Journal Logo

Applied Sciences: Biodynamics

An analysis of hip joint loading during walking, running, and skiing

VAN DEN BOGERT, ANTON J.; READ, LYNDA; NIGG, BENNO M.

Author Information
Medicine & Science in Sports & Exercise: January 1999 - Volume 31 - Issue 1 - p 131-142
  • Free

Abstract

Total hip replacement (THR) is one of the most frequent and successful surgical operations for a serious disability (16,20). In early years, the procedure was typically performed on elderly patients, but with improvements in materials, geometry, and fixation, THR have become commonplace in younger patients (21). After THR surgery, many patients are typically able to return to participation in activities that were too painful before surgery (15,45).

Aseptic loosening is the most common complication after hip replacement surgery (16). Early reports of radiographic loosening of the femoral component were approximately 20% at 5 yr and 30 to 40% at 10 yr after surgery (42,44). In more recent reports, with improved fixation, the rate of femoral stem loosening is as low as 2% to 7% at 10 yr or more (2,28). However, recent studies have also reported a much higher incidence of acetabular loosening than femoral stem loosening (2). Several studies have indicated a relationship between high activity levels and increased rates of implant failure (7,17,22). Others have found that low impact activity did not increase the risk of loosening (15,45). Dubs and coworkers (15) found a higher incidence of revision surgery among the group of patients with no physical activity compared with a group who participated in sports. It has even been speculated that physical activity reduces the risk of implant failure by preventing bone resorption around the stem of the implant (46). There is general agreement that THR patients should not participate in high-impact activities such as running (44,48), but there is controversial opinion as to whether patients should participate in activities such as cross-country and alpine skiing. The National Institutes of Health recently recommended (29): "Investigations are needed into [...] activity limitations, or types of physical effort that contribute to extended prosthesis survival. Physical conditioning activities-[...] exercises that enhance bone integrity without affecting fixation-need to be studied as they relate to the anticipated life-style and occupational objectives of the patient."

Data on loading of the hip joint have previously been collected using two different methods: direct measurement and inverse dynamics analysis. Direct measurement involves the use of an instrumented prosthesis, which measures the internal hip joint forces and transmits the data telemetrically (4,12). This invasive procedure is costly and has only been applied in a small number of elderly THR patients. Consequently, only data on daily living activities such as walking, jogging, and stair climbing has been published and this only for elderly subjects. Inverse dynamics analysis is noninvasive and involves measurements of the movement of the body segments and the forces between foot and ground. A mathematical model, using equations of motion for a multirigid segment system, is used to determine intersegmental forces. Models for muscle lines of action are then used to distribute the intersegmental loads to the various structures crossing the joint. Because the number of unknown forces exceeds the number of equations of motion, additional assumptions are required to achieve a solution (1). This method, pioneered by Paul (33) and developed further by others (10,39), has been used to estimate bony contact forces at the hip joint for walking, stair climbing, and descending and rising from a chair. This method requires a gait laboratory and is not generally applicable to the analysis of sports activities. Hip joint loading has, to our knowledge, not been quantified in activities such alpine or cross-country skiing.

Prosthetic implants are typically tested under loading conditions similar to gait. However, with replacements now being performed on younger and more active patients, there are increased demands on the prosthesis. Hip joint forces have only been quantified for activities of daily living (4,10,12). Little is known about hip joint loading during the variety of activities that patients may now participate in, and the knowledge of the actual loading conditions during activities such as skiing will allow development of test protocols for implants that correspond to actual loading during sport activities.

The objective of the present study was to quantify hip joint loading during walking, running, and cross-country and alpine skiing. Because walking is typically recommended as a "safe," and running as an "unsafe" activity for a THR patient, the loading conditions of the various skiing activities are compared with walking and running. This information will be used to make recommendations to patients with reference to "safe" and "unsafe" activities and to provide the loading conditions for future mechanical testing of hip joint prosthetic components.

MATERIALS AND METHODS

Data collection. Nine male subjects volunteered to participate in the study and gave written informed consent in accordance with the policy statement of the American College of Sports Medicine. The average age of the subjects was 41 yr, with three subjects older than 65 yr. All subjects were physically active and intermediate to expert alpine skiers. The subjects had normal, healthy hip joints. As described in more detail elsewhere (5), the measuring equipment included four triaxial accelerometers (ENTRAN, model EGAXT -*-10) attached to a semirigid frame tightly strapped to the upper body. Three accelerometers were mounted on the back, and one accelerometer was mounted on the front of the frame (Fig. 1). The 12 channels of accelerometer data were recorded by a portable data acquisition system (Tattletale 6F, Onset Computer Corporations, North Falmouth, MA) carried in a pack around a subject's waist. With a 500-kB memory capacity, the data logger could store 69 s of continuous recording at a sampling rate of 300 Hz. The data logger was triggered by a telemetry signal allowing the subject to be in motion at the start of data collection. The mass of the entire system including frame, accelerometers and data logger was 1.5 kg. Each subject performed four different activities: walking, running, and alpine and cross-country skiing. The alpine skiing involved four different conditions based on the hill terrain and radius of ski turns. The cross-country skiing involved two different techniques, classic and skating. These eight different activities or conditions are summarized in Table 1.

Figure 1
Figure 1:
Front and back view of subject wearing the accelerometer system. Three triaxial accelerometers are mounted on the back, one on the front. Signals are recorded by a miniature data logger carried in the belt pack. The trunk coordinate system with the origin at the hip joint is also shown.
TABLE 1
TABLE 1:
Summary of activities and conditions included in the study.

For each subject, the study required two data collection sessions on separate days, one in the laboratory and one on the ski hill. In the laboratory session the accelerometer frame was first fitted to the subject. Measurements from the frame to anatomical landmarks were taken to ensure that the accelerometers would have the same locations in the second data collection session at the ski hill. Three to four reflective markers were attached at known positions to each triaxial accelerometer, and 3-D coordinates of these markers were measured during a well-defined standing position using an automated 3-D video system (Motion Analysis Corp., Santa Rosa, CA). Additionally, coordinates for the center of mass of the upper body and left hip joint center were determined from markers attached to anatomical landmarks (sternum, anterior superior iliac spines, greater trochanter). After completing these "subject calibration" measurements, the reflective markers were removed.

Accelerometer data were then collected for walking and running activities. All data were collected at a frequency of 300 Hz and a duration of 5 s for running and 10 s for walking. Three trials were obtained for each activity. Trials that were not within 10% of the required speed (Table 1) were repeated. At the beginning and end of each data file, signals were recorded in the previous well-defined standing position. This was to check for any drift in the accelerometer signals and to obtain the reference signals for data processing. The data were then transferred to a portable Toshiba laptop computer (T1850C) for storage.

Skiing data were collected at a local ski area (Canada Olympic Park) covered with manmade, groomed snow. Two trials of each ski activity were collected for a duration of 22 s. The subject was asked to perform the alpine skiing on one leg, that is, lifting the inside ski as they made each turn. The cross-country skiing was performed on groomed cross-country trails at the base of the ski hill. The classical cross-country trials were performed without ski poles. The subject started skiing with poles, but the poles were dropped before the start of data collection. To have a visual record of the movements analyzed, the skiing trials were also filmed using a Hitachi video system (30 frames·s−1). The video was synchronized with the accelerometer signals by a LED attached to the camera lens and triggered by the same signal as the data logger.

Because the local ski hill did not offer the variety of conditions available in the mountains, particularly moguls, a third data collection session with only two subjects was performed. For this session, data were collected at a large mountain resort (Lake Louise). Each of the two subjects performed trials in the following ski conditions: large and small moguls, natural and manmade snow, and beginner and steep terrain. Additionally, data were collected for a reference trial on a intermediate, groomed slope with the skier performing short radius turns.

Data analysis. For analysis, the data were transferred to a SUN SparcStation IPC computer. Intersegmental forces and moments at the hip joint, acting on the upper body, were calculated from the accelerometer data according to the method described in detail elsewhere (5). Briefly, the method was based on rigid-body kinematics and dynamics. With four triaxial accelerometers attached in known position and orientation to one rigid body (trunk segment), the linear acceleration, angular velocity, and angular acceleration of the segment could always be uniquely estimated by a nonlinear least-squares method. These three kinematic quantities were expressed in a body-fixed, moving coordinate system, and the linear acceleration included the gravitational acceleration vector in this coordinate system. Rigid-body dynamics was then used to calculate the intersegmental force and moment transmitted through the left hip joint. The least squares solution of the overdetermined system of equations resulted in residuals that indicated to which extent the accelerometer signals were consistent with the rigid-body assumption.

The coordinate system was a trunk segment coordinate system with the origin at the left hip joint: (x) was positive lateral, (y) was positive posterior, and (z) was positive superior (Fig. 1). Positive force components (Fx, Fy, Fz) should therefore be interpreted as lateral, posterior, and superior forces on the upper body, or medial, anterior, and inferior forces on the femur. Positive moment components (Mx, My, Mz) should be interpreted as muscle moments causing flexion, abduction, and internal rotation at the hip joint, respectively. The location of the hip joint center was calculated from anatomical landmark coordinates according to the recommendation by Bell et al. (3). Force and moment data were smoothed using cubic splines (49) at a cutoff frequency of 20 Hz for all activities, to obtain reliable measurements of peak loading.

The ski video data were reviewed to determine the phase of predominantly left limb support. It was found that this was perfectly correlated to the mediolateral component of the intersegmental force. During single support on the left leg, the mediolateral component peaked at a negative value. During the transfer of weight from the left to the right ski the mediolateral component quickly crossed zero and gradually increased to a positive peak. Zero was assumed to be symmetrical support on both legs. This was applicable to both alpine and cross-country skiing. During walking the phase of single support could be identified by the triple peaks in the superior-inferior force component. The second or middle and largest peak represented the summation of the force from the two legs during double support. The first peak represented the end of single leg support and the third peak represented the beginning of single leg support on the opposite limb (see Fig. 2 in results).

Figure 2
Figure 2:
The 3-D components of the intersegmental hip joint forces for one subject for each of the activities, before spline smoothing. Forces are expressed with respect to an upper-body coordinate system: x, positive lateral; y, positive posterior; z, positive superior. The horizontal solid bars indicate the phases of single left limb support for which the analysis is valid.

A Matlab program (Math Works Inc., Natick, MA) was used to digitize the peak force and moment magnitude of each cycle, calculated according to equations 1 and 2, respectively. Equations [1]-[2]

For each cycle, the peak magnitude of the intersegmental force (Fmag) was digitized. This value also typically corresponded to the peak moment magnitude (Mmag). Values for the corresponding force (Fx, Fy, Fz) and moment (Mx, My, Mz) components were obtained at this peak. A total of 10 cycles were digitized and averaged for each activity except for the long radius, steep terrain skiing. For this condition, only six cycles were digitized because steep terrain was limited at the local ski area. For running, high errors were associated with the impact phase (5); thus, peak loading was obtained for the active peak only. Impact and active forces have been described as the high- and low-frequency phases, respectively, of the force-time function, and it has been shown that the internal joint loading is higher for the active than for the impact phase in running (31).

For each subject, average values for the peak force and moment magnitude were obtained for each activity. BMDP statistical software was used to perform a one-way multivariate ANOVA with repeated measures (within factor) to test for significant differences between the loading conditions of the various activities. The two variables, Fmag and Mmag, were used in the statistical analysis. A level of significance (P level) of 0.01 was chosen. Post hoc contrasts were performed to make comparisons between each activity and walking. Contrasts were also performed between steep and flat slope terrain, short and long radius ski turns, and classic and skating cross-country techniques, respectively.

Joint contact forces were calculated from the mean magnitude of the intersegmental hip joint forces and moments. Most activities were dominated by an extensor moment, with the exception of walking which also included a large abductor moment. Nemeth and Ohlsen (30) determined in vivo moment arms for three hip extensor muscles over a range of hip flexion angles. Table 2 provides the hip flexion angles for each of the activities, approximated from the video data, and the corresponding average hip extensor moment from Nemeth and Ohlsen (30).

TABLE 2
TABLE 2:
Hip flexion angles and moment arms for the various activities.

For walking, the abductor moment arm was also considered. From the data of Dostal and Andrews (14), the gluteus medius muscle had an abductor moment arm of 60 mm at 20° of hip flexion. This moment arm is equal to the extensor moment arm selected for walking. Using these moment arms (d) average joint contact forces (Fjoint) were calculated for each activity using equation (3). It was assumed that the muscle forces and joint contact force act on the pelvis in opposite directions. This method provides an estimate of the magnitude of the contact forces but not the direction. Furthermore, no antagonistic muscle activity was assumed. Equation [3]

RESULTS

The root mean square (RMS) value of the residuals in the least-squares method, averaged over all samples in a recording, which indicates to which extent the accelerometer signals were consistent with the rigid body assumption, was lowest for walking (0.15-0.25 m·s−2), highest for running (0.8-0.9 m·s−2), and intermediate values were obtained for alpine and cross-country skiing (0.3-0.5 m·s−2). The impact phase of running contained samples with very high residuals (up to 8 m·s−2).

An example of the components of the intersegmental hip joint forces and moments (no smoothing) for one subject are shown in Figures 2 and 3, respectively. The age of the subject was 66 yr old, representative of the age typical of hip prosthetic patients. Graphs are displayed for walking, running, and alpine and cross-country skiing activities for a duration of 5 s. For alpine skiing, results for the lowest (long radius, flat slope) and highest (short radius, steep slope) loading activities are shown. Although the magnitudes of the intersegmental force and moment were used in the analysis to determine peak loading, the (x, y, z) components clearly indicate the left and right limb loading cycles. The horizontal bars on the graph indicate the phase of left limb support. The impact peaks during running are high for the subject shown in Figure 2, especially at the end of the trial where the subject is decelerating. As described previously (5) these peaks were associated with high potential errors and were not used for statistical analysis. The peak of the active (push-off) phase was used instead.

Figure 3
Figure 3:
The 3-D components of the intersegmental hip joint moments for one subject for each of the activities, before spline smoothing. Forces are expressed with respect to an upper-body coordinate system: x, flexion-extension; y, ab-adduction; z, internal-external rotation. The horizontal solid bars indicate the phases of single left limb support for which the analysis is valid.

Results from the peak loading analysis for the same subject are given in Table 3. Each value is an average of 10 cycles, except for the long radius, steep skiing activity where only six cycles were used. The results for this subject were very similar to the other older subject (69 yr old) but differed slightly from the younger subjects. The two older subjects had lower magnitudes of loading (Fmag and Mmag) than the younger subjects for all of the skiing activities. This was particularly apparent for the short radius, steep slope alpine condition and the cross-country skating condition.

TABLE 3
TABLE 3:
Peak intersegmental loads (average of 6 or 10 loading cycles) in one typical subject.

The average force and moment magnitudes for all subjects are shown in Figure 4. Standard deviations, indicating variability within the group, are shown for each activity. Walking and running had small variability between subjects, but there was high variability between subjects for all of the skiing activities. All activities had significantly higher intersegmental hip joint forces than walking except for the classic cross-country ski condition. All activities had significantly higher hip joint moments than walking. The highest forces were found for running and short radius ski turns on a steep slope. The highest moments were found for the short radius, steep slope skiing condition.

Figure 4
Figure 4:
The average peak force and moment magnitude results for all subjects. Standard deviation bars indicate the variability between subjects. Results for activities significantly different than walking are indicated by an asterisk.

In the statistical analysis, comparisons were also made between the different ski conditions. Long radius turns were found to have significantly lower hip joint forces and moments than short radius turns. Additionally, hip joint forces and moments for flat slope terrain were significantly lower than for steep slope terrain. No significant differences in hip joint loading were found between the two cross-country skiing techniques.

Values for the intersegmental forces and moments shown in the bar graphs are given in Table 4, which also includes the components along the three anatomical axes. Large differences in the mediolateral (x) component are apparent between activities. Alpine skiing and cross-country skating technique had higher mediolateral forces than the other activities. When interpreting the anterior-posterior (y) and superior-inferior (z) forces, it should be kept in mind that force components are reported in an upper body coordinate system of which the orientation is continuously changing during the movement. For all activities, the moments were predominantly hip extensor moments (negative Mx). Walking and running also had relatively large abductor moment (positive My) contributions.

TABLE 4
TABLE 4:
Average peak intersegmental loads for all 9 subjects.

For the data collection session in the mountains with two subjects, the intersegmental forces and moments were higher for large mogul skiing than for any of the skiing conditions at the local ski hill. For the small mogul condition at the mountain ski area, the forces and moments were higher in one subject and lower in the other subject than the short radius, steep slope condition at the local ski hill. Forces and moments obtained for different snow conditions (soft natural snow vs manmade hard packed snow) did not appear to be different.

Results from the calculation of hip joint contact forces are given in Figure 5 for each of the activities. The joint contact forces were calculated from the mean peak force and moment results of all subjects and expressed in terms of body weight (BW). The mean body weight of the subjects was 72.4 kg. Joint contact forces are also given for two of the ski conditions from the data collection session in the mountains: large and small mogul skiing. Walking had the lowest joint contact forces (2.5 × BW). Alpine skiing had about the same joint contact force as running (5.2 × BW) except for long radius turns on a flat slope, which had significantly lower hip joint contact forces (4.1 × BW) and short radius turns on a steep slope (7.8 × BW) and mogul skiing (8.3-12.4 × BW), which had significantly higher hip joint contact forces. The two cross-country techniques and one skiing condition (long radius, flat slope) all had intermediate force values (4.0-4.6 × BW).

Figure 5
Figure 5:
Average peak joint contact forces for each of the activities, normalized to body weight. Standard deviation is shown for variability between subjects or between trials for each of the two subjects who performed mogul skiing.

DISCUSSION

In this study we used a new method, based on accelerometry, to determine intersegmental force and moment at the hip joint. Compared with conventional methods using ground reaction forces and kinematic data, this method is convenient because it allows data collection outside the laboratory, recordings of long duration, and fully automated analysis. This new method has been described in detail elsewhere (5). However, some important aspects are summarized here. Several major assumptions were made in order to obtain the intersegmental forces and moments:

  1. Head, arms, and trunk are combined into one rigid "upper body" segment.
  2. The accelerometers are all rigidly connected to this "upper body" segment.
  3. No force and moment is acting at the contralateral hip joint.
  4. No other external forces are acting at the upper body.

The implications of these assumptions for analysis of walking and running have been investigated previously (5). Assumption 4 is obviously correct for these movements. Assumption 3 requires that the double support phase of walking should not be analyzed, but this did not limit the present study since peak hip joint loading occurs during the single support phase. Secondly, during the single support phase, the force and moment generated by the swing leg were incorrectly attributed to the stance leg which introduced an error due to the accelerations of the swing leg segments and the large mass of the leg which was not included in the mass of the upper body. Assumption 2 was evaluated by examining to which extent the 12 accelerometer signals were consistent with a rigid-body movement. Only for the impact phase of running were significant errors found, which could result in errors in hip joint loading of up to 50% of upper body weight. The only error introduced by assumption 1 was neglecting a moment about the vertical axis, generated by the swinging of the arms. A comparison between the accelerometry method and a conventional inverse dynamics analysis using force plate, video, and a model of the lower extremities indicated that all assumptions together resulted in a systematic underestimation of the intersegmental force and moment during walking and running by about 20% (5).

For alpine skiing, no problems are expected with assumption 1, due to the lesser amount of arm swing. Cross-country skiing should, in this regard, be comparable to running. The rigidity of the accelerometer configuration (assumption 2) during all skiing activities was satisfactory, introducing errors less than 0.5 m·s−2, which corresponds to about 25-N intersegmental force. The validity of assumption 3 for alpine skiing depends on the distribution of a skier's weight between the two skis, downhill and uphill ski. Müller (27) reported that during a parallel turn, the weight on the downhill ski varied between 75% and 85% of the total weight, depending on the conditions and slope of the hill. In the present study, the subjects were asked to concentrate on skiing with all of their weight on the downhill ski. It is reasonable to assume that our subjects had 85% or more of their weight on the downhill ski. Because the mass of the entire leg is about 16.1% of total body weight (47), this could mean that the uphill leg supports its own weight and little force or moment is transmitted through the uphill hip joint. We therefore expect that assumption 3 is more valid for alpine skiing than for running. A similar situation exists during cross-country skiing. Smith (41) showed that at the time of peak loading, the contralateral leg carries about 10-15% of the total force. The one exception to this is the mogul skiing condition where the weight distribution has been found to be almost equal between the two skis (27). Assumption 4 is possibly incorrect because of the ground reaction force acting on the ski pole. In alpine skiing, the pole is used mostly for timing and not a propulsive force; therefore, the forces would be minimal. However, in cross-country skiing, the poling action is used as a propulsive force. In the present study, no poles were used for classic technique. For skating technique, poles were necessary to perform the proper technique, but the subjects were asked to pole on the right side; this meant that during the left limb cycle, the poling action occurred only at the very beginning of the cycle. The peak force and moment magnitudes were digitized at the end of the cycle. The timing of the poling force could also be seen in the force-time results as a large anterior-posterior (y) force. The only other external force acting on the trunk segment was air resistance, which is estimated to be between 10 and 35 N, depending on body position, for a skier at a speed of 10 m·s−1(38).

The final step in the analysis, the calculation of joint contact forces from the intersegmental loads, required the assumption that no antagonistic muscle forces were present. To satisfy a joint moment, antagonistic activity requires a corresponding increase in agonistic activity, leading to a substantial increase in joint contact force (19). Specifically in the present study, a hip joint extensor moment was predominant in all of the activities analyzed. This moment would be satisfied by the hip joint extensor muscles (i.e., biceps femoris, semimembranosus, semitendinosus, adductor magnus, and gluteus maximus). Antagonistic muscle forces would be generated by the hip joint flexors (i.e., adductor longus, rectus femoris, gracilis, sartorius, and iliacus) (13,35). After examining EMG studies in the literature on the various activities, it was concluded that antagonistic activity can be classified as low for walking (13,35), medium for running (26) and cross-country skiing, (23) and high for alpine skiing (9,25). Due to the assumption of no antagonistic activity, the joint contact forces calculated in the present study are underestimated, especially for alpine skiing. It should be noted however, that the contribution to the joint contact force vector which is due to muscle activity tends to be aligned with the long axis of the femur, while the contribution due to upper body weight and accelerations (the first term in equation 3) may have any direction. The implications of loading direction will be discussed below.

Age of the subject has been found to be an important factor to consider when analyzing gait kinetics and kinematics. Crowninshield et al. (11) reported significantly lower hip joint moments and hip contact forces in a group of older subjects (60-80 yr) compared with a group of younger subjects (22-30 yr) for the same velocity of gait. In the present study, there were not enough older subjects to make a statistical comparison, but very little difference was found in the results between the older subjects and younger subjects during walking and running. However, for the skiing activities, in particular the highest loading condition (short radius, steep slope) and cross-country skating condition, the magnitude of loading was considerably lower for the two older subjects. This indicates that the skiers can control the magnitude of loading by performing the activities less dynamically. However, even with a more controlled style of skiing the hip joint moments for these two activities were similar to or higher than for running.

Table 5 shows a comparison of the joint contact force results from the present study for walking and running with values previously reported in the literature. Note that literature data for running were only available for much lower running speeds (1.9-2.5 m·s−1) than the present study (3.5 m·s−1). The average joint contact forces calculated in the present study for walking (2.5 BW) and running (5.2 BW) were at the low range of values reported in the literature. The results for walking show that the inverse dynamics studies using a model of the lower extremity typically give high values compared with direct measurements using an instrumented prosthesis. Brand et al. (6) explained this as a sensitivity to the modelling assumptions; in particular, the muscles were assumed to act along a straight line between origin and insertion, which tended to result in severely underestimated moment arms and therefore overestimated muscle forces. In the present study, we used an average extensor moment arm of approximately 60 mm, based on recent literature (14,30). If a smaller, previously reported moment arm value (35 mm) for the extensor muscles is used (30), then the calculation of joint contact force in our study would be 3.7 and 7.8 times body weight for walking and running, respectively, which is more within the range of literature values. Literature data collected with instrumented prostheses may not represent the normal population because of the age of the subjects. Furthermore, many of the values reported in the literature are peak values for one subject. The values reported in the present study are averaged over 10 cycles and 9 subjects. These methodological considerations make it difficult to compare data from the literature; it is more meaningful to compare the results between the activities in the present study where the same subjects, equipment, assumptions, and analysis were used for all activities (Fig. 5).

TABLE 5
TABLE 5:
Peak hip joint contact forces reported in the literature for walking and running.

The results of the study will now be interpreted with respect to the potential effects of these activities on the fixation of the components of a hip prosthesis. The following aspects are relevant:

  1. Impact loading.
  2. Magnitude of loading.
  3. Direction of loading.

Impact loading of the joint, defined as a rapid increase in force (31) (<50 ms to peak), was only seen in recordings of running (Figs. 2 and 3). All other activities showed a gradual increase of force. In our study, the impact loading during running could not be quantified reliably using the accelerometry technique. Relative movements between accelerometers and trunk have probably led to overestimation of impact forces in the hip joint, if anything, because the analysis interpreted any movement of the accelerometers as movement of the entire upper body. This is supported by the absence of visible impact loading on an instrumented prosthesis (4) during jogging at 2.2 mm·s−1. We therefore conclude that impact loading of the hip was not present in any of the skiing activities and that impact loading is not a reason to advise against skiing after hip replacement.

Considering the magnitude of loading (Fig. 5), it can be concluded that cross-country skiing and controlled alpine skiing (with the exception of short turns on a steep slope) produce joint contact forces that are between walking and running and therefore not excessive. More dynamic skiing activities, such as short turns on a steep slope and mogul skiing, produce loads that are 50-100% larger than running and should not be recommended.

There are considerable differences in the direction of hip joint loading as indicated by the force and moment components (Table 4) between the different activities. Direction of the moment vector indicates which muscle groups are active. If it is assumed that muscles are aligned with the long axis of the femur, only the magnitude of moment is relevant for the joint contact force vector. Because the muscle forces (the second term in equation 3) are the main contribution to the joint contact force, the joint contact force will be approximately aligned with the long axis of the femur. The nonaxial components of the joint contact force are small but may be important for the implant. Information about these components can be obtained from the intersegmental force vector (Fx, Fy, Fz). As shown in Table 4, alpine skiing had much higher anterior-posterior and mediolateral forces than both walking and running. Cross-country skiing also had higher values for these components but not as high as alpine skiing. Studies in the literature on mechanical loading of hip joint prostheses have found the direction of loads acting on the head of the femur to be important (8). In particular, there has been a recent focus in the literature on torsional loading of the head of the femur (24,32,36,43). Torsional loading is due to a posteriorly directed force acting on the head of the femur, which tends to twist the prosthetic stem into retroversion (36). This force has been found to be relatively small during walking but two to three times higher during activities such as stair climbing and rising from a chair (10,12). These same activities have been found to result in moderate to severe thigh pain particularly in patients with cementless prosthetic stems, which is believed to be a symptom of femoral stem loosening (37), the most frequent complication in total hip replacements. Mechanical testing of the prostheses under torsional loading has resulted in loosening of the femoral stem at loads within the physiological range of normal activity (36). Cementless prostheses, more commonly used for young, active patients have been found to fail at much lower torsional loads than cemented prostheses (32).

Interpretation of our results with respect to femoral stem loosening requires a coordinate transformation from the upper body coordinate system to a femur coordinate system. Table 6 provides the anterior-posterior (AP) component of the intersegmental force vector acting on the trunk segment in the sagittal (y-z) plane, the approximate hip flexion angle for each activity, and the AP component Fy of the force acting on the femoral head, expressed in the femur coordinate system. Using action = −reaction, this component is calculated as: Equation [4] where α is the flexion angle. All activities had a small force component acting on the femur posteriorly. This is the direction of loading associated with femoral stem loosening. Using direct force measurements, Davy et al. (12) measured a force component of 400 N acting posteriorly on the head of the femur for stair climbing. For all of the activities in the present study, this component of the force vector was lower than that. Skiing on the flat slope was comparable to walking, and cross-country skiing using classical style had only half the Fy of walking. Those two activities therefore do not lead to increased risk of femoral stem loosening due to torsional loading. However, very large Fy values are present in the trunk (pelvis) coordinate system for all skiing activities, indicating that skiing may create more problems in the acetabular component of a hip prosthesis than any other activity. This effect may be aggravated by the presence of large muscle forces and large flexion angles during those activities.

TABLE 6
TABLE 6:
The anterior-posterior component of the intersegmental force vector in the sagittal (y-z) plane for both the trunk and femur segment coordinate systems and the hip flexion angle for each of the activities.

Few mechanical studies have included an analysis of prosthetic loading in the mediolateral direction because this force component is minimal during gait. However, in alpine skiing the mediolateral forces were large, over 0.5 times body weight for the short radius, steep slope skiing condition (Table 4). Hampton et al. (18) used a finite element model of the prosthetic stem to analyze the stress distribution of three orthogonal loads applied to the ball of the stem. They reported that the tensile stress on the medial surface for a laterally applied load was 4.5 times larger than the maximum tensile stress on the lateral surface due to longitudinal loading. They speculated that this large tensile stress could initiate cracks on the medial surface. This study indicates high mediolateral forces during alpine skiing and cross-country skating that may have a negative effect on the prosthesis.

Table 7 shows a summary of the variables relevant for hip prosthesis survival. The value for each variable is compared to walking because walking is assumed to be a "safe" activity for a hip prosthetic patient. Note that the mediolateral force component during walking was extremely small, which tends to exaggerate the magnitude of that variable for the other activities. For alpine skiing, only the lowest (long turns, flat slope) and highest (short turns, steep slope) loading conditions are included. If running and walking are assumed to be "unsafe" and "safe" activities for a hip prosthetic patient, respectively, then based on the results shown in Table 7, alpine skiing on steep slopes, moguls or performing short radius turns are not recommended for a hip prosthetic patient. Controlled alpine skiing (long radius turns on a flat slope) and cross-country skiing appear to be relatively safe activities. However, the analyzed skiing activities showed higher loading in directions where walking showed low loading. In particular, the skiing activities had higher mediolateral and anterior-posterior forces than walking. Little is known about the effects of high loading in these transverse directions on a hip joint prosthesis. Further mechanical testing of prosthetic devices with loading conditions specific to these activities is needed to assess the potential effect of increases in these force components on artificial hip prostheses and to allow an interpretation with respect to positive or negative effects of controlled alpine and cross-country skiing for a hip prosthetic patient. Specifically, the acetabular component of a THR should be tested using forces obtained in this study during alpine skiing, because it experiences a large increase in A-P loading when alpine skiing is compared to walking. Another factor that has not been mentioned so far and was not analyzed in the present study is the risk of falling during skiing. It is assumed a hip joint replacement patient who returns to skiing is an intermediate to expert skier who has skied most of his or her life and skis under very controlled conditions. The risk of this type of skier falling would be small but definitely a risk of the sport.

TABLE 7
TABLE 7:
Summary of the variables relevant for hip prosthesis survival.

REFERENCES

1. An, K.-N., E. Y. S. Chao, and K. R. Kaufman. Analysis of muscle and joint loads. In: Basic Orthopaedic Biomechanics, V. C. Mow and W. C. Hayes (Eds.). New York: Raven Press, 1991, pp. 1-50.
2. Barrack, R. L., R. D. Mulroy, and W. H. Harris. Improved cementing techniques and femoral component loosening in young patients with hip arthroplasty: a 12-year radiographic review. J. Bone Joint Surg. 74B:385-389, 1992.
3. Bell, A. L., D. R. Pedersen, and R. A. Brand. A comparison of the accuracy of several hip center location prediction methods. J. Biomech. 23:617-621, 1990.
4. Bergmann, G., F. Graichen, and A. Rohlmann. Hip joint loading during walking and running, measured in two patients. J. Biomech. 26:969-990, 1993.
5. Bogert, A. J. van den, L. Read, and B. M. Nigg. A method for inverse dynamic analysis using accelerometry. J. Biomech. 29: 949-954, 1996.
6. Brand, R. A., D. R. Pedersen, D. T. Davy, K. G. Heiple, and V. M. Goldberg. Comparison of hip joint force calculations and measurements in the same patient. Trans. 35th Ann. Meeting Orthop. Res. Soc., Las Vegas, 1989, p. 96.
7. Chandler, H. P., F. T. Reineck, R. L. Wixson, and J. C. McCarthy. Total hip replacement in patients younger than thirty years old. J. Bone Joint Surg. 63A:1426-1434, 1981.
8. Cheal, E. J., M. Spector, and W. C. Hayes. Role of loads and prosthesis material properties on the mechanics of the proximal femur after total hip arthroplasty. J. Orthop. Res. 10:405-422, 1992.
9. Clarys, J. P., J. Publie, and E. Zinzen. Ergonomic analyses of downhill skiing. J. Sports Sci. 12:243-250, 1994.
10. Crowninshield, R. D., R. C. Johnston, J. G. Andrews, and R. A. Brand. A biomechanical investigation of the human hip. J. Biomech. 11:75-85, 1978.
11. Crowninshield, R. D., Brand, R. A., and R. C. Johnston. The effects of walking velocity and age on hip kinematics and kinetics. Clin. Orthop. Rel. Res. 132:140-144, 1978.
12. Davy, D. T., G. M. Kotzar, R. H. Brown, et al. Telemetric force measurements across the hip after total arthroplasty. J. Bone Joint Surg. 70A:45-50, 1988.
13. Devita, P. The selection of a standard convention for analyzing gait data based on the analysis of relevant biomechanical factors. J. Biomech. 27:501-508, 1994.
14. Dostal, W. F., and J. G. Andrews. A three-dimensional biomechanical model of hip musculature. J. Biomech. 14:803-812, 1981.
15. Dubs, L., N. Gschwend, and U. Munzinger. Sport after total hip arthroplasty. Arch. Orthop. Trauma Surg. 101:161-169, 1983.
16. Evans, B. G., E. A. Salvati, M. H. Huo, and O. L. Huk. The rationale for cemented total hip arthroplasty. Orthop. Clin. North Am. 24:599-610, 1993.
17. Galante, J. O. Causes of fractures of the femoral component in total hip replacement. J. Bone Joint Surg. 62A:670-673, 1980.
18. Hampton, S. J., T. P. Andriacchi, and J. O. Galante. Three dimensional stress analysis of the femoral stem of a total hip prosthesis. J. Biomech. 13:443-448, 1980.
19. Herzog, W., and L. Read. Anterior cruciate ligament forces in alpine skiing. J. Appl. Biomech. 9:260-278, 1993.
20. Huiskes, R. Biomechanics of artificial-joint fixation. In: Basic Orthopaedic Biomechanics, V. C. Mow and W. C. Hayes (Eds.). New York, Raven Press, 1991, pp. 375-442.
21. Hungerford, D. S., and L. C. Jones. The rationale for cementless total hip replacement. Orthop. Clin. North Am. 24:617-626, 1993.
22. Kilgus, D. J., F. J. Dorey, G. A. M. Finerman, and H. C. Amstutz. Patient activity, sports participation, and impact loading on the durability of cemented total hip replacements. Clin. Orthop. Rel. Res. 269:25-31, 1991.
23. Komi, P. V., and R. W. Norman. Preloading of the thrust phase in cross-country skiing. Int. J. Sports Med. 8(Suppl.)48-54, 1987.
24. Kotzar, G. M., D. T. Davy, J. Berilla, and V. M. Goldberg. Torsional loads in the early postoperative period following total hip replacement. J. Orthop. Res. 12:945-955, 1995.
25. Maxwell, S. M., and M. L. Hull. Measurement of strength and loading variables on the knee during alpine skiing. J. Biomech. 22:609-624, 1989.
26. Montgomery, W. J., M. Pink, and J. Perry. Electromyographic analysis of hip and knee musculature during running. Am. J. Sports Med. 22:272-278, 1994.
27. Müller, E. Analysis of the biomechanical characteristics of different swinging techniques in alpine skiing. J. Sport Sci. 12:261-278, 1994.
28. Mulroy, R. D. J., and W. H. Harris. The effect of improved cementing techniques on component loosening in total hip replacement: an 11-year radiographic review. J. Bone Joint Surg. 72B:757-760, 1990.
29. National Institutes of Health. Total hip replacement. NIH Consensus Statement. 12(5):23-34, 1994.
30. Nemeth, G., and H. Ohlesen. In vivo moment arm lengths for hip extensor muscles at different angles of hip flexion. J. Biomech. 18:129-140, 1985.
31. Nigg, B. M. Biomechanical aspects of running. In: Biomechanics of Running Shoes, B. M. Nigg (Ed.). Champaign, IL: Human Kinetics Publishers, 1986, pp. 1-25.
32. Nunn, D., M. A. R. Freeman, K. E. Tanner, and W. Bonfield. Torsional stability of the femoral component of hip arthroplasty: response ot an anteriorly applied load. J. Bone Joint Surg. 71B:452-455, 1989.
33. Paul, J. P. Load actions on the human femur in walking and some resultant stresses. Exp. Mech. 11:121-125, 1971.
34. Paul, J. P. Approaches to design: force actions transmitted by joints in the human body. Proc. R. Soc. Lond. B 192:163-172, 1976.
35. Perry, J. Gait Analysis: Normal and Pathological Function. Thorofare, NJ: SLACK, Inc., 1992, pp. 111-129.
36. Phillips, T. W., T. Nguyen, and S. D. Munro. Loosening of cementless femoral stems: a biomechanical analysis of immediate fixation with loading vertical, femur horizontal. J. Biomech. 24:37-48, 1991.
37. Phillips, T. W., and S. S. Messieh. Cementless hip replacement for arthritis. J. Bone Joint Surg. 70B:750-755, 1988.
38. Read, L., and W. Herzog. External loading at the knee joint for landing movements in alpine skiing. Int. J. Sports Biomech. 8:62-80, 1992.
39. Röhrle, H., R. Scholten, C. Sigolotto, W. Sollbach, and H. Kellner. Joint forces n the human pelvis-leg skeleton during walking. J. Biomech. 17:409-424, 1984.
40. Rydell, N. W. Forces acting on the femoral head prosthesis. Acta Orthop. Scand. 37(Suppl. 88):1-132, 1966.
41. Smith, G. A. Biomechanical analysis of cross-country skiing techniques. Med. Sci. Sports Exerc. 24:1015-1022, 1992.
42. Stauffer, R. N. Ten-year follow-up study of total hip replacement. J. Bone Joint Surg. 64A:983-990, 1982.
43. Sugiyama, J., L. A. Whiteside, and A. D. Kaiser. Examination of rotational fixation of the femoral component in total hip arthroplasty. Clin. Orthop. Rel. Res. 249:122-129, 1989.
44. Sutherland, C. J., A. H. Wilde, and L. S. Borden. A ten year follow-up of one hundred consecutive Muller curved-stem total hip replacement arthroplasties. J. Bone Joint Surg. 64A:970-982, 1982.
45. Visuri, T., and R. Hokkanen. Total hip replacement: its influence on spontaneous recreation exercise habits. Arch. Phys. Med. Rehabil. 61:325-328, 1980.
46. Weinans, H. Mechanically induced bone adaptations around orthopaedic implants. Doctoral thesis, University of Nijmegen, The Netherlands, 1991.
47. Winter, D. A. Biomechanics of Human Movement. New York: Wiley, 1979.
48. White, J. No more bump and grind: exercise and total hip replacement. Physician Sportsmed. 20:223-228, 1992.
49. Woltring, H. J. A FORTRAN package for generalized, cross-validitory spline smoothing and differentiation. Adv. Eng. Software 8:104-113, 1986.
Keywords:

HIP JOINT; SKIING; JOINT LOADING; ACCELEROMETRY

© 1999 Lippincott Williams & Wilkins, Inc.