EACH YEAR, 750,000 AMERICANS have a stroke,

Transcription

1 1997 Accelerometer Monitoring of Home- and Community-Based Ambulatory Activity After Stroke Elaina Haeuber, MS, Marianne Shaughnessy, PhD, Larry W. Forrester, PhD, Kim L. Coleman, MS, Richard F. Macko, MD From the Baltimore Veterans Affairs Medical Center Geriatrics Research, Education, and Clinical Center, Baltimore, MD (Shaughnessy, Macko); Division of Gerontology (Haeuber, Shaughnessy, Macko); and Departments of Neurology (Haeuber, Macko) and Physical Therapy (Forrester), University of Maryland School of Medicine, Baltimore, MD; University of Maryland School of Nursing, Baltimore, MD (Shaughnessy); and Cyma Corp, Seattle, WA (Coleman). Supported in part by the Baltimore Veterans Administration Geriatrics Research, Education, and Clinical Center, the Veterans Affairs Medical Center Rehabilitation Research and Development (merit review) and the National Institute on Aging (grant no. R29 AG14487), and a Veterans Affairs Rehabilitation Research Development and Career Development Award. Coleman is an employee and minority stockholder in Cyma, which produces the StepWatch Activity Monitor. At the time of data collection, analysis, and interpretation, she had no relationship with Cyma. A commercial party with a direct financial interest in the results of the research supporting this article has conferred or will confer a financial benefit on the author or 1 or more of the authors. Reprint requests to Richard Macko, MD, Dept of Neurology, University of Maryland School of Medicine, 22 N Greene St, Baltimore, MD , rmacko@grecc.umaryland.edu /04/ $30.00/0 doi: /j.apmr ABSTRACT. Haeuber E, Shaughnessy M, Forrester LW, Coleman KL, Macko RF. Accelerometer monitoring of homeand community-based ambulatory activity after stroke. Arch Phys Med Rehabil 2004;85: Objectives: To investigate the utility of a novel microprocessor-linked Step Watch Activity Monitor (SAM) to quantify ambulatory activity after stroke and to evaluate the validity and reliability of conventional accelerometers to measure freeliving physical activity in this population. Design: Cross-sectional with repeated measures of 2 separate 48-hour recordings in 17 persons wearing an anklemounted SAM and Caltrac, a hip-mounted mechanical accelerometer. Setting: Home and community. Participants: Seventeen subjects with chronic hemiparetic gait after stroke. Interventions: Not applicable. Main Outcome Measures: The SAM derived stride counts per day and Caltrac estimated the daily caloric expenditure of physical activity. Results: SAM data revealed that stroke patients had a mean strides per day standard deviation of and demonstrated a broad range of daily activity profiles ( strides). SAM test-retest reliability was high across separate monitoring periods (r.96, P.001). Although Caltrac also revealed a broad range of daily activity calories ( kcal/d; range, kcal/d), reliability was poor (r.044, P not significant) and Caltrac accounted for only 64% of the ambulatory activity quantified by the SAM. Conclusions: Microprocessor-linked accelerometer monitoring, but not conventional accelerometers, are accurate and highly reliable for quantifying ambulatory activity levels in stroke patients. These findings support the utility of personal status monitoring of ambulatory activity as an outcomes instrument and metric in programs to increase physical activity and cardiovascular health after stroke. Key Words: Gait; Hemiplegia; Outcome assessment (health care); Rehabilitation; Walking by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation EACH YEAR, 750,000 AMERICANS have a stroke, making it the leading cause of adult disability in the United States. 1,2 Nearly two thirds have residual neurologic deficits, which impair mobility and promote a sedentary lifestyle, leading to further functional declines because of physical deconditioning and learned nonuse. 1,3,4 A major rehabilitation goal for many stroke survivors is a return to home- and communitybased ambulatory activity, which may reinforce recovery and improve cardiovascular health. 3,5 Despite this goal, little is known about the physical activity profiles of stroke patients. This is because of methodologic limitations on how to measure accurately free-living activity in this population. There are currently no instruments proven valid and reliable for quantifying home- and community-based ambulatory function or for estimating free-living physical activity in the gait-impaired stroke population. A variety of portable devices, including pedometers and accelerometers, have been used to measure ambulatory activity and to estimate the energy expenditure of free-living physical activity. These instruments provide information about physical activity patterns, either in the form of energy expenditure or steps and estimated distance traversed during a specified monitoring period. Although such monitors have proven utility in diverse populations without neurologic disease, 6-13 few studies have evaluated their use in stroke patients, where abnormal gait patterning may render the recordings of such devices unreliable. The StepWatch Activity Monitor a (SAM) is a pager-sized microprocessor-linked accelerometer with adjustable filtering thresholds for motion and cadence parameters. Adjustable calibration enables this programmable monitor to recognize stance and swing components of numerous gait patterns in animals and humans, providing time-integrated ambulatory activity profiles across user-defined epochs. 14 The SAM is worn on 1 ankle, and hence records the occurrence of 1 complete gait cycle, or stride. We recently reported 15 that the SAM, but not a conventional mechanical pedometer, is highly accurate in laboratory-based timed walk tests of gait-impaired stroke patients. However, no other studies to our knowledge have established the feasibility or reliability of continuous home- and community-based monitoring of ambulatory activity in stroke patients by using this technology. Caltrac b is a commonly used belt-worn mechanical accelerometer that estimates total caloric expenditure by adding calories from physical activity to an estimate of the resting metabolic rate. This device has proven reliable and valid for measuring physical activity in various populations in both the laboratory and field settings, including in older individuals and

2 1998 ACTIVITY MONITORING AFTER STROKE, Haeuber those with peripheral arterial occlusive disease. 8,16 Caltrac has also been validated against the technique of doubly labeled water, the criterion method for estimating the energy expenditure of physical activity. 9 However, the validity and reliability of conventional accelerometer-derived estimates of free-living energy expenditure are not established in people with gait abnormalities after stroke. The purpose of our study was to investigate the feasibility and reliability of continuous microprocessor-linked accelerometer recording of ambulatory activity in community-dwelling stroke patients with a broad spectrum of gait-deficit severity, assistive device, and orthosis requirements. Simultaneous Caltrac recordings across the repeated 48-hour outpatient monitoring epochs were used to evaluate the validity and reliability of conventional accelerometer estimates of free-living physical activity in this population. METHODS Participants older than 50 years, with remote ischemic stroke ( 6mo), were recruited from the Baltimore Veterans Affairs Medical Center (VAMC) and the University of Maryland (UM) Medical System. All had completed conventional physical therapy and were left with residual hemiparetic gait deficits with some preserved capacity for ambulation, albeit with an assistive device (ie, walker or cane) and/or standby aid. Chronic hemiparetic stroke participants were selected to optimize test-retest reliability studies within a heterogeneously gait-impaired, but neurologically stable, stroke cohort. Baseline evaluations included a medical history and physical examination, a Mini-Mental State Examination to screen for dementia (score, 22), and a Center for Epidemiologic Studies Depression screen (score, 16). 17,18 Exclusion criteria included congestive heart failure (New York Heart Association class II), unstable angina, dementia, global or severe receptive aphasia (unable to follow 2-point commands), peripheral vascular disease, or other major medical conditions beyond the stroke that could markedly limit home- and community-based ambulatory function. The Baltimore VAMC and the UM institutional review board approved the study, and all participants provided written informed consent. Based on our earlier study in hemiparetic stroke patients, 15 the SAM was programmed using initial generic calibration and a 12-second recording epoch. Before outpatient monitoring, each participant performed two 1-minute timed floor walks at their self-selected and fastest comfortable paces, with 30-second interval rests in between. Self-selected and fastest comfortable paces were selected to assess whether the customizable calibration settings were accurate across the anticipated range of gait performance for each subject. The SAM counts for strides taken were compared with visual counts made by using a hand tally counter. If the SAM showed less than 94% accuracy against visual counts, adjustments were made to the calibration settings and the participants repeated the 1-minute timed walks. Once calibration was complete, the SAM was programmed and attached by using 2 adjustable elastic straps with Velcro closure to the nonparetic ankle above the medial malleolus if the left leg and above the lateral malleolus if the right leg. Caltrac was programmed with the participant s age, weight, height, and gender, to estimate resting metabolic rate, and attached to the nonparetic hip at waist level along the midaxillary line. The Caltrac display is cumulative and updated every 2 minutes, to show the estimated total (gross) energy expenditure and activity energy expenditure. After attachment of the monitors, participants were instructed on how to remove and replace the monitors during sleep and bathing. Within 3 weeks, each participant repeated the 48-hour monitoring period wearing the 2 monitors. On completion of each 48-hour monitoring period, SAM data were downloaded by using an infrared docking port (SAM Dock Program 1.6.L a ). Data were expressed as total stride counts from which strides per 24-hour period were calculated. The number of activity calories expended during the monitoring period was recorded from Caltrac. Intraclass correlation coefficients were calculated to analyze the test-retest reliability of the SAM and Caltrac measures derived from the two 48- hour monitoring periods conducted on separate weeks. Pearson correlation coefficients were used to analyze the strength of the relationship between Caltrac estimates of the energy expenditure of free-living physical activity and the total ambulatory activity for each recording period derived by SAM. RESULTS We tested 17 participants who had a mean age standard deviation (SD) of 65 6 years, including 11 with right- and 6 with left-hemisphere infarctions. The mean latency since stroke was 41.6 months (range, 9 120mo). Two participants had mild deficits requiring no assistive device during routine ambulatory activities at home or in the community, 11 used a single-point cane, and 4 had greater deficit severity requiring a walker (n 1) or quad cane (n 3). The latter 4 used a wheelchair for any sustained mobility or travel outside the home. Nine participants used an ankle-foot orthosis. The mean self-selected floor walking velocity calculated from the 1-minute calibration walks was.73.30m/s (range, m/s). The initial generic cadence and motion sensitivity settings (cadence 80, motion 12) for the SAM proved 94% accurate in all cases for the 1-minute timed walks at self-selected and fastest comfortable paces. Therefore, no calibration adjustments were needed. SAM and Caltrac data for the 2 separate 48-hour monitoring periods are shown in table 1. SAM data revealed that stroke patients had a mean of total strides per day, with subjects displaying a broad range of ambulatory activity profiles, ranging from 336 to 6472 total strides per day. Ambulatory activity profiles derived by using the SAM showed excellent test-retest reliability for total stride counts per 24 hours across the separate 48-hour monitoring periods (r.96, P.001). Caltrac recordings also revealed a broad range of activity energy expenditure from 83 to 1222kcal/d, with the mean estimated activity energy expenditure of kcal/d. In contrast with the SAM, simultaneously conducted measures with Caltrac revealed poor test-retest reliability for activity calories (r.44, P not significant). Simple regression analysis revealed a positive relationship between the Caltrac estimates of activity energy expenditure and SAM-derived total stride counts across both the first 48-hour monitoring period (r.77, P.001) and the second monitoring period (r.82, P.001) (fig 1). Hence, Caltrac activity energy expenditure accounts for only 64% (r 2 ) of the actual ambulatory activity measured as total stride counts by the SAM. DISCUSSION The main finding of our study is that microprocessor-based accelerometer recordings using the SAM were accurate and showed high test-retest reliability for measuring total daily ambulatory activity in stroke patients. Further, we report that conventional belt-mounted mechanical accelerometer estimates of daily activity calories were not reliable in this population and accounted for only 64% of the variance in ambulatory activity that was measured by the microprocessor-linked accelerometer recordings. These findings show the utility of micro-

3 ACTIVITY MONITORING AFTER STROKE, Haeuber 1999 *P.001. Table 1: Test-Retest Reliability of the SAM and Caltrac to Measure Daily Activity Profiles in 17 Gait-Impaired Stroke Participants, Based on 2 Separate 48-Hour Monitoring Periods Subject No. SAM Steps/d Caltrac kcal/d Test Period 1 Test Period 2 Test Period 1 Test Period Mean SD Test-retest reliability (r).96*.44 processor-linked accelerometer recordings to quantify ambulatory activity in hemiparetic stroke patients. Accelerometers have been applied to measure physical activity and estimate energy expenditure in numerous populations. However, their validity depends on the gait characteristics of the study population. In healthy populations, Caltrac is accurate in measuring energy expenditure compared with both doubly labeled water and indirect calorimetry. Bassett et al 8 found that energy expenditure measured by using Caltrac during performance of brisk walking and activities of daily living (ADLs) correlated with that measured by indirect calorimetry (r.580). Caltrac also provides reliable estimates of energy expenditure during treadmill walking at various speeds. 19 The most compelling evidence for the validity of Caltrac is its strong correlation with free-living energy expenditure measured by using the doubly labeled water technique, the criterion standard for field monitoring. Montoye et al 20 reported a correlation of r equal to.94 between accelerometer readings Fig 1. Scatterplot showing the relationship between SAM total stride counts and Caltrac activity calories for each of the two 48- hour monitoring periods in 17 hemiparetic stroke patients. within a 2-week period in healthy adults. Accelerometers are also proven reliable in selected chronic disease populations. Sieminski et al 16 reported a correlation of r equal to.84 between Caltrac readings during two 48-hour monitoring periods in subjects with peripheral arterial occlusive disease. These studies show that conventional accelerometers are both valid and reliable in populations without significant gait abnormalities. Although conventional accelerometers are accurate in healthy populations, their utility in neurologic populations has not been established. We found that Caltrac has poor test-retest reliability in stroke patients and accounts for less than two thirds of the variance in ambulatory activity measured by the SAM. These findings indicate that conventional accelerometer recordings lack validity and should not be considered as a physical activity or outcomes monitoring instrument in this population. Caltrac, as well as other accelerometers (ie, the computer science and applications monitor, the Tritrac), do not account for asymmetries in gait. 6,8,10-12 People with short strides or shuffling gait have less vertical displacement of the center of gravity and lower-energy expenditures than when walking similar distances using longer strides and smoother gaits. 16 Other studies 10,21 indicate that Caltrac is less accurate in estimating energy expenditure during low-level activities. Hemiparetic gait after stroke is slow and characterized by abnormal gait biomechanics, including asymmetric temporaldistance and force parameters and increased cycle-to-cycle variability. 22 Such factors exemplify how the usual conservation of mechanical energy between whole-body potential and translational kinetic energy may not be optimal for proficient gait, further confounding interpretation of Caltrac estimates of energy expenditure that are derived from algorithms in healthy patients. 23,24 Hence, the limitations of conventional mechanical accelerometers with abnormal gait biomechanics likely account for their poor reliability and inadequate construct validity for measuring energy expenditure of ambulatory activity in stroke patients.

4 2000 ACTIVITY MONITORING AFTER STROKE, Haeuber In contrast to Caltrac, we report excellent test-retest reliability in stroke patients for SAM recordings of total stride counts (r.96) measured during 2 separate 48-hour monitoring periods. These findings corroborate our earlier study 15 in which the SAM, but not a conventional mechanical pedometer, proved highly accurate and reliable during 6-minute walks in hemiparetic stroke patients. Because the SAM was designed with an electronic filter to reject extraneous signals, one can adjust parameters of threshold, cadence, and motion, to enhance sensitivity to movements associated with a wide range of gait styles. 14 This may account for its accuracy in diverse study populations, including orthopedic and obese patients with total hip replacement, frail elderly, persons walking with a total contact case, healthy children, and those with Duchenne s muscular dystrophy. 13,14,25-28 In contrast with Caltrac, in which a subject pressing buttons can easily disrupt data collection, 8 the SAM is programmed by using a dock and computer and is less obtrusive because it is mounted on the ankle rather than the waist. 14 Such features may have contributed to the lack of technical problems and good tolerability by participants using the SAM in our field-testing studies. Collectively, these studies show the feasibility, validity, and reliability of the SAM for recording ambulatory activity in gait-impaired stroke patients. Portable monitoring of ambulatory activity may have important clinical applications as a mobility outcomes instrument and metric to measure physical activity after stroke. Conventional stroke outcome instruments consist of questionnaires indexing mobility function, including contemporary stroke-specific instruments and standardized examinations or observer-rated scales that categorize functional dependence. 3,29-33 Quantitative and laboratory-based measures, such as timed walks, gait biomechanics, and performances on standardized simulated ADL tasks, constitute another important category for quantifying recovery of physical performance capacity after stroke. However, none of these methodologies directly measures ambulatory activity in the home and community setting. Microprocessor-linked accelerometer technology affords an opportunity to determine the extent to which stroke patients incorporate their motor gains into daily ambulatory activity across the stroke recovery period. Such information could prove useful in optimizing outpatient rehabilitation of mobility recovery. Personal status monitoring of ambulatory activity will further reveal whether people who have had a stroke meet the consensus recommendations for regular daily physical activity forwarded by the American Heart Association and numerous public health agencies Most stroke survivors assume a sedentary lifestyle and have profound cardiovascular deconditioning, which may accelerate atherosclerosis and which increases the risk for recurrent cardiovascular events and stroke Periodic monitoring of daily physical activity by using microprocessor-linked technology could serve as a metric in health promotion programs to help stroke survivors improve their cardiovascular health and fitness. Our findings are limited by small sample size and inclusion of stroke survivors with only chronic hemiparetic gait deficits. We did not evaluate the SAM during the subacute phases of stroke rehabilitation and thus cannot determine its sensitivity relative to other mobility outcomes instruments. The SAM was not tested in stroke patients with bilateral gait impairment or with predominant cerebellar gait patterns. Caution should be taken in extrapolating our findings to patients with such complex gait patterns, in which further customization of SAM parameters and individualized validation may be needed. Furthermore, volunteers in this study were all participants in an investigation of treadmill aerobic exercise after stroke, which may introduce selection bias. Their ambulatory activity profiles may not be representative of community-dwelling stroke patients and are unlikely to reflect the activity patterns during earlier phases of stroke recovery. Similarly, caution must be taken in interpreting results of the Caltrac recordings. Although low mean activity calorie values may suggest reduced activity profiles in these patients, this cannot be equated to a reduced metabolic cost of free-living physical activity. The energy cost of hemiparetic gait is elevated 1.5- to 2-fold compared with normal gait. 41,42 Hence, accelerometer estimates of caloric expenditure derived by using algorithms based on normal populations would be expected to underestimate the true energy expenditure in stroke patients. Studies performed by using doubly labeled water or portable indirect calorimetry are needed to understand better energy metabolism and its relation to physical activity in this population. CONCLUSIONS We report that microprocessor-linked accelerometer recordings using the SAM, but not conventional accelerometers, are accurate and highly reliable for quantifying ambulatory activity in chronic hemiparetic stroke patients with a broad range of gait-deficit severity and varied ambulatory assistive-device and orthosis requirements. Further studies are needed to determine their sensitivity as an outcomes instrument across the stroke recovery period and their utility as a metric in rehabilitation programs to enhance mobility recovery, cardiovascular health, and fitness after stroke. References 1. Gresham GE, Dawber TR. Residual disability in survivors of stroke: the Framingham Study. N Engl J Med 1975;293: Williams GR, Jiang JG, Matchar DB, Samsa GP. Incidence and occurrence of total (first ever and recurrent) stroke. Stroke 1999; 30: Gresham GE, Duncan PW, Adams HP, et al. Post-stroke rehabilitation. Clinical practice guideline no. 16. Rockville: US Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research; AHCPR Publication No Liepert J, Bauder H, Miltner WH, Taub E, Weiller C. Treatmentinduced cortical reorganization after stroke in humans. Stroke 2000;31: World Health Organization. ICIDH-2. International classification of impairments, activities and participation: a manual of dimensions of disablement and functioning. Beta-1: Draft for field trials. Geneva: WHO; Ainsworth BE, Bassett DR Jr, Strath SJ, et al. Comparison of three methods for measuring the time spent in physical activity. Med Sci Sports Exerc 2000;32(Suppl):S Bassett DR Jr, Ainsworth BE, Leggett SR, et al. Accuracy of five electronic pedometers for measuring distance walked. Med Sci Sports Exerc 1996;28: Bassett DR Jr, Ainsworth BE, Swartz AM, Strath SJ, O Brien WL, King GA. Validity of four motion sensors in measuring moderate intensity physical activity. Med Sci Sports Exerc 2000; 32(Suppl 9):S Gardner A, Poehlman E. Assessment of free-living daily physical activity in older claudicants: validation against the double labeled water technique. J Gerontol A Biol Sci Med Sci 1998;53:M Hendelman D, Miller K, Baggett C, Debold E, Freedson P. Validity of accelerometry for the assessment of moderate intensity physical activity in the field. Med Sci Sports Exerc 2000;32 (Suppl 9):S Leenders NY, Sherman WM, Nagaraja HN. Comparisons of four methods of estimating physical activity in adult women. Med Sci Sports Exerc 2000;32: Nicols JF, Patterson P, Early T. A validation of a physical activity monitor for young and older adults. Can J Sports Sci 1992;17:

5 ACTIVITY MONITORING AFTER STROKE, Haeuber Shepherd EF, Toloza E, McClung CD, Schmalzried TP. Step activity monitor: increased accuracy in quantifying ambulatory activity. J Orthop Res 1999;17: Coleman KL, Smith DG, Boone DA, Joseph AW, del Aguila MA. Step activity monitor: long-term, continuous recording of ambulatory function. J Rehabil Res Dev 1999;36: Macko RF, Haeuber E, Shaughnessy M, et al. Microprocessorbased ambulatory activity monitoring in stroke patients. Med Sci Sports Exerc 2002;34: Sieminski DJ, Cowell LL, Montgomery PS, Pillai SB, Gardner AW. Physical activity monitoring in patients with peripheral arterial occlusive disease. J Cardiopulm Rehabil 1997;17: Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12: Parikh RM, Eden DT, Price TR, Robinson RG. The sensitivity and specificity of the Center for Epidemiologic Studies Depression Scale in screening for post-stroke depression. Int J Psychiatr Med 1988;18: Pambianco G, Wing RR, Robertson R. Accuracy and reliability of the Caltrac accelerometer for estimating energy expenditure. Med Sci Sports Exerc 1990;22: Montoye HJ, Washburn R, Servais S, Ertl A, Webster JG, Nagle FJ. Estimation of energy expenditure by a portable accelerometer. Med Sci Sports Exerc 1983;15: Miller DJ, Freedson PS, Kline GM. Comparison of activity levels using the Caltrac accelerometer and five questionnaires. Med Sci Sports Exerc 1994;26: Harris-Love M, Forrester L, Macko RF, Silver KH, Smith GV. Hemiplegic gait parameters in overground versus treadmill walking. Neurorehabil Neural Repair 2001;15: Cavagna GA, Saibene FP, Margaria R. External work in walking. J Appl Physiol 1963;18: Caldwell GE, Forrester LW. Estimates of mechanical work and energy transfers: demonstration of a rigid body power model of the recovery leg in gait. Med Sci Sports Exerc 1992;24: Resnick B, Nahm ES, Orwig D, Zimmerman SS, Magaziner J. Measurement of activity in older adults: reliability and validity of the Step Activity Monitor. J Nurs Meas 2001;9: Hartsell H, Fitzpatrick D, Brand R, Frantz R, Saltzman C. Accuracy of a custom-designed activity monitor: implications for diabetic foot ulcer healing. J Rehabil Res Dev 2002;39: McDonald CM, Walsh DD, Widman L, Walsh SA, Coleman K. Use of the Step Activity Monitor for continuous objective threeday physical activity monitoring in children [abstract]. Dev Med Child Neurol 1999;80:S Widman L, Walsh DD, Walsh SA, Coleman K, McDonald CM. Use of the Step Activity Monitor for continuous objective physical activity assessment in children with Duchenne muscular dystrophy [abstract]. Dev Med Child Neurol 1999;80:S Kelly-Hayes M, Robertson JT, Broderick JP, et al. The American Heart Association Stroke Outcome Classification: executive summary. Circulation 1998;97: Mahoney FI, Barthel D. Functional evaluation: the Barthel index. Md State Med J 1965;14: Collen FM, Wade DT, Robb GF, Bradshaw CM. The Rivermead Mobility Index: a further development of the Rivermead Motor Assessment. Int Disabil Stud 1991;13: Duncan PW, Wallace D, Lai SM, Johnson D, Embretson S, Laster LJ. The stroke impact scale version 2.0. Evaluation of reliability, validity, and sensitivity to change. Stroke 1999;30: Keith RA, Granger CV, Hamilton BB, Sherwin FS. The functional independence measure: a new tool for rehabilitation. Adv Clin Rehabil 1987;1: Physical activity and cardiovascular health. NIH Consensus Development Panel on Physical Activity and Cardiovascular Health. JAMA 1996;276: Pate RR, Pratt M, Blair SN, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995;273: Fletcher GF, Blair SN, Blumenthal J, et al. Statement on exercisebenefits and recommendations for physical activity programs for all Americans. A statement for health professionals by the Committee on Exercise and Cardiac Rehabilitation of the Council on Clinical Cardiology, American Heart Association. Circulation 1992;86: Lee IM, Paffenberger RS. Physical activity and stroke incidence: the Harvard Alumni Health Study. Stroke 1998;29: Sacco R, Gan R, Boden-Albala B, et al. Leisure-time physical activity and ischemic stroke risk: The Northern Manhattan Stroke Study. Stroke 1998;29: Potempa K, Lopez M, Braun LT, Szidon JP, Fogg L, Tincknell T. Physiological outcomes of aerobic exercise training in hemiparetic stroke patients. Stroke 1995;26: Ryan AS, Dobrovolny CL, Silver KH, Smith GV, Macko RF. Cardiovascular fitness after stroke: role of muscle mass and gait deficit severity. J Stroke Cerebrovasc Dis 2000;9: Gerston J, Orr W. External work of walking in hemiparetic patients. Scand J Rehabil Med 1971;3: Corcoran PJ, Jebsen RH, Brengelmann GL, Simons BC. Effects of plastic and metal leg braces on speed and energy cost of hemiparetic ambulation. Arch Phys Med Rehabil 1970;51: Suppliers a. Cyma Corp, th Ave NE, Ste C, Seattle, WA b. Muscle Dynamics Fitness Network, Hamilton Ave, Torrance, CA