In both radiographic analyses and implant-retrieval studies, the rate of polyethylene wear has traditionally been reported as a function of time. Tremendous variability in rates of wear has been observed2,4-6,8,10-12,16,20,27,28. The reported rates of radiographic wear of the polyethylene bearing of Charnley total hip replacements have ranged from zero to 0.6 millimeter per year4-6,10,12,16,28. However, wear is not a function of time. A prosthetic joint is similar to an automobile tire in that wear is a function of use, or the number of cycles. In laboratory tests, wear rates are compared on the basis of a measured number of cycles7,8. In clinical studies, the fundamental variable of the number of cycles has been replaced by duration in situ. As joint-replacement prostheses have been implanted in a wide range of patients, the variability in use of the joint needs to be assessed.
It has been stated that walking is the single most important physical activity affecting the wear of knee and hip prostheses21. Methods for measuring activity include calorimetry, job classification, survey procedures, physiological markers, behavioral observation, and mechanical or electronic monitors such as pedometers13,14,17-19.
The pedometer is a step-counter consisting of a pendulum that swings with each step and registers cycles on a counting mechanism. The pendulum mechanism is driven by oscillation of the pelvis during gait. Pedometers were used as early as the 1920s15. Modern pedometers are small devices worn inconspicuously on the beltline. They are inexpensive, are easy to use, and have been found to be reliable for assessments of walking activity1.
The pedometer may provide a satisfactory means of quantifying the use of prosthetic joints in the lower extremity. The goals of the present study were to assess the accuracy of a pedometer with regard to counting steps for a variety of walking activities, to establish a database of gait cycles for patients who have had at least one total joint arthroplasty of the hip or knee, and to compare the walking activity of subgroups of patients.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding sources were The Piedmont Fund of Los Angeles Orthopaedic Foundation and Sportline Company, Campbell, California.
†Joint Replacement Institute at Orthopaedic Hospital, 2400 South Flower Street, Los Angeles, California 90007. E-mail address for Dr. Schmalzried: 102467.3476@compuserve.com.
‡Harbor-University of California at Los Angeles Medical Center, 1000 West Carson Street, Torrance, California 90509.
An electronic, digital pedometer (Sportline, Campbell, California) was used to record the number of steps taken daily. This device was chosen because of its comparatively low cost (less than fifteen dollars) and ease of use. The device records the combined steps of both lower extremities.
Accuracy
Twenty-five healthy volunteers were asked to walk briskly around a track for a distance of 400 meters to simulate walking in the community. The subject, accompanied by one of the investigators, counted aloud the number of steps taken while the investigator counted simultaneously. The number of steps recorded by the pedometer for the 400 meters and the number of steps counted by the subject and the investigator (the actual number of steps taken) were both recorded. Each subject completed two trials of 400 meters.
The volunteers were then asked to walk ten meters at a slow pace to simulate walking in the household. Again, both the subject and an investigator counted the actual number of steps, and this number was compared with the number recorded on the pedometer. The subject completed two trials of this distance.
To evaluate the performance of the pedometer on stairs, the actual number of steps taken while ascending and then while descending a single flight of twelve stairs was compared with the number recorded by the pedometer. As with the previous measurements, the subject completed two trials of ascending and descending the stairs.
The accuracy was calculated as 100 - ([|the number of steps recorded - the actual number of steps|/the greater of the two values] x 100). For example, if the subject took sixteen steps and the pedometer registered fourteen, the accuracy would be 100 - ([|14 - 16|/16] x 100) = 87.5 per cent. If the difference between the number of steps recorded and the actual number of steps was negative, the pedometer undercounted the number of steps. If the difference was positive, the pedometer overcounted the number of steps.
The accuracy of the pedometer averaged 96 per cent (range, 82 to 99 per cent) for the fifty trials of 400 meters of brisk walking. The pedometer overcounted twenty-four times and undercounted twenty-six times. The greatest instance of overcounting was 15 per cent for an eighty-six-year-old woman, and the greatest instance of undercounting was 18 per cent for a thirty-two-year-old woman who was obese.
For the fifty trials of the ten-meter slow walk, the average accuracy was 92 per cent (range, 70 to 100 per cent). The pedometer overcounted twenty-six times, undercounted fourteen times, and was exact ten times. The greatest instance of overcounting was 25 per cent for a thirty-one-year-old woman, and the greatest instance of undercounting was 30 per cent for a forty-four-year-old woman.
With regard to ascending the flight of twelve stairs, the pedometer registered an average of 11.5 steps, for an average accuracy of 96 per cent. The pedometer overcounted fourteen times, undercounted twenty-one times, and was exact fifteen times. For descending the flight of twelve stairs, the pedometer registered an average of 11.7 steps, for an average accuracy of 98 per cent. The pedometer overcounted sixteen times, undercounted seventeen times, and was exact seventeen times.
Quantitative Assessment of Walking Activity in Patients Who Had Had a Joint Replacement
This study was not based on a consecutive series of patients. Instead, the patients came from the practices of two of us (T. P. S. and H. C. A.). The criteria for inclusion in this study were interest on the part of the patient, a duration of at least six months since a total hip or total knee replacement, the ability to walk independently in the community, and a body habitus and gait that were conducive to use of a pedometer. Experience obtained during the accuracy phase of testing of the device indicated that potential sources of inaccuracy were related to improper placement of the device either too close to the midline of the body or on clothes that rode too far off the waistline or were too stiff, such as a thick belt, and consequently pelvic oscillations were not transmitted uniformly. In obese subjects, the soft tissue over the iliac crest may diminish the amplitude of pelvic oscillations transmitted to the pedometer, resulting in undercounting. Small or so-called shuffling steps may not be recorded accurately. Other investigators have made this observation26. We consistently informed the patients of these limitations and did not enroll any patient whose body habitus or gait pattern inhibited function of the pedometer.
Written instructions and a recording sheet were distributed at the time of enrollment in the study. We modified the Charnley functional classification4 such that any patient who had osteoarthrosis affecting an uninvolved hip or knee or had had an arthroplasty of any two joints was designated as class B. All patients were instructed and supervised by one of the investigators regarding the proper use of the pedometer. If the pedometer was recording inaccurately, its position was adjusted to improve accuracy. In addition to recording steps, the pedometer may also record other movements of the torso or lower extremities, such as rising from or assuming a seated position. However, these motions make a comparatively small contribution to the total pedometer recording.
The patients were asked to wear the pedometer from the time that they got up in the morning until they went to bed at night. Each patient recorded the time that the pedometer was put on in the morning as well as the time that it was taken off at night and recorded the number of steps per day as counted by the pedometer. The patients were asked to wear the pedometer for seven days that were typical or representative of their usual activity. A day during which the pedometer was not used for a period of time, such as if it were not transferred to different clothes, did not qualify as a representative day.
The pedometer protocol was completed by 111 patients who had had at least one total hip or knee replacement. The average age was fifty-nine years (range, twenty-three to eighty-two years) overall, sixty years (range, twenty-five to eighty years) for the men, and fifty-seven years (range, twenty-three to eighty-two years) for the women. The average number of steps per day was determined for each patient. The patients wore the pedometer for an average of fifteen hours a day for eight days.
The relationship between the average number of steps per day and independent variables was examined with linear regression analysis. The Student t test was performed to analyze differences in the numbers of steps taken by different groups of patients.
The 111 patients averaged 4988 steps per day. The least active patient, a seventy-one-year-old woman, averaged 395 steps per day whereas the most active patient, a seventy-year-old former Olympic gymnast, averaged 17,718 steps per day. This range represents a forty-five-fold difference in the activity levels of the patients. Extrapolation of the data indicates that the patients were averaging approximately 0.9 million cycles per year (calculated as [the number of steps per day/2] x 365) for each joint of the lower extremity. The most active patient averaged 3.2 million cycles per year whereas the least active patient averaged 72,000 cycles per year.
Regression analysis revealed that the independent variable of age was significantly associated with activity (p = 0.048) (Fig. 1); however, there was a high degree of variability (standard deviation, 3040 steps per day). The equation derived from the linear regression for average steps per day based on age was: 7525 - (43.3 x age of the patient in years).
The forty-nine patients who were less than sixty years old averaged 5732 steps per day, and the sixty-two patients who were sixty years old or more averaged 4400 steps per day (Table I). This 30 per cent difference in walking activity was significant (p = 0.023). The fifty-seven men averaged 5579 steps per day, and the fifty-four women averaged 4364 steps per day (Table I). This 28 per cent difference in walking activity was also significant (p = 0.037). The twenty-three men who were less than sixty years old were, on average, 40 per cent more active than the remaining patients; again, this was a significant difference (p = 0.011).
The sixty-four patients in Charnley4 class A averaged 5428 steps per day; the thirty-three patients in class B, 4429 steps per day; and the fourteen patients in class C, 4277 steps per day. The data indicate a tendency toward greater activity for the patients who were in Charnley class A, but with the numbers available no significant difference was detected at the 95 per cent confidence level (p = 0.081). The average activity of the patients who had rheumatoid arthritis (4537 steps per day) was only slightly less than that of the patients with other diagnoses (5020 steps per day) (p = 0.668). The lack of a significant difference is probably due to the small number of patients who had rheumatoid arthritis (eight) and to the inclusion criteria, which led to the selection of patients who had higher function. The data also suggest that the patients who had a total hip replacement were more active (5194 steps per day) than those who had a total knee replacement (3514 steps per day) (p = 0.056) (Table I). This trend should be interpreted with caution in view of the comparatively small number of patients who had a total knee replacement (fourteen).
The average weight of the 111 patients was seventy-six kilograms (range, forty-four to 123 kilograms); the men weighed an average of eighty-five kilograms (range, fifty-eight to 123 kilograms) and the women, an average of sixty-seven kilograms (range, forty-four to 109 kilograms). Regression analysis revealed no association between weight and the average number of steps per day in the series as a whole (p = 0.75), but when the men and women were analyzed separately there was a trend for walking activity to decrease with increasing weight, especially for the women (p = 0.17), with a stronger association when the weight of the women was adjusted for height (weight divided by height) (p = 0.099).
Although it is generally believed that the activity level of the patient influences the wear and survival of prosthetic joints3,12,20, there has been a paucity of studies on the measurement of such activity. Seedhom et al.22 used a pedometer to quantify the walking activity of nine healthy elderly individuals who were on vacation; none had a prosthesis. The average daily activity of these subjects extrapolated to approximately one million gait cycles per year. This appears to be the source of the estimation used in laboratory wear simulations that one million cycles is equivalent to one year of in vivo use7. Wallbridge and Dowson used a pedometer to study the walking activity of eleven randomly selected patients who had a hip prosthesis; some had had a hemiarthroplasty. The average number of cycles per day was 3975, which extrapolates to 1.45 million cycles per year. Wallbridge and Dowson suggested that the average number is of limited value and that the most important information is provided by the most active patients. The most active patient in their study was a forty-four-year-old woman who averaged about 2.1 million cycles per year.
The average accuracy of the pedometer used in this study was found to be 92 per cent or better for a variety of walking activities; the lowest accuracy for the 400-meter walk was 82 per cent and that for the ten-meter walk was 70 per cent. There did not appear to be any consistent pattern of overcounting or undercounting. The accuracy of the pedometer was high for ascending and descending stairs, probably because the gait pattern was determined by the design of the stairs and was therefore very uniform.
Our study of 111 non-randomized volunteers demonstrated an average activity of slightly more than 0.9 million cycles per year. Because of the limitations of our methods, these results are only an approximation of joint use. It is not practical to wear the pedometer twenty-four hours a day, and consequently some gait cycles will be missed. The number of gait cycles that we report is probably an underestimation of the actual number of cycles; thus, our findings are consistent with an estimate on the order of one million cycles per year7.
The most important result of this study, however, is documentation of the wide range of walking activity. The most active patient in the study subjected the prosthesis to about 3.2 million cycles per year. The average number of steps per day for this patient was more than 3.5 times the average and forty-five times more than that for the least active patient in the study.
Young patients and men have generally been considered at increased risk for aseptic loosening3,9,23,24, and this appears to be related, at least in part, to the greater average activity of these patients. In the current study, age was significantly associated with walking activity (p = 0.048), but there was a high degree of variability (standard deviation, 3040 steps per day). On average, the patients who were less than sixty years old were more active than those who were sixty years old or more (p = 0.023). There was no significant difference in age between the men and the women, but the men were significantly more active than the women (p = 0.037). The men who were less than sixty years old were, on average, 40 per cent more active than the other patients in the study (p = 0.011). The most active patient in the present study, interestingly enough, was seventy years old. Some elderly patients are very active and place exceptional demands on the prosthesis from a wear perspective. This indicates that age is not the salient criterion for selection of a prosthesis in so-called demand-matching programs.
There was a tendency for the patients in Charnley4 functional class A to be more active than those in class B or C. There was no significant decrease in the walking activity of the patients who had rheumatoid arthritis, but this is probably because of the small number of patients who had rheumatoid arthritis and to the inclusion criteria, which led to the selection of patients who had a higher level of function. The data suggest that relative obesity has a downward effect on activity, especially for women. It is likely that true obesity is associated with reduced activity, but our inclusion criteria eliminated truly obese patients. The data also suggest that the patients who had a total hip replacement were more active than those who had a total knee replacement, but this result should be interpreted with caution in view of the small number of patients who had a total knee replacement.
The patients in the current study engaged in less walking activity than healthy, normal subjects21,22,25. The slope of the linear regression line in the current study (steps per day = 7525 - [43.3 x age of the patient in years]) was less steep than that reported by Wallbridge and Dowson for healthy, normal subjects (steps per day = 18,000 - [175.6 x age of the patient in years]). This suggests that the presence of a hip or knee prosthesis is associated with a greater decrease in the activity of a young patient than it is in an older patient. An alternative or additional explanation could be that our inclusion criteria led us to select relatively more active older patients.
Studies of total joint replacement, including radiographic analyses of implant wear, generally include data on the ages of the patients, and these data have been used as an indirect assessment of activity. Because of the association between age and activity documented by pedometer studies25, the linear regression equation of the present study as well as that of Wallbridge and Dowson for healthy, normal subjects can be used to convert the average age of the patients in a cohort study to an average number of steps per day for that cohort. This increases the validity of comparisons of rates of wear in different studies by accounting for differences in age.
Clinical studies have documented a wide range in the rate of polyethylene wear2,4-6,8,10-12,16,20,27,28. In virtually every study, there are hips that have no measurable wear and hips that have wear that is several times more than the average for that study. The in vivo wear of polyethylene is multifactorial. Attention has generally been focused on variables related to the reconstruction of the hip, such as the type of fixation of the implant as well as design and manufacturing variables related to the femoral head or the polyethylene bearing, or both2,8,11,12,16,20,27. The pedometer data in the present study indicate that individual differences in the activity of the patients are a substantial source of variability. All other factors being equal, a forty-five-fold difference in the rate of wear as well as a maximum rate of more than three times the average can be accounted for simply by differences in the activity of the patients.
The pedometer has many potential applications for orthopaedic research. Pedometer data can be used in conjunction with radiographic assessments to standardize rates of wear of joint prostheses as a function of gait cycles rather than time. This quantitative approach may aid in the prediction of the survival of joint replacements. Pedometer data may also be helpful for quantifying walking ability in outcome studies.