Accuracy of five electronic pedometers for measuring distance walked

The electronic pedometer is a simple device used to measure walking distance. This instrument has the potential to provide epidemiologists with a lowcost, objective measure of walking behavior, which accounts for a substantial fraction of the calories expended in physical activity^(5,7). Recently, Ainsworth et al.⁽²⁾ reported that some physical activity questionnaires may yield imprecise estimates of daily walking distance. For instance, on the Harvard alumnus questionnaire, subjects recalled walking an average of 0.6 miles·d^-1, while in their daily activity logs the walking distance was approximately 3.0 miles·d^-1. Richardson et al.⁽¹³⁾ showed similar patterns from the Minnesota Leisure Time Physical Activity Survey. These studies^(2,13) have served as the impetus for carrying out a validation study to determine whether electronic pedometers could accurately track distance and number of steps over a range of different conditions.

The 1993 Summary Statement of the Centers for Disease Control and Prevention (CDC) and the American College of Sports Medicine (ACSM) states that Americans should accumulate at least 30 min of moderate-intensity physical activity on most days of the week ⁽¹²⁾. They further recommend that this goal can be met by walking two miles briskly. The Japanese recommend that adults walk at least “10,000 steps/day” to maintain optimum health ^(5,16). These goals emphasize walking behaviors, and the Japanese recommendation can only be monitored through the use of electronic pedometers. Electronic pedometers are increasingly being used in physical activity research. In one recent study, obese diabetic patients were given electronic pedometers and instructed to walk at least 10,000 steps·d^-1(16). They walked an average of 19,200 ± 2,100 steps·d^-1 (mean± SE) with beneficial effects on weight loss and insulin sensitivity⁽¹⁶⁾. The sale of pedometers in sporting good stores shows their acceptance by the general population. Thus, practitioners may find that these devices are useful as a motivational tool in exercise prescription.

The question is, how accurate are these commercial pedometers in measuring distance walked? The electronic pedometers tested in the present study are worn on the belt or waistband. They all operate on the same principle-a horizontal, spring-suspended lever arm moves up and down with vertical accelerations of the hip. With each step, the lever arm makes an electrical contact and one event is recorded (Fig. 1). Previous research on the older, mechanical style pedometers was carried out by Gayle et al. ⁽⁴⁾, Montoye ⁽¹¹⁾, Washburn et al. ⁽¹⁵⁾, Bassey ⁽³⁾, and Kemper and Verschuur ⁽⁸⁾. In general, researchers have concluded that the large errors involved with the gear-driven, mechanical pedometers make them unsuitable for precise work ^(9,11). Limited information is available on the accuracy of the newer electronic pedometers ^(5,10). The purpose of the present study was to examine the accuracy of five brands of electronic pedometers for measuring distance walked and number of steps. The goals of this three-part study were: (a) to examine the accuracy of five different pedometers over a 3.03-mile (4.88-km) sidewalk course; (b) to compare the accuracy of the pedometers on different surfaces (concrete sidewalk versus rubber track); and(c) to examine the effects of walking speed on pedometer accuracy during treadmill walking at 2.0-4.0 mph (54-107 m·min^-1).

METHODS

Pedometers

Five models of commercially available electronic pedometers were tested(Fig. 2)-Freestyle Pacer 798 (Camarillo, CA), Eddie Bauer Compustep II (Redmond, WA), L.L. Bean Pedometer (Freeport, ME), Yamax Digi-walker DW-500 (Tokyo, Japan), and the Accusplit Fitness Walker (San Jose, CA). All of these pedometers measure vertical oscillations, and all, excluding the Yamax DW-500, utilize the individual subject's stride length to compute the distance walked. The decision to use the Yamax DW-500 was based upon the fact that this model is being used by researchers at the Aerobics Center in Dallas, the Department of Epidemiology and Biostatistics at the University of South Carolina, and the Center for Health Promotion in Indians in New Mexico. Several other Yamax models (DW-400, DW-450, DW-480, DW-550, DW-580) do have a stride length function, but they were not selected for study inclusion. The Yamax DW-500 was selected based upon its growing use in research studies.

Subjects

To obtain results that would be generalizable to a broad population, both male and female adults (18-65 yr of age) were asked to participate in the study. Twenty persons (13 females and 7 males) volunteered for Part I of the study. A subset of 10 of them also volunteered for Part II. Another group of 10 different subjects (five females and five males) participated in Part III of the study. Written informed consent was obtained from all subjects prior to their participation in the study. Age was recorded, and height and weight were measured in one layer of street clothing (without shoes). To characterize the subjects' physical profile, skinfold measurements were taken at three sites using a Lange skinfold caliper, and percent body fat was estimated from the formulas provided by Jackson and Pollock ⁽⁶⁾. The subjects' physical characteristics are shown in Table 1.

Part I-Accuracy of Pedometers on Sidewalks

The accuracy of the pedometers in measuring distance walked and number of steps taken was determined by having the subjects cover a 3.03-mile (4.88-km) level course on cement sidewalks. The sidewalk course length was measured twice and determined to be 3.02879 and 3.02784 miles by using a calibrated measuring wheel (Rolatape Corporation, Model 400, Spokane WA). Each subject completed six different trials on different days, wearing the same pair of shoes for all trials. Table 2

Subjects wore two pedometers of the same brand (on left and right sides) for each trial. Pedometer placement was standardized by placing them on the belt or waistband, approximately 5-7 cm from the umbilicus. This placement was consistent with the manufacturer's instructions (which all specified the“belt or waistband” and the mid-line of the thigh or “as close to the center of the body as possible”).

The stride length was measured to compute distances walked. To determine stride length, subjects were asked to take 20 strides at their normal walking pace, in an indoor hallway. The distance covered was divided by 20, to compute the average stride length. This value was entered into the electronic pedometers. The L.L. Bean and Accusplit pedometers, however, did not allow the stride length to be entered in this fashion. Instead, the manufacturer's instructions were followed. This involved a calibration routine where the subject walked a fixed distance while the pedometer calculated the stride length. Pedometers with a variable sensitivity switch were always placed in the highest setting. Some pedometers had two sensitivity settings-“walk” and “run.” The Pacer had a number of settings ranging from “+” (more sensitive) to “-”(less sensitive). Adjusting the pedometer sensitivity setting alters the amount of vertical acceleration at the hip that is needed to register a footstrike.

The actual number of steps taken during the 4.88-km walk was determined using a hand-tally counter. Subjects clicked the counter once for each left foot heelstrike, and the final number was doubled to give the total number of steps. Each subject walked at his/her self-selected velocity, and the amount of time it took to complete the walk was recorded.

Part II-Effects of Walking Surface on Pedometer Accuracy

To examine the effects of variations in the “hardness” of walking surfaces, 10 of the original 20 subjects completed additional trials with all five pedometers on rubberized track surface. A 400-m outdoor track was used for this purpose. To obtain a distance equal to 3.03 miles (4.88 km) the subjects walked 12 laps around the track, doing two laps in lane 1 (400 m·lap^-1) and 10 laps in lane 2 (407.5 m·lap^-1). They walked counter-clockwise around the track, approximately 10 inches (25 cm) to the right of the inside lane marker.

Part III-Effects of Walking Speed on Pedometer Accuracy

To examine the effects of walking speed on the accuracy of three pedometers(Freestyle Pacer, Eddie Bauer, and Yamax) 10 different subjects walked at various speeds on a motor-driven treadmill (Quinton model Q55XT, Seattle, WA). The treadmill speed was calibrated by measuring the belt length (3.190 m) and measuring the time it took to complete 25 revolutions of the treadmill belt. A carpenter's level was used to calibrate the treadmill grade to 0.0%, according to the manufacturer's instructions. The accuracy of the carpenter's level itself was checked by turning it horizontally 180°, and observing that the bubble was still centered.

The order in which the pedometers were worn was randomized to account for the possibility of an “order effect.” All pedometers were worn on the right side of the body using the landmarks specified previously. Stride length was first measured in the hallway at a “normal walking speed,” and then on the treadmill at 3.0 mph (80.4 m·min^-1). The latter value was entered into the pedometer. Five speeds were used (2.0, 2.5, 3.0, 3.5, and 4.0 mph). The actual number of steps taken was determined by using a hand-tally counter, as previously described. Each stage lasted 5 min, and the stages were interrupted just long enough to increase the treadmill speed and record data. Dependent variables were stride length, stride frequency, the percentage of footstrikes recorded, and the percentage of actual distance covered.

Statistical Analysis

In Part I of this study (Accuracy of Pedometers on Sidewalks), comparisons of distance and percentage of actual steps recorded were analyzed using contrasts ⁽¹⁾. Planned contrasts were carried out between each of the five different pedometer brands, and between the same-brand pedometers worn on the left and right sides of the body.

In Part II (Effects of Walking Surface on Pedometer Accuracy), comparisons of percent difference scores for distance (sidewalk-track) were analyzed using contrasts. Planned contrasts were carried out between each of the five pedometer brands, and between the same-brand pedometers worn on the left and right sides of the body.

In Part III (Effects of Walking Speed on Pedometer Accuracy), comparisons of percentage of actual steps recorded and percentage of actual distance were analyzed using contrasts. We wanted to determine if there was a significant difference between the three pedometer brands (Yamax, Pacer, and Eddie Bauer), at each velocity. Hence, three contrasts were carried out at five walking velocities, for a total of 15 contrasts. Two-sided comparisons and an overall significance level of P = 0.05 were used throughout.

RESULTS

Part I: Accuracy of Pedometers on Sidewalks

The mean scores for 20 subjects were computed for each brand of motion sensor. Figure 3 shows the mean distance recorded by each pedometer over the 3.03-mile (4.88-km) walking course, using pedometers worn on either the left or right side of the body. There were significant differences in distance walked between pedometers (see significance bars in figure legend). The Yamax, Pacer, and Accusplit were significantly more accurate than the L.L. Bean and Eddie Bauer models. The two Yamax pedometers showed very close agreement, while the Pacer showed significantly higher distance scores for the pedometer worn on the left side of the body(P = 0.0003) versus the right side, and the Accusplit showed a trend in this direction (P = 0.0657). For the three pedometers that displayed steps (Eddie Bauer, Pacer, and Yamax), there was no statistical difference in the percent of actual steps recorded. No differences were observed between trials in walking speed over the 4.88-km sidewalk course.

Part II: Effects of Walking Surface on Pedometer Accuracy

In general, the effect of walking surface (track versus sidewalk) on distance recorded was not significant (Fig. 4). However, the distance value for the left-side Pacer pedometer was 14% higher on the track, compared with the sidewalk (P = 0.0220). All other devices, regardless of which side of the body they were worn on, showed similar values for sidewalk and track surfaces (P > 0.05).

Part III: Effects of Walking Speed on Pedometer Accuracy

Three pedometer brands were compared to determine if there were differences in the percentage of steps recorded and the percentage of actual distance, at various speeds. At slow-to-moderate speeds (2.0-3.0 mph or 54-80 m·min^-1), the Yamax pedometer was more accurate than the Eddie Bauer and Pacer brands. At faster speeds (4.0 mph or 107 m·min^-1) the Eddie Bauer and Pacer accuracy improved, showing better agreement with the Yamax pedometer.

The average stride length at each of the walking speeds was as follows: 0.59 ± 0.03 m·stride^-1 (54 m·min^-1); 0.65± 0.02 m·stride^-1 (67 m·min^-1); 0.72± 0.03 m·stride^-1 (80 m·min^-1); 0.78± 0.04 m·stride^-1 (94 m·min^-1); and 0.86± 0.05 m·stride^-1 (107 m·min^-1) (mean± SD).

DISCUSSION

Part I of this study was designed to test the ability of pedometers to record walking distance and steps taken on a sidewalk course. The electronic pedometers were found to provide a reasonably accurate estimate of the distance walked and number of steps taken. The fact that several pedometers underestimated distance is probably due to a failure to register each footstrike. The Yamax was the most accurate, recording 100.7% (left) and 100.6% (right) of all steps taken. The Pacer recorded 102.6% and 87.5% of actual steps for the left and right pedometers, respectively. The Eddie Bauer recorded 94.0% (left) and 91.6% (right) of actual steps.

When the individual subject's stride length was taken into account, the average distance for two pedometers of the same brand was always within 0.53 km for the 4.88-km course (an 11% difference). The Yamax pedometer differed by only 0.05 km for the 4.88-km course (a 1% difference). The Yamax also had the smallest between-subject standard deviation (0.36 km). All of these pedometers respond to vertical accelerations of the body, and their sensitivity is determined by the “threshold” for vertical acceleration that is needed to trigger an event ⁽³⁾. Therefore, some brands are more sensitive than others (i.e., they have a lower threshold acceleration), resulting in an increased ability to register actual steps.

The variability in results from two units of the same model for the Pacer and Accusplit pedometers was a cause for concern. Earlier work on mechanical pedometers by Gayle et al. ⁽⁴⁾, Washburn et al.⁽¹⁵⁾, and Saris and Binkhorst ⁽¹⁴⁾ showed differences between two units of the same model, which they attributed to variations in spring tension. Our results indicate that the same concern exists with some of the newer electronic pedometers. Only the Yamax pedometer was very consistent between units. The inaccuracy between units of the same model is probably due to quality control in manufacturing. When all of the pedometers were considered together, the devices worn on the left side of the body did not differ significantly from those worn on the right. Even though the subjects made several right-hand turns when walking around the sidewalk course, this was not evidenced in significant differences between right and left pedometers. Hence, we conclude that it does not matter which side of the body they are worn on.

In Part II of the study, we compared differences in the walking surface on the accuracy of the pedometers. The major finding was that different walking surfaces did not affect pedometer accuracy. We had hypothesized that cushioned surfaces might decrease the impact force, and thus lessen the vertical acceleration experienced by the pedometers worn on the hip. However, most of the pedometers gave similar readings for distance, regardless of whether the subjects walked on a concrete sidewalk or cushioned, rubberized track surface.Figure 5

In Part III, we examined the effect of walking speed on three pedometers that provided information on the number of steps taken (Yamax, Eddie Bauer, and Pacer). The Yamax was more accurate at slower speeds compared with the other pedometers. This may reflect a high level of quality control-in the proposed Japanese Industrial Standards set by the Ministry of Industry and Trading Regulations, the maximum permissible rate of miscounting steps is 0.3%⁽⁵⁾. The Yamax met this specification during treadmill walking at speeds over 80 m·min^-1. However, at very slow speeds(e.g., 54 m·min^-1) all three brands tended to underestimate distance. This was due to a “failure to register” footstrikes, which was especially apparent in the Eddie Bauer pedometer (it registered only 40% of the steps taken at 2.0 mph). The inaccuracy results from the fact that vertical accelerations measured at the hip are less pronounced at slow speeds. Similarly, at faster speeds (e.g., 107 m·min^-1) the pedometers also underestimated distance. This error was due to a lengthening of the actual stride length rather than a miscounting of steps.

The electronic pedometer would appear to be useful for validating questions about “distance walked” on physical activity questionnaires. Despite the fact that variations in walking speed will affect the results, if a subject walks between 2.0 mph (54 m·min^-1) and 4.0 mph (107 m·min^-1), the most accurate pedometer brand (Yamax) should yield distance values that are within 20% of the actual distance(Fig. 6). In addition, if most of the walking is performed at intermediate speeds, the pedometer value should be within 10% of the actual distance. Respondents are reported to underestimate daily walking distance by as much as 5-fold on physical activity surveys⁽²⁾. Thus, even with an error rate of 10-20%, pedometers are likely to be considerably more accurate than a respondent's subjective estimate of walking distance. Electronic pedometers could be used as a criterion measure of “distance walked” to validate questions about walking on physical activity surveys.

Several other points are worth noting regarding the comparison of these pedometers. The Eddie Bauer and L.L. Bean have a maximum distance of 99.9 miles, while the Yamax has a maximum step count of 99,999 steps (approximately 50 miles or 80 km). The other pedometers are able to record maximum distances of 999.9 miles. Battery life according to the manufacturer varied substantially: Accusplit (7 yr), Eddie Bauer (“long-life battery”), Yamax (3 yr), L.L. Bean (not specified), and Pacer (1 yr). One Accusplit and one Pacer pedometer began to malfunction during the course of the study, and thus several trials had to be redone. This indicates that there may be differences in durability among the brands. However, we did not test a large enough sample of each brand to draw conclusions about durability. For epidemiologic studies, there may be an advantage to models that have a plastic cover that can be closed to prevent accidental resetting. The plastic cover on the Yamax and the Accusplit restricted the subject from viewing the display when it was closed. This could be advantageous in“blinding” the subject to the distance walked.

The generalizability of this study has some potential limitations. The results were determined on adults who were within the normal range for body weight and body fatness. Therefore, one should be cautious when extrapolating these findings to children or the extremely obese. Further research is needed to validate the pedometers in these groups. Additional studies are needed to determine the efficacy of pedometers in correlating walking behaviors with health outcomes in public health settings.

In summary, this study tested the accuracy of electronic pedometers. During overground walking at self-selected velocities, most pedometers were accurate to within 11% for distance. However, one pedometer brand (Yamax) measured the number of steps and distance to within 1% of actual on the 4.88-km sidewalk course. The walking surface (sidewalk versus track) did not affect the accuracy of most pedometers. Treadmill studies showed that the Yamax pedometer was more accurate than the Eddie Bauer and Pacer at slow-to-moderate walking speeds, though no differences were seen at the fastest speed (4.0 mph or 107 m·min^-1). The results indicate that some of the newer electronic pedometers have a greater absolute accuracy than the old-style mechanical pedometers. Electronic pedometers could prove to be useful in epidemiological studies of physical activity in free-living populations.