FALL-RELATED HIP FRACTURES contribute substantially. Age-Related Changes in Spatial and Temporal Gait Variables

Transcription

1 31 Age-Related Changes in Spatial and Temporal Gait Variables Penny C. Grabiner, BSN, S. Tina Biswas, BSE, Mark D. Grabiner, PhD ABSTRACT. Grabiner PC, Biswas ST, Grabiner MD. Agerelated changes in spatial and temporal gait variables. Arch Phys Med Rehabil 2001;82:31-5. Objective: To extend recent findings describing the effect of age on spatial and temporal gait variables. Design: Experimental. Setting: A gait analysis laboratory. Participants: Two experiments with healthy nonfallers were conducted. Experiment 1 included 33 subjects (n 15, yr; n 18, yr); and experiment 2 included 24 subjects (n 14, yr; n 10; yr). Interventions: The effect of age, walking velocity, shoe condition, and performance of an attention-splitting task on gait variables was investigated. Main Outcome Measures: Temporal and spatial gait variables were quantified using an instrumented surface across which subjects walked. The independent variables were walking velocity variability, stride length variability, stride width variability, and stride time variability. Results: Stride width variability of older adults was significantly larger than that of younger adults in both experiments. The remaining gait variables demonstrated nonsystematic or no age-related differences. Conclusions: With the exception of stride width variability, the variability of the remaining gait variables of interest were insensitive to the speed at which subjects walked, whether the subjects were wearing shoes or not, and performing an attention-splitting task while walking. These findings contribute to an emerging interpretive framework established by similar work published by others regarding gait variability. Key Words: Gait; Walking; Biomechanics; Rehabilitation by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation FALL-RELATED HIP FRACTURES contribute substantially to mobility loss, institutionalization, and death in older adults. Over 250,000 hip fractures occur annually in older adults in the United States and impose an approximately $8 billion burden to the health care system. 1 Fewer than 2% of falls by older adults result in a hip fracture. 2,3 However, more than 90% of hip fractures in older adults result from a fall. 4,5 Therefore, an important avenue to reducing the incidence of fall-related fractures is to identify older adults who are candi- From the Clinical Biomechanics and Rehabilitation Laboratory, Department of Biomedical Engineering, Lerner Research Institute, Cleveland Clinic Foundation, Cleveland, OH. Accepted in revised form January 24, Supported by the National Institutes of Health (grant no. RO1AG10557). No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Mark D. Grabiner, PhD, Dept of Biomedical Engineering, ND-2, Lerner Research Institute, 9500 Euclid Ave, Cleveland Clinic Foundation, Cleveland, OH, grabiner@bme.ri.ccf.org /01/ $35.00/0 doi: /apmr dates for injurious falls. Because most falls, and therefore fall-related hip fractures, occur during locomotion, 6 age-related changes in locomotion-specific variables have been hypothesized to contribute to falls in older adults. However, until recently, this hypothesis had not been tested directly. 7,8 Although Pavol et al 7,8 provided insights into the biomechanics of some types of falls and recovery efforts that could not otherwise be obtained, their protocol is not appropriate for clinical use. The variability of spatial and temporal measures of gait, which is appropriate for clinical use, has been the topic of several papers Recently, 2 research groups specifically analyzed the variability of stride kinematics in the context of falls by older adults. Increased variability of step time retrospectively distinguished older adults who had fallen from older adults who had not fallen. 12 Decreased variability of stride width prospectively predicted falls by older adults that occurred during locomotor activities. 13 The findings of these studies are compelling and warrant further study because of the scope of the social and medical problems associated with falls by older adults. The present study sought to extend the findings of Maki 13 and Hausdorff et al 12 to include a broader range of common locomotor conditions. Subjects in the Maki and Hausdorff studies walked at a self-selected velocity wearing either slippers or shoes, respectively. The purpose of our first of 2 experiments was to describe age-related differences in the variability of stride kinematics as a function of walking velocity and shoe condition. People commonly increase and decrease walking speed during daily activities. Similarly, it is not uncommon for people to walk without shoes, especially in the home. The ubiquity of these conditions served as the rationale for their study. The purpose of the second experiment was to characterize the effect of age on the extent to which performance of an attention-splitting task affected the variability of stride kinematics. Walking while performing an attention-splitting task has been associated with increased risk of falling by older adults. 14,15 In this study, we were particularly interested in those variables on which age had a significant effect. METHODS A total of 33 subjects participated in experiment 1 and 24 subjects participated in experiment 2 (table 1). The older subjects were healthy and independent community-dwellers who reported not having fallen during the year before participation. The older adults were screened for the presence of neuromuscular, musculoskeletal, and other systemic exclusion factors. Both groups of subjects were of comparable height and weight. The methods had been reviewed and approved by the institutional review board of the Cleveland Clinic Foundation and all subjects provided informed consent. The older adults were paid for their participation. Spatial and temporal stride kinematics were measured using a commercially available instrument a composed of a meter (12 2-ft) walkway that physically resembles a carpet. The walkway is made up of 6 contiguous instrumented pads. Each pad is composed of sensors arranged in a matrix. The center-to-center distance of the sensors is 1.27cm (0.5in). The data were digitized at 100Hz and stored on a

2 32 EFFECTS OF AGE ON GAIT VARIABILITY, Grabiner Age (yr) Height (cm) Weight (lb) Gender Table 1: Descriptive Characteristics of Participants Younger Adults (n 18) Experiment 1 Experiment 2 Older Adults (n 15) Younger Adults (n 10) Older Adults (n 14) women, 10 men 11 women, 4 men 5 women, 5 men Values presented as mean standard deviations. 8 women, 6 men personal computer. The spatial and temporal stride kinematic variables were extracted offline using the digitized data using GAITRite TM a and custom-written software. Experiment 1 The protocol for experiment 1 included 2 conditions during which the subjects walked with and without shoes. Both shoe conditions consisted of 5 trials performed at each of 3 walking speed conditions. The number of trials was selected to provide a minimum of approximately 20 steps per condition. This number of steps was consistent with that previously reported using similar methods. 13 The walking speed conditions and order of their administration were walking at normal speed, walking at slow speed, and walking at fast speed. For the normal speed trials, the subjects were instructed to walk at a self-selected and typical speed. Subjects attempted to repeat this speed during each of the 5 trials. For the slow and fast conditions, subjects were asked to select subjectively their own speed and to repeat this speed during the 5 trials in each condition. No feedback of the walking speed was provided to the subjects during the protocol. For the first half of the protocol, subjects walked in their own shoes. In the second half of the protocol, subjects repeated the same 3 speed conditions but without shoes. Each trial was initiated approximately 1.22 meters (4ft) before the beginning of the walkway to exclude effects of gait initiation on the measurement. The subjects were instructed to continue walking for several steps beyond the end of the walkway to exclude effects of gait termination. Experiment 2 The protocol for experiment 2 consisted of 2 conditions during which subjects wore their own shoes. In the first condition, the subjects were instructed to walk the length of the walkway at a self-selected speed. The second condition required the subjects to walk the length of the walkway while carrying an 8-ounce cup placed in a saucer. The cup was filled with water to within 3mm of the rim. The subjects were instructed to use 1 hand to hold the cup and saucer while walking and to not spill the water from the cup. The subjects were allowed to switch hands between trials. Several subjects asked if they should maintain the same pace they used during the first condition. The standard response was that the goal was not to spill the water. Each trial was initiated one to two meters from the beginning of the walkway, and ended one to two meters past the end of the walkway. In each condition, subjects performed 10 trials. Data Analysis Analysis of the data consisted first of deriving the (average) walking velocity, stride width, stride length, and stride time (fig 1). The walking velocity was determined by dividing the distance traveled between the first and last foot strikes of a trial by the elapsed time. Stride length was computed in a manner similar to Maki, 13 as the distance between heel locations of 2 consecutive footfalls of the same foot. To reduce the influence of between-subject body dimension differences on stride velocity and stride length, these values were expressed as a percentage of body height. Stride width was also computed in a manner similar to Maki as the perpendicular distance between the line of progression and the heel location of the contralateral foot. Stride time was the time required to complete 1 stride. The mean value for each dependent variable was computed using the total number of strides collected during all of the trials of a given condition. Standard deviations of the 3 variables were used to represent the variability of the dependent variables. 13 The statistical analyses were primarily directed at identifying those dependent variables that were significantly influenced by age and those dependent variables for which age significantly interacted with other independent variables. For experiment 1, the data were analyzed with a (age shoe condition speed) analysis of variance (ANOVA) with repeated measures on the factors of shoe condition and walking speed. For experiment 2, data were analyzed with a 2 2 (age attention condition) ANOVA with repeated measure on the factor of attention. These analyses were performed using SPSS software, version 7.0. b Where indicated, post hoc multiple comparisons were performed using Bonferroni adjusted t tests. The maximum probability level accepted as statistically significant was RESULTS We confirmed that the subjects did walk slower and faster than their self-selected normal velocity during those conditions. This was important because we expected that the walking velocity conditions would influence the variability of the gait variables. The differences among the walking velocities during the normal, slow, and fast speed conditions were significant (p.001), but the age-related differences in walking velocity were not significant. Furthermore, the effect of the shoe condition was not significant. Collapsed across age and shoe conditions, the slow, normal, and fast normalized walking velocities standard deviation were 0.55% 0.10%, 0.74% Fig 1. The manner in which the spatial and temporal stride kinematics were derived. Stride time was the elapsed time between 2 consecutive footfalls of the same foot.

3 EFFECTS OF AGE ON GAIT VARIABILITY, Grabiner 33 Table 3: Summary of Statistical Results, Experiment 2 Variability Measured on Age Main Effect Attention- Splitting Task Interaction Age Attention-Splitting Task Walking velocity Stride width Step length Stride time Fig 2. Stride width variability (cm) of the young (gray bars) and older adults (black bars) collapsed across shoe conditions for the 3 walking velocity conditions. 0.10%, and 0.92% 0.13% body height per second, respectively. The walking velocity variability of the older adults was not significantly different from that of the young adults. Furthermore, the effects of the interaction between age and shoe condition, and age and walking velocity, on walking velocity variability were not significant. Older adults walked with significantly (p.007) larger stride width variability than the young adults (fig 2). The stride width variability standard deviations for the older and young adults was and , respectively. The effects of the interaction between age and shoe condition, and age and walking velocity, on stride width variability were not significant. Notably, the step width standard deviation of the older and young adults, cm and cm, respectively, was not significantly influenced by any of the experimental conditions. Older adults walked with significantly (p.002) larger step length variability (0.01% 0.007% body height) than the young adults (0.0078% 0.006% body height). The effects of the interaction between age and shoe condition, and age and walking velocity, on step length variability were not significant. The stride time variability of the young and older adults was initially found to be affected differently by the shoe condition (age shoe interaction, p.03). After collapsing across walking velocity conditions, further analysis revealed that the stride time variability of the young subjects was unaffected by the shoe condition, whereas the stride time variability of the older adults was significantly increased by shoes. However, the data of the older adults had a large standard deviation and prompted further examination of the data. Four older subjects were observed to have increased stride time variability by twelve- to fourteenfold when wearing shoes. Without these 4 subjects, the main effect of age and the age by shoe interaction terms were no longer significant. Table 2 summarizes the statistical findings of experiment 1. In experiment 2, the walking velocity variability of the young and older adults was found to be affected differently by the performance of the attention-splitting task (p.04). Compared with the control condition, the walking velocity variability of the young adults increased significantly, over 60%, during the attention-splitting task. In contrast, the walking velocity variability of the older adults during the control condition was not different from that found in the attentionsplitting condition. In experiment 2, similar to the results of experiment 1, the older adults walked with significantly larger stride width variability than the young adults (p.004). However, the effect of the attention-splitting task on these variables was not significant. Collapsed across attention condition, the stride width variability of the older adults was about 41% larger than that of the young subjects (table 2). In contrast to experiment 1, the older adults in experiment 2 walked with a significantly larger stride width than the younger adults. The stride width values standard deviations for the older and young adults, collapsed across attention condition, were cm and cm, respectively. In contrast to experiment 1, the older adults in experiment 2 walked with significantly smaller step length variability than the young adults (p.05). However, the effect of the attention-splitting task on these variables was not significant. Collapsed across attention condition, the stride length variability of the older adults was about 23% smaller than that of the young subjects. In experiment 2, similar to the result of experiment 1, stride time variability was not sensitive to age (p.17). Table 3 summarizes the statistical findings of experiment 2. DISCUSSION This study sought to explore further the previously described relations between age and the variability of gait variables. Based on the retrospective and prospective prediction of falls by older adults using such variables, we think that these relationships have potential clinical value. The purposes of the Variability Measure (p) Table 2: Summary of Statistical Results, Experiment 1 Main Effect Interaction Age Shoes Velocity Age Shoes Age Velocity Age Shoes Velocity Walking velocity Stride width Step length Stride time

4 34 EFFECTS OF AGE ON GAIT VARIABILITY, Grabiner experiments were to characterize age-related differences in the variability of gait variables as a function of walking velocity, shoes, and performance of an attention-splitting task while walking. In particular, it was of interest to identify those variables on which the factor of age exerted a significant influence. The most consistent finding from the 2 experiments was that stride width variability was larger in older adults than in young adults. The variability of the other gait variables demonstrated nonsystematic or no age-related differences. Few published data exist with which the present variability data may be compared. Our data are consistent with some, but not all, reports in the literature. Our most consistent and positive result was found for stride width variability. Some of the available data suggest that decreased stride with variability accompanies increasing age and pathology. Decreased stride width variability predicted future falls by older adults living in self-care residences 13 and is decreased by Parkinson s disease, 10 a patient group that is highly susceptible to falls. In a sample 16 of healthy younger and healthy older adults, the differences in stride width variability were not significant, a finding that contrasts with our findings in experiments 1 and 2 in which healthy older adults walked with significantly larger stride width variability than young adults, on average by about 15%. Our second consistent, although negative, outcome was that age-related step time differences were not significant. This finding is similar to those of previously published studies. 12,16 However, stride time variability is significantly larger in older adults with a self-reported history of falls. 12 Notably, in another study, 13 stride time variability was not associated with future falls or a preexisting fear of falling. A limitation of the present study was that questions related to gender differences could not be adequately addressed. This important aspect requires further attention in light of the higher incidence of fall-related hip fractures by women. 17 In experiment 1, the small number of older men who participated (n 4) did not make an age by gender analysis attractive. In experiment 2, the numbers of older men and women were somewhat small, 6 and 8, respectively, but more balanced. A comparison of step width and step width variability between men and women confirmed the inadequate statistical power for gender-related comparisons. The step width standard deviation on men ( cm) was significantly larger than that of women ( cm). However, the associated power, 0.72, was below the generally used value of The step width variability of the men ( cm) was not different from that of the women ( cm). The associated power, 0.18, was much too low to detect the 18% difference as significant. Further calculation revealed that detecting between-gender differences of this size with similar variability would require a sample of 43 men and 43 women. Although it is not possible to state the size of between-gender differences that are clinically significant, further studies of gait variability must consider how large a difference is desired to detect as statistically significant. The number of subjects is a means to control the statistical power of a study. A methodology question, the solution to which we are presently pursuing, relates to the number of steps that are required to obtain a representative and stable measure of gait variables. This question also raises the issue of statistical power. A contributor to inadequate power may have been the failure to collect an adequate number of steps to provide a stable measure of variability. The number of strides collected in the present study was about 20, similar to that of Maki 13 and about twice that reported by Gabell and Nayak. 16 In contrast, Hausdorff et al 12 collected hundreds of steps and reported a coefficient of variation (100 SD/mean) of less than 5% for stride time. For the various conditions measured in the present study, the coefficient of variation for stride time ranged from 31% to more than 100%. The average coefficient of variation computed for the Maki s variables (his table 2) is approximately 52%. Despite the differences in the technologies used to assess stride time, the accuracy with which the technologies can measure time should be good. We suggest that the between-study differences in the coefficient of variation reflect the number of strides collected in each study. If so, the potential utility of measuring the variability of gait variables causes the issue of sample size to be of concern. Specifically, the methodology question requiring attention is how many steps are required to compute a representative measure of variability of gait variables. This question has not yet been directly addressed in the literature. Another question of general import that has not been addressed in the literature is the interpretation of the variability of gait variables. A traditional interpretation of motor performance variability is that it reflects control system instability. In this view, increased motor skill is associated with lower variability in the performance of a motor task and vice versa. In contrast, variability plays an essential and functional role in an adaptive control system. 18 A striking example of physiologically important variability relates to cardiac electrophysiology. Diminished heart rate variability predicts increased risk of sudden death and mortality after a myocardial infarction. 19 Whether increased or decreased variability of gait variables is an outcome of age- or pathology-related processes or, conversely, is an adaptive control system response to the age- or pathology-related processes, is still pending. CONCLUSION The most consistent finding of the present study is that the healthy older participants walked with significantly larger step width variability and generally larger stride width. This is of interest given the findings of Maki 13 who showed that stride width variability is predictive of falls in some older adults. The potential of measures of variability to distinguish between older adults who fall or will fall from older adults who have not fallen and will not likely fall has clinical importance. The ease with which these variables may be measured increases the clinical attractiveness. However, the relevant published studies collectively suggest the need for further systematic study. Future efforts should include study of the stability of age-related changes of variability, the presence of age-related threshold effects, gender-related differences, and, subsequently, whether the trajectory of these changes diverges in a statistically significant and clinically relevant fashion with increased predisposition to falls. These studies will contribute to understanding the age-related mechanisms that give rise to gait changes and will suggest interventions to ameliorate the predisposition to falls. From a clinical standpoint, methods to derive this type of data have the advantage of being time- and cost-effective to administer. Continued research and development in this area seems justified in light of the evidence and given the present social, medical, and economic burdens associated with falls, and particularly of fall-related hip fractures in older adults. References 1. Greenspan SL, Myers ER, Maitland LA, Kido TH, Krasnow MB, Hayes WC. Trochanteric bone mineral density is associated with type of hip fracture in the elderly. J Bone Miner Res 1994;9: Michaelson J, Meyers A, Jinnah R, Cox Q, Van Natta M. Epidemiology of hip fractures among the elderly. Risk factors for fracture type. Clin Orthop 1995;311: Nevitt MC, Cummings SR, Hudes ES. Risk factors for injurious falls: a prospective study. J Gerontol A Biol Sci Med Sci 1991; 46:M164-M170.

5 EFFECTS OF AGE ON GAIT VARIABILITY, Grabiner Cummings S, Black D, Nevitt M, Browner W, Cauley C, Genant H, et al. Appendicular bone density and age predict hip fractures in women. JAMA 1990;263: Grisso JA, Kelsey JL, Strom BL, Chiu GY, Maislin G, O Brien LA, et al. Risk factors for falls as a cause of hip fractures in women. New Engl J Med 1991;324: Berg WP, Alessio HM, Mills EM, Tong C. Circumstances and consequences of falls in independent community-dwelling older adults. Age Aging 1997;26: Pavol MJ, Owings TM, Foley KT, Grabiner MD. The sex and age of older adults influence the outcome of induced trips. J Gerontol A Biol Sci Med Sci 1999;54:M103-M Pavol MJ, Owings TM, Foley KT, Grabiner MD. Gait characteristics as risk factors for falling from trips induced in older subjects. J Gerontol A Biol Sci Med Sci. In press. 9. Blin O, Ferrandez AM, Serratrice G. Quantitative analysis of gait in Parkinson patients: increased variability of stride length. J Neurolog Sci 1990;98: Charlett A, Weller C, Purkiss AG, Dobbs SM, Dobbs RJ. Breadth of base whilst walking: effect of aging and parkinsonism. Age Aging 1998;27: Hausdorff JM, Forman DE, Ladin Z, Goldberger AL, Rigney DR, Wei JY. Increased walking variability in elderly persons with congestive heart failure. J Am Geriatr Soc 1994;42: Hausdorff JM, Edelberg HK, Mitchell SL, Goldberger AL, Wei JY. Increased gait unsteadiness in community-dwelling elderly fallers. Arch Phys Med Rehabil 1997;78: Maki BE. Gait changes in older adults: predictors of falls or indicators of fear? J Gerontol 1997;45: Chen HC, Schultz AB, Ashton-Miller JA, Giordani B, Alexander NB, Guire KE. Stepping over obstacles: dividing attention impairs performance of old more than young adults. J Gerontol A Biol Sci Med Sci 1996;51:M116-M Lundin-Olsson L, Nyberg L, Gustafson Y. Attention, frailty, and falls: the effect of a manual task on basic mobility. J Am Geriat Soc 1998;46: Gabell A, Nayak USL. The effect of age on variability in gait. J Gerontol 1984;39: Grisso JE, Chiu GY, Maislin G, Steinmann WC, Portale J. Risk factors for hip fractures in men: a preliminary study. J Bone Miner Res 1991;6: Riccio GE. Information in movement variability about the qualitative dynamics of posture and orientation. In: Newell KM, Corcos DM, editors. Variability and motor control. Champaign (IL): Human Kinetics; p Klieger RE, Miller JP, Bigger JT, Moss AJ, Multicenter Post- Infarction Research Group. Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am J Cardiol 1987;59: Suppliers a. CIR Systems Inc, PO Box 4402, Clifton, NJ b. SPSS, Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL