both mechanical stabilization of the body (Maki et al., 1984) and sensory (“haptic”) feedback about the body position and movement (Jeka, 1997), proactive handrail use reduces the likelihood that loss of balance will occur and improves the ability to recover equilibrium in the event that loss of balance does occur. However, it remains to be determined whether such cueing would be equally effective in a “real-life” environment, particularly under conditions where persons are repeatedly exposed to the verbal prompt. It seems possible that some people may begin to disregard the prompt after repeated exposure, and in fact, could begin to find the prompt to be an irritant. Perhaps, the most productive application of this cueing technique would be in public settings where the environment is likely to be unfamiliar to a large proportion of person, and/or settings where the person is likely to be distracted from using the handrail. Use of the verbal handrail cueing in such settings would be analogous to the proximity-triggered verbal prompts that are currently used to warn users of “moving sidewalks” in airports that they are nearing the end of the moving sidewalk. Verbal cueing could also be useful in residential or institutional settings that are populated by persons with cognitive deficits that might lead to a tendency to forget to use the handrail. In settings where repeated exposure to the verbal cueing might become an irritant for some people, the cueing system could be modified so that it is only activated for certain “high-risk” individuals [e.g. via radio-frequency identification (RDIF) tags that could be worn by such individuals]. A potential limitation of this study pertains to the fact that the participants were all generally healthy and independent community-dwellers. We elected to focus this initial study on such a cohort in view of epidemiological evidence which suggests that even relatively healthy older adults may be at elevated risk of experiencing serious injuries due to falls on stairs (Startzell et al., 2000). However, it is clear that further research is needed to study the effectiveness of the cueing in persons with more significant cognitive and neuromusculoskeletal impairments due to specific diseases and disorders. Another potential limitation arises because the verbal-cue condition was added to the study after completing the randomized assignment of subjects for the other three cueing conditions. As a consequence, the verbal-cue subjects were recruited consecutively. However, the recruitment advertisements and locations were the same as for the other groups; hence, there was no reason to expect that the verbal-cue subjects would differ in any systematic way, and the comparison of the characteristics of the four groups supported the absence of any inter-group differences. Finally, it should be noted that the conclusions of this study are dependent on the specific manner in which the verbal and visual cueing was implemented. Although the details of the current implementation were based on the existing literature regarding optimal attention-capture and hazard-warning systems (Scovil et al., 2007), we cannot rule out the possibility that other implementations could be more effective.

5. Conclusion The present results support the viability of using unimodal verbal cueing to reduce fall risk by increasing proactive handrail use.

414

S.M. McKay et al. / Accident Analysis and Prevention 59 (2013) 407–414

Conversely, the results do not strongly support the use of visual cueing, either alone or in combination with verbal cueing. Although it appears that visual cueing may have some benefit in improving the accuracy of perturbation-evoked grasp reactions, it also had adverse consequences, namely, delays in the timing of these reactions and reduction in proactive handrail use. Further research is needed to determine the underlying reason for these adverse effects, but it seems likely that the unfamiliarity of the visual cueing may have distracted and/or confused some subjects. Further research is also needed to address effects of verbal and multimodal cueing on perturbation-evoked grasp reactions, which could not be addressed in the present study due to the predominating tendency to grasp proactively in response to the verbal cue. Ultimately, there is a need to evaluate the effectiveness of the handrail cueing systems in a wide range of populations and in a variety of “reallife” settings. Such studies could take advantage of recent efforts to develop inexpensive, automated video systems to detect handrail use, missteps and falls in daily life (Snoek et al., 2009). Acknowledgments This work was supported by operating grant #MOP-13355 and Mobility In Aging Teams grant #MAT-91865 from the Canadian Institutes of Health Research (CIHR), and by a summer internship award from the Ontario Neurotrauma Foundation. In addition, SMM was supported, in part, by a CIHR Strategic Training Post-Doctoral Fellowship in Health Care, Technology and Place. Infrastructure support was provided by the Sunnybrook Research Institute and by the Toronto Rehabilitation Institute, where equipment and space have been funded with grants from the Canada Foundation for Innovation, Ontario Innovation Trust and the Ministry of Research and Innovation. The authors thank Dr. Jay Pratt for his expert advice, and gratefully acknowledge assistance provided by: Areeba Adnan, Kenneth Cheng, Simon Jones, Emily King, Tracy Lee, Adrian Liggins, Marlene Luis, Avril Mansfield, Aaron Marquis, Stephanie Middleton, Amy Peters and Carol Scovil. References Abrams, R.A., 1992. Coordination of eye and hand for aimed limb movements. In: Proteau, L., Elliot, D. (Eds.), Vision and Motor Control. Elsevier, Amsterdam, pp. 129–152. Ball, K., Owsley, C., 1993. The useful field of view test: a new technique for evaluating age-related declines in visual function. Journal of the American Optometric Association 64, 71–79. Carnahan, H., Marteniuk, R.G., 1991. The temporal organization of hand, eye, and head movements during reaching and pointing. Journal of Motor Behavior 23 (2), 109–119. Chapman, G.J., Hollands, M.A., 2006. Evidence for a link between changes to gaze behaviour and risk of falling in older adults during adaptive locomotion. Gait and Posture 24, 288–294. Cheng, K.C., McKay, S.M., King, E.C., Maki, B.E., 2012a. Does aging impair the capacity to use stored visuospatial information or online visual control to guide reachto-grasp reactions evoked by unpredictable balance perturbation? Journal of Gerontology 67, 1238–1245. Cheng, K.C., McKay, S.M., King, E.C., Maki, B.E., 2012b. Reaching to recover balance in unpredictable circumstances: is online visual control of the reach-to-grasp reaction necessary or sufficient? Experimental Brain Research 218, 589–599. Ghafouri, M., McIlroy, W.E., Maki, B.E., 2004. Initiation of rapid reach-and-grasp balance reactions: is a pre-formed visuospatial map used in controlling the initial arm trajectory? Experimental Brain Research 155, 532–536. Jeka, J.J., 1997. Light touch contact as a balance aid. Physical Therapy 77, 476–487. Kemps, E., Newson, R., 2006. Comparison of adult age differences in verbal and visuo-spatial memory: the importance of ‘pure’, parallel and validated measures. Journal of Clinical and Experimental Neuropsychology 28 (3), 341–356. King, E.C., Lee, T.A., McKay, S.M., Scovil, C.Y., Peters, A.L., Pratt, J., Maki, B.E., 2011. Does the “eyes lead the hand” principle apply to reach-to-grasp movements

evoked by unexpected balance perturbation? Human Movement Science 30, 368–383. King, E.C., McKay, S.M., Cheng, K.C., Maki, B.E., 2010. The use of peripheral vision to guide perturbation-evoked reach-to-grasp balance-recovery reactions. Experimental Brain Research 207, 105–118. King, E.C., McKay, S.M., Lee, T.A., Scovil, C.Y., Peters, A.L., Maki, B.E., 2009. Gaze behavior of older adults in responding to unexpected loss of balance while walking in an unfamiliar environment: a pilot study. Journal of Optometry 2, 119–126. Land, M.F., 2006. Eye movements and the control of actions in everyday life. Progress in Retinal and Eye Research 25 (3), 296–324. Laurienti, P.J., Burdette, J.H., Maldjian, J.A., Wallace, M.T., 2006. Enhanced multisensory integration in older adults. Neurobiology of Aging 27 (8), 1155–1163. Lord, S.R., Dayhew, J., 2001. Visual risk factors for falls in older people. Journal of American Geriatric Society 49 (5), 508–515. Maki, B.E., Bartlett, S.A., Fernie, G.R., 1984. Influence of stairway handrail height on the ability to generate stabilizing forces and moments. Human Factors 26, 705–714. Maki, B.E., Cheng, K.C.C., Mansfield, A., Scovil, C.Y., Perry, S.D., Peters, A.L., McKay, S., Lee, T., Marquis, A., Corbeil, P., Fernie, G.R., Liu, B., McIlroy, W.E., 2008. Preventing falls in older adults: new interventions to promote more effective change-insupport balance reactions. Journal of Electromyography and Kinesiology 18 (2), 243–254. Maki, B.E., McIlroy, W.E., 2005. Change-in-support balance reactions in older persons: an emerging research area of clinical importance. Neurologic Clinics 23 (3), 751–783. Maki, B.E., McIlroy, W.E., 2007. Cognitive demands and cortical control of human balance-recovery reactions. Journal of Neural Transmission 114, 1279–1296. Maki, B.E., Perry, S.D., McIlroy, W.E., 1998. Efficacy of handrails in preventing stairway falls: a new experimental approach. Safety Science 28 (3), 189–206. Maki, B.E., Perry, S.D., Scovil, C.Y., Mihailidis, A., Fernie, G.R., 2006. Getting a grip on stairs: research to optimize effectiveness of handrails. In: Pikaar, R.N., Koningsveld, E.A.P., Settels, P.J.M. (Eds.), Proceedings IEA2006 Congress. Elsevier, Amsterdam, pp. 4669–4674 (article #778). Maki, B.E., Sibley, K.M., Jaglal, S.B., Bayley, M., Brooks, D., Fernie, G.R., Flint, A.J., Gage, W., Liu, B.A., McIlroy, W.E., Mihailidis, A., Perry, S.D., Popovic, M.R., Pratt, J., Zettel, J.L., 2011. Reducing fall risk by improving balance control: development, evaluation and knowledge-translation of new approaches. Journal of Safety Research 42, 473–485. Maylor, E.A., Wing, A.M., 1996. Age differences in postural stability are increased by additional cognitive demands. Journal of Gerontology 51, P143–P154. McIlroy, W.E., Maki, B.E., 1993. Changes in early ‘automatic’ postural responses associated with the prior-planning and execution of a compensatory step. Brain Research 631, 203–211. Munoz, D.P., Broughton, J.R., Goldring, J.E., Armstrong, I.T., 1998. Age-related performance of human subjects on saccadic eye movement tasks. Experimental Brain Research 121 (4), 391–400. Myers, A.M., Fletcher, P.C., Myers, A.H., Sherk, W., 1998. Descriminative and evaluative properties of the activities-specific balance confidence (ABC) scale. Journal of Gerontology 53A, M287–M294. Owsley, C., Ball, K., Mcgwin Jr., G., Sloane, M.E., Roenker, D.L., White, M.F., Overley, E.T., 1998. Visual processing impairment and risk of motor vehicle crash among older adults. Journal of the American Medical Association 279 (14), 1083–1088. Owsley, C., Mcgwin Jr., G., 2004. Association between visual attention and mobility in older adults. Journal of the American Geriatrics Society 52 (11), 1901–1906. Patla, A.E., Vickers, J.N., 1997. Where and when do we look as we approach and step over an obstacle in the travel path? NeuroReport 8, 3661–3665. Powell, L.E., Myers, A.M., 1995. The activities-specific balance confidence (ABC) scale. Journal of Gerontology 50A (1), M28–M34. Quant, S., McIlroy, W.E., Verrier, M.C., Maki, B.E., 2000. Effect of attention on compensatory upper-limb balance reactions. In: Proceedings of the Tri-Joint Congress 2000 (Canadian Associations for Physical Therapy, Occupational Therapy and Speech-Language Pathology and Audiology), Toronto, Canada. Salthouse, T.A., 1992. Reasoning and spatial abilities. In: Craik, F.I.M., Salthouse, T.A. (Eds.), The Handbook of Cognition and Aging. Lawrence Erlbaum Associates, Hillsdale, NJ, pp. 167–204. Scott, N.W., Mcpherson, G.C., Ramsay, C.R.M.K.C., 2002. The method of minimization for allocation to clinical trials: a review. Controlled Clinical Trials 23, 662–674. Scovil, C.Y., Corbeil, P., Lee, T.A., McKay, S.M., Peters, A.L., Maki, B.E., 2007. A novel handrail cueing system to prevent falls in older adults. Gerontechnology 6, 224–229. Scovil, C.Y., King, E.C., Maki, B.E., 2009. Determining the visual angle of objects in the visual field: an extended application of eye trackers. IEEE Transactions on Biomedical Engineering 56, 910–912. Snoek, J., Hoey, J., Stewart, L., Zemel, R., Mihailidis, A., 2009. Automated detection of unusual events on stairs. Journal of Image and Vision Computing 27, 153–166. Startzell, J.K., Owens, D.A., Mulfinger, L.M., Cavanagh, P.R., 2000. Stair negotiation in older people: a review. Journal of the American Geriatrics Society 48, 567–580. Vickers, J.N., 1996. Visual control when aiming at a far target. Journal of Experimental Psychology – Human Perception and Performance 22 (2), 342–354.