Despite the well-known benefits of physical activity (PA) for all segments of society, older adults are considered among the least active (12). Older adults in nursing homes (NH) may be particularly inactive, spending up to 80% of their waking time either lying or sitting (17). Accurate quantification of older adult PA is necessary for surveillance, screening, intervention, and evaluation purposes (18); however, very low levels of PA present a measurement challenge (28). In field settings, PA behaviors have been typically assessed using self-report methods (e.g., surveys and diaries). For older adults, the limitations of these approaches include recall bias (21), floor effects (28), and insensitivity to incidental daily walking behaviors (1,20). An alternate approach, which has rapidly evolved in the past decade, is objective monitoring of PA using body-worn instruments such as accelerometers and pedometers (26).
An accelerometer measures acceleration of movement (from which intensity may be inferred) in up to three dimensions (9). At least three studies to date have used accelerometers to assess PA behaviors and/or detect change in PA level as a result of exercise interventions in older populations (2,13,17). Pedometers, meanwhile, detect vertical accelerations of the hip and are used to measure steps taken during ambulatory activity (6). This is an important consideration given that walking is a preferred leisure-time PA for older adults (30) and is necessary for basic mobility, in both community and NH environments (25). These devices have been used to document the number of steps per day taken by community-dwelling older adults, including individuals living with chronic health problems (29). Of the two types of motion sensors, pedometers may be the most feasible (i.e., practical and inexpensive) for ascertaining PA in the field (3,9).
With increased interest in pedometers, it is necessary to critically examine their utility and limitations in different populations and settings. Only one study to date, by Cohen-Mansfield et al. (7), has examined the feasibility of using pedometers in a NH setting, reporting that the devices were easy to use and well tolerated. Other researchers have speculated, however, that the slow walking speed of some older adults may preclude the use of pedometers with this population (13). In younger populations, pedometers have underestimated the number of steps taken by as much as 25% at slow (<54 m·min−1 or 0.9 m·s−1) walking speeds (4,10). Because this walking speed is lower than that (1.0 m·s−1) associated with independent living (8), pedometer error should be inconsequential when assessing PA among healthy, community-dwelling older adults. However, slow walking speed is more characteristic of NH residents; reports range between 0.46 and 0.63 m·s−1 (14,17). The shuffling gait pattern of some older adults may also contribute to pedometer error in detecting actual steps taken (2,13,22). In fact, gait impairments (such as hemiparesis and ataxia) were found to compromise pedometer accuracy with community-dwelling stroke patients (16).
To date, pedometers have not been thoroughly tested with older adult populations. Therefore, the purpose of the present study was to examine the effects of walking speed and gait disorders on the accuracy of pedometers with NH residents relative to apparently healthy older adults living in the community.
Twenty-six residents (5 males, 21 females) were recruited from two NH facilities, whereas 28 community-dwelling older adults (5 males, 23 females) were recruited from a local seniors’ recreation center (SC). Uniform exclusion criteria were: a recent cardiovascular event, vestibular disorder or fracture (past 4 months), uncontrolled hypertension, severe visual or hearing impairment, and dementia. Study participants were asked to review a letter of information and provide informed consent. Ethical approval for this study was obtained from the Office of Research Ethics at the University of Waterloo.
Participant characteristics, collected via background questionnaire, are presented in Table 1. The questionnaire was self-completed by the SC sample and completed by NH staff on behalf of the residents.
A Yamax Digiwalker pedometer (Model DW-200, Yamax Corporation, Tokyo, Japan) was used in this study. Yamax-brand pedometers record within 1% of all steps taken under controlled conditions (4) and have shown a strong relationship (r = 0.80–0.93) with more expensive accelerometers (5,27). All units were checked for calibration before use as recommended elsewhere (26). Briefly, the researcher walked a short distance at normal walking pace and counted steps taken with a hand-tally counter (26). Percent error for all pedometers was <5%. Participants wore the pedometer attached to clothing at the waist, centered over the dominant foot as per the manufacturer’s instructions.
A 13-m distance was measured in a quiet hallway in each facility. After a single, normal pace practice trial (not wearing pedometers), all participants performed one walking trial of the course with pedometers at three different self-paced speeds: slow, normal, and fast (14). Specifically, participants were instructed to walk one length of the course, using usual gait aids, accordingly: 1) walk rather slowly; 2) walk at a normal pace, neither fast nor slow; and 3) walk rather fast but without overexerting yourself. The number of pedometer-detected steps was recorded at the end of each trial, and the device was reset to zero before the next trial. Actual steps taken were observed and recorded using a hand-tally counter. All walking trials were videotaped for verification purposes. To reduce the effects of acceleration and deceleration, walking speed was ascertained during the middle of each trial (14). To determine walking speed, a photoelectric cell was placed at the beginning and end of the middle 7 m of the walking course. The cell at the end of this distance contained a timing device. When the participant passed through the infrared beam at the starting point, the timing device started. The device stopped when the participant passed through the beam at the end of 7 m. Walking speed (m·s−1) was calculated by dividing 7 m by the time in seconds.
To examine possible related disorders, gait was evaluated during the practice walking trial using the gait subscale of Tinetti’s Performance-Oriented Mobility Assessment or POMA (25). Interrater reliability of the POMA, measured via percent agreement, has been reported as 85 ± 10% (25). A correlation of gait scores with physical performance test scores supports the concurrent validity of the subscale (Pearson r = 0.78) (19). The examiner observes and rates eight components during walking: initiation of gait, step length and height, step symmetry and continuity, path deviation, trunk sway, and walk stance. Each component is scored as normal (1 point) or abnormal (0 points), with exception of two items that have a maximum of two points. The total gait score can range from 0 to 12 (with higher scores indicating less impairment). In the NH setting, each resident was independently assessed by two raters (the researcher and the facility’s kinesiologist). Having established consistent interrater reliability of the gait subscale with NH residents (ICC = 0.84; 95% CI = 0.67, 0.93, P < 0.0001) and anticipating that there would be less variation in gait score within a community sample, only one rater conducted the gait assessment for the SC group.
Descriptive data are presented as means, SD, 95% confidence intervals (CI) for all continuous variables. Percent error was calculated as ([pedometer steps − observed steps]/observed steps) × 100. A positive value indicated overcounting (extra steps detected), and a negative value indicated undercounting of the pedometer (missed steps). Values close to zero indicated more accurate pedometer results.
Independent t-tests and chi square analyses were used to evaluate differences between the NH and SC groups on the continuous and categorical variables, respectively. POMA gait scores were not normally distributed so Spearman rank order correlations were computed to assess the strength of the relationship between gait score and pedometer percent error and gait score and walking speed (at normal pace) across NH and SC groups combined. A two (NH vs SC) by three (walking trial) repeated-measures ANCOVA was performed to examine differences in walking speed. Age was used as a covariate to control for its potential confounding effect on walking speed (11). A two (NH vs SC) by three (walking trial) repeated-measures ANOVA was used to examine differences in pedometer error. If the overall F were significant for main effects, post hoc tests were performed. Supplementary analyses were conducted if a significant interaction effect emerged. For all analyses, significance was set at an alpha level of P < 0.05. Data were analyzed using the Statistical Package for the Social Sciences (SPSS), version 10.0.
As shown in Table 1, the SC sample was significantly younger, more independently mobile (none relied on gait aids), and took fewer prescribed medications than the NH sample (all P < 0.0001). The average score for the gait subscale of the POMA in the NH sample was 8.4 ± 1.9 (95% CI = 7.7, 9.2) which was significantly lower (P < 0.0001) than for the SC sample (11.9 ± 0.3; 95% CI = 11.7, 12.0). Higher gait scores indicate less impairment. At normal pace, gait score was found to be moderately, yet significantly, correlated with pedometer error (rs = 0.46, P < 0.0001) and highly correlated with walking speed (rs = 0.70, P < 0.0001) for NH and SC combined. Although shuffling gait patterns were only observed in one NH resident (most had normal step height), those using walkers demonstrated little vertical hip movement.
Table 2 shows the slow, normal, and fast pace walking speeds of each group. There was no significant difference between walking speed during the practice trial (without pedometers) and the normal pace trial for either group (P = 0.87 for NH and P = 0.73 for SC). The results of the ANCOVA revealed a significant group by trial interaction [F (1,81) = 7.9, P < 0.01]. After supplementary analyses (ANCOVA and repeated-measures ANOVA), significant differences emerged for all pairwise comparisons. Adjusting for age, the community sample walked significantly faster (P < 0.0001) than the NH sample at all three paces. Average walking speed significantly increased from slow to fast pace (P < 0.0001) for both groups.
Table 3 presents the performance of the Yamax pedometers relative to observed steps taken (i.e., percent error). Pedometer accuracy improved (decrease in percent error) for both samples as walking speed increased [F (1,84) = 34.9, P < 0.0001]. At each pace, the magnitude of the error was greater for the NH versus the SC sample [F (1,52) = 26.5, P < 0.0001). Significant differences emerged for all pairwise comparisons (P < 0.001).
Given that Bassett et al. (4) reported greater pedometer accuracy at speeds >54 m·min−1 (equivalent to 0.9 m·s−1), the present data were examined to determine whether any of the NH residents met this criterion. Only four residents walked faster than 0.9 m·s−1 (at both their normal and fast paces), none of whom required walking aids and all of whom were considered by NH staff to be among the most physically independent residents. Their gait scores were also significantly better (P < 0.05) compared with the other 22 NH residents (10.4 ± 2.4; 95% CI = 6.7, 14.0 vs 8.1 ± 1.6; 95% CI = 7.4, 8.8). Pedometer recordings were more accurate for these four NH residents. During the normal walking pace trial, the pedometer detected an excess of only 3% on average for these residents, compared with missing 66% of the steps taken by the remainder of the sample.
This study confirms previous findings that the Yamax pedometer underestimates the number of steps taken at slower walking speeds (4,10,15). Specifically, Bassett et al. (4) reported that walking speeds <54 m·min−1 (equivalent to 0.9 m·s−1) were associated with 25% pedometer error. The average walking speeds for the present NH sample (0.64 m·s−1 at normal pace and 0.8 m·s−1 at fast pace) are comparable to previous findings from institutional settings (14) but slower than the slowest speeds found under controlled conditions with younger populations (4,10). The self-selected walking speeds of the SC sample, meanwhile, were characteristic of independent older adults (8,24). The finding that the pedometer missed significantly fewer steps at normal and fast paces for both groups (compared with slow pace) provides further evidence of the impact of walking speed on pedometer error.
The results of the present study also validate the assumption that gait disorders compromise pedometer accuracy (2,13,16,22). Correlative analysis showed that the pedometer missed fewer steps when worn by participants with a higher score on the gait subscale of the POMA. Participants who walked faster at normal pace also tended to have higher gait scores. It is noteworthy that pedometer error was considerably lower for the four NH residents whose normal walking speed was above 0.9 m·s−1 and who had fewer gait disorders compared with the other residents in the sample.
Comparing the two samples verified that the natural walking behavior of residents contributed to the pedometer’s inability to detect steps taken. The NH adults walked significantly slower than the SC adults at all paces and had significantly greater gait impairment scores (scores of the latter group indicated almost no gait disorders). Pedometer accuracy improved considerably when tested under the same conditions with the community-dwelling group.
The Yamax pedometer model used in this study contains a horizontal, spring-suspended lever arm that deflects with the usual up-and-down motion of the hips during walking. A force ≥ 0.35 g causes an electrical circuit to close and record a single step taken (27). Less forceful movements are insufficient to close the circuit and are therefore not registered as steps taken (4). This sensitivity threshold (i.e., ≥0.35 g) appears to be appropriate for assessing normal paced walking in ostensibly healthy older adults but not walking behavior typical of frail older adults. Although it is tempting to call for manufacturers to lower the sensitivity threshold of pedometers, the trade-offs are important to consider. Namely, if the instrument has greater sensitivity to detect slow walking or compensate for gait impairments, then the researcher must be willing to accept decreased specificity or the ability to discriminate between steps taken and forces generated from external vibrations including, but not limited to, travel in motorized vehicles (15). Further, varying the sensitivity threshold from study to study or population to population will make it more difficult to interpret and compare findings.
Only one other study to date has examined the impact of gait disorders on pedometer performance. Macko et al. (16) compared the accuracy of a step activity monitor (SAM), a type of accelerometer positioned at the ankle, versus a waist-borne pedometer in a sample of 16 community-dwelling, gait-impaired stroke patients (mean age 67 yr). During 1-min floor walks at self-selected paces, approximately 14% and 1.5% of actual steps were missed by the pedometer and the SAM, respectively. The average gait speed of the sample was 0.74 m·s−1. In contrast to the present study, Macko et al. (16) did not examine pedometer accuracy with respect to walking speed and gait disorders. However, there was evidence from their findings to similarly suggest that these factors may have influenced pedometer performance.
Before concluding that the SAM is superior to the pedometer, it is important to consider that cadence sensitivity and motion parameters of the SAM were preset for slow walkers. When the SAM was < 90% accurate, the researchers changed the sensitivity setting and the trial was repeated (16). No similar attempts were made to improve pedometer accuracy by adjusting its placement. Furthermore, according to the manufacturer, the SAM (with its docking station and customized software) costs over $3000. In contrast, at $10–50 per unit, pedometers are much less expensive (29).
In the only other pedometer study in NH (7), the researchers assessed the wandering/pacing behavior of 10 cognitively impaired residents (mean age 84 yr) using a shoe-borne pedometer and three other motion sensor devices. Resident behavior, wearing a different device each day was observed for 10 min·h−1 over 12 h. The pedometer, clipped to the shoe, was reportedly fidgeted with and removed more frequently than the other devices (7); however, accuracy was not assessed. In pilot testing, we clipped Yamax pedometers to the ankle and shoe. Because this device was designed to be waist-borne, threshold forces exceeding 0.35 g were easily achieved when the pedometer was closer to the ground, thus resulting in an overestimation of steps. Under field conditions, we were also concerned that shaking, tapping, and other fidgeting would affect the accuracy of recordings (overestimation). Accordingly, we decided to restrict our examination of pedometer accuracy to the traditional, waist-borne attachment site, as specified by the manufacturer.
One limitation of the present study is that the NH sample consisted of higher functioning residents (both physically and cognitively) selected by staff. Although the findings cannot be generalized to nonambulatory, cognitively impaired residents, it is unlikely that pedometers would be used with such individuals, even if more accurate instruments become available. Similarly, the present findings concerning community-dwelling older adults are restricted to members of recreational SC, who tend to be healthier, more active, and more likely to volunteer than seniors in general (18).
It must also be mentioned that the distance for the self-paced walk test in the present study was reduced to 13 m, from the original 20 m (11), following the protocol modification by Lazowski et al. (14) to enable completion by frail, NH residents. Over such a short distance, miscounts of only a few steps can exaggerate percent error (23). Shepherd et al. (23) found greater pedometer error when middle-aged subjects were asked to walk 10 m slowly (15.5% error) versus 400 m briskly (2.3% error). In short walking trials, miscounting the first and last steps may also increase error because, unlike a mid-walk step, both the swing phase and weight transfer are shorter and produce less movement of the hip (23). Given the substantial magnitude of error that emerged concerning pedometer use with NH residents, it is unlikely that miscounting due to the short distance of the walk test would appreciably affect the findings.
In conclusion, the present findings indicate that pedometers are not acceptably accurate for quantifying physical activity in frail, institutionalized older adults with characteristically slow walking paces (i.e., <0.9 m·s−1) and gait impairments. Although pedometers consistently underestimate steps taken during slow walking, healthy community-dwelling older adults normally walk at a faster pace; therefore, pedometry can be confidently used with this segment of the population. Future research is needed to develop accurate and feasible options for quantifying physical activity in mobility-impaired populations.
The authors do not have a professional relationship with companies or manufacturers who may benefit from the results of the present study. The results of the study do not constitute endorsement of the product by the authors or ACSM.
The authors would like to acknowledge the assistance of Shannon Keenor with data collection.
1. Ainsworth, B. E., A. S. Leon, M. T. Richardson, D. R. Jacobs, and R. S. Paffenbarger. Accuracy of the college alumnus physical activity questionnaire. J. Clin. Epidemiol. 46: 1403–1411, 1993.
2. Algase, D. L., B. Kupferschmid, C. A. Beel-Bates, and E. R. A. Beattie. Estimates of stability of daily wandering behavior among cognitively impaired long-term care residents. Nurs. Res. 46: 172–178, 1997.
3. Bassett, D. R. Validity and reliability issues in objective monitoring of physical activity. Res. Q. Exerc. Sport 71: 30–36, 2000.
4. Bassett, D. R., B. E. Ainsworth, S. R Leggett, et al. Accuracy of five electronic pedometers for measuring distance walked. Med. Sci. Sports Exerc. 28: 1071–1077, 1996.
5. Bassett, D. R., B. E. Ainsworth, A. M. Swartz, S. J. Strath, W. L. O’Brien, and G. A. King. Validity of four motion sensors in measuring moderate intensity physical activity. Med. Sci. Sports Exerc. 32: S471–S480, 2000.
6. Bassey, E. J., H. M. Dallosso, P. H. Fentem, J. M. Irving, and J. M. Patrick. Validation of a simple walking mechanical accelerometer (pedometer) for the estimation of walking activity. Eur. J. Appl. Physiol. 56: 323–330, 1987.
7. Cohen-Mansfield, J., P. Werner, W. J. Culpepper, M. Wolfson, and E. Bickel. Assessment of ambulatory behavior in nursing home residents who pace or wander: A comparison of four commercially available devices. Dement. Geriatr. Cogn. Disord. 8: 359–365, 1997.
8. Cunningham, D. A., D. H. Paterson, J. E. Himann, and P. A. Rechnitzer. Determinants of independence in the elderly. Can. J. Appl. Physiol. 18: 243–254, 1993.
9. Freedson, P. S., and K. Miller. Objective monitoring of physical activity using motion sensors and heart rate. Res. Q. Exerc. Sport 71: 21–29, 2000.
10. Hendelman, D., K. Miller, C. Baggett, E. Debold, and P. Freedson. Validity of accelerometry for the assessment of moderate intensity physical activity in the field. Med. Sci. Sports Exerc. 32: S442–S450, 2000.
11. Himann, J. E., D. A. Cunningham, P. A. Rechnitzer, and D. H. Paterson. Age-related changes in speed of walking. Med. Sci. Sports Exerc. 20: 161–166, 1988.
12. King, A. C., J. Rejeski, and D. M. Buchner. Physical activity interventions targeting older adults: A critical review and recommendations. Am. J. Prev. Med. 15: 316–333, 1998.
13. Kochersberger, G., E. Mcconnell, M. N. Kuchibhatla, and C. Pieper. The reliability, validity, and stability of a measure of physical activity in the elderly. Arch. Phys. Med. Rehabil. 77: 793–795, 1996.
14. Lazowski, D. A., N. A. Ecclestone, A. M., Myers, et al. A randomized outcome evaluation of group exercise programs in long-term care institutions. J. Gerontol. A Biol. Sci. Med. Sci. 54: M621–M628, 1999.
15. Le Masurier, G., and C. Tudor-Locke. Comparison of pedometer and accelerometer accuracy under controlled conditions. Med. Sci. Sports Exerc. 35: 867–871, 2003.
16. Macko R. F., E. Haeuber, M. Shaughnessy, et al. Microprocessor-based ambulatory activity monitoring in stroke patients. Med. Sci. Sports Exerc. 34: 394–399, 2002.
17. Macrae, P. G., J. F. Schnelle, S. F. Simmons, and J. G. Ouslander. Physical activity levels of ambulatory nursing home residents. J. Aging Phys. Act. 4: 264–278, 1996.
18. Myers, A. M. Program Evaluation for Exercise Leaders. Champaign IL: Human Kinetics, 1999, pp. 1–149.
19. Reuben, D. B., and A. L. Siu. An objective measure of physical function of elderly outpatients: the Physical Performance Test. JAGS 38: 1105–1112, 1990.
20. Richardson, M. T., A. S. Leon, D. R. Jacobs, B. E. Ainsworth, and R. Serfass. Comprehensive evaluation of the Minnesota Leisure Time Physical Activity Questionnaire. J. Clin. Epidemiol. 47: 271–281, 1994.
21. Sallis, J. F., and B. E. Saelens. Assessment of physical activity by self-report: Status, limitations, and future directions. Res. Q. Exerc. Sport 71: 1–14, 2000.
22. Schmalzried, T. P., E. S. Szuszczewicz, M. R. Northfield, et al Quantitative assessment of walking activity after total hip or knee replacement. J. Bone Joint Surg. 80-A: 54–59, 1998.
23. Shepherd, E. F., E. Toloza, C. D. Mcclung, and T. P. Schmalzried. Step activity monitor: increased accuracy in quantifying ambulatory activity. J. Orthop. Res. 17: 703–708, 1999.
24. Steffen, T. M., T. A. Hacker, and L. Mollinger. Age- and gender-related test performance in community-dwelling elderly people: Six-minute Walk Test, Berg Balance Scale, Timed Up & Go Test, and gait speeds. Phys. Ther. 82: 128–137, 2002.
25. Tinetti, M. E. Performance-oriented assessment of mobility problems in elderly patients. JAGS 34: 119–126, 1986.
26. Tudor-Locke, C. Taking steps toward increase physical activity: using pedometers to measure and motivate. Res. Digest 3: 1–8, 2002.
27. Tudor-Locke, C., B. E. Ainsworth, R. W. Thompson, and C. E. Matthews. Comparison of pedometer and accelerometer measures of free-living physical activity. Med. Sci. Sports Exerc. 34: 2045–2051, 2002.
28. Tudor-Locke, C., and A. M. Myers. Challenges and opportunities in measuring physical activity in sedentary adults. Sports Med. 31: 91–100, 2001.
29. Tudor-Locke, C., and A. M. Myers. Methodological considerations for researchers and practitioners using pedometers to measure physical (ambulatory) activity. Res. Q. Exerc. Sport 72: 1–12, 2001.
30. Yusuf, H. R., J. B. Croft, W. H. Giles, et al. Leisure-time physical activity among older adults: United States, 1990. Arch. Int. Med. 156: 1321–1326, 1996.