Knee adduction moment and development of chronic knee pain in elders

Authors


Abstract

Objective

To determine whether the adduction moment at the knee during locomotor activity contributes to the development of future chronic knee pain.

Methods

We studied 132 community-dwelling elders who had undergone a full kinetic and kinematic motion analysis while performing 4 different activities: standing, walking, rising from a chair, and descending stairs. We contacted the participants 3–4 years after their baseline locomotion analysis and identified those who reported no knee pain at the time of motion analysis but who subsequently developed new chronic knee pain at followup. We examined whether the development of new chronic knee pain was associated with higher peak adduction moment at the knee during activities, measured at baseline.

Results

Of the 132 elders evaluated in 1995–1996, 118 (89%) were contacted in 1999. Of the 118 contacted, 80 (mean age 75 years; 78% women) had no lower extremity prosthetic joints at baseline, no known underlying inflammatory arthritis at baseline nor followup, and no baseline knee pain. At followup, 7 had developed new chronic knee pain defined as pain or stiffness on most days of the month and with walking 2 blocks or using stairs. Compared with those who did not develop knee pain, those who did develop new chronic knee pain had higher baseline adduction moments for all activities (P = 0.01), ranging from 8% higher during chair rise to 39% higher during stair descent.

Conclusion

We found that greater adduction moment at the knee during activities contributes to the development of future chronic knee pain. Our results suggest that biomechanical factors may play an important role in the pathogenesis of knee pain and should be studied further.

INTRODUCTION

Chronic knee pain contributes significantly to disability among elders. It has been estimated that 25% or more of older persons have chronic knee pain, when defined as pain occurring on most days of a recent month (1). Little is known, however, of risk factors for knee pain. Chronic knee pain among elders is most frequently associated with osteoarthritis (OA) and tends to occur with or be aggravated by activity. Yet, there has been relatively little attention on biomechanical parameters of daily activities that may affect the risk of developing chronic knee pain or knee OA.

The biomechanical variable best studied for its association with knee pain and OA is the adduction moment across the knee. During walking, the most common daily activity, forces acting on the leg produce an adduction moment that tends to adduct the knee into the varus position (2). This moment is correlated with the compressive force across the medial compartment of the knee, the compartment most often affected in knee OA (3). During walking, 60–70% of weight-bearing force passes through this compartment (2). The adduction moment has been reported by some to be the largest moment arm during gait (2).

Increasing evidence shows that the adduction moment has clinically important consequences to the knee joint. Ogata et al placed springs attached to wires across the knee joints of rabbits; the knees in which the spring tension caused elevated adduction moments experienced accelerated cartilage breakdown, suggestive of early osteoarthritis in the medial compartment compared with contralateral unoperated and unaffected knees (4). In patients undergoing high tibial osteotomy, all of whom had valgus alignment after surgery, a high preoperative adduction moment was correlated with a poor operative clinical outcome and an eventual return to varus alignment (5). Those with knee OA have greater adduction moment at the knee than control subjects (6). Others have shown that greater adduction moment at the knee is associated with greater severity of knee OA (7). Higher adduction moment at the knee has also been shown to be predictive of radiographic progression of knee OA (8). However, there are no data as to whether the adduction moment affects the risk of developing chronic knee pain.

We took advantage of comprehensive kinetic and kinematic motion analysis laboratory data on >100 elders drawn from the community, most of whom had no complaints of knee pain. We investigated whether the knee adduction moment measured during different activities was associated with later development of chronic knee pain.

SUBJECTS AND METHODS

Study participants.

All subjects provided informed consent in accordance with the institutional policy on human research. Participants of this study were subjects in the Strong For Life program, a clinical trial of ambulatory elders in a 6-month home-based resistance exercise program where kinetic and kinematic motion analysis data were collected at baseline in 1995–1996 (9). To be eligible for the Strong for Life program, subjects had to be functionally limited, noting at least 1 problem on the Short Form 36 physical function scale (excluding the vigorous activity item). Eligibility criteria also included the following: being 60 years or older, cognitively intact, having permission from his/her primary care physician to participate in this study, being able to ambulate independently at least 25 feet, and passing an in-home cardiovascular exercise safety test. Subjects were excluded if they had a terminal illness, uncontrolled hypertension, acute renal failure, a history of neurologic disease, or any report of acute pain impairing test performance.

Kinetic and kinematic motion analysis.

All study participants underwent a full kinetic and kinematic laboratory motion analysis at baseline while performing 4 different activities: standing, rising from a chair (chair rise), walking, and descending stairs (stair descent) (10). Details of the motion analysis assessment and its reproducibility have been described and reported previously (11–14). Three-dimensional (3D), full-body motion analysis data were collected using a bilateral Selspot II optoelectric camera system (Selective Electronics, Partille, Sweden) and 2 piezoelectric force platforms (Kistler Instruments, Winterthur, Switzerland) at 152 Hz. The optoelectronic cameras captured illuminated signals from arrays of 3–5 infrared light emitting diodes (LEDs) embedded in rigid plastic disks secured to 11 body segments (head, trunk, pelvis, and both arms, thighs, shanks, and feet) (11). TRACK™ software (Massachusetts Institute of Technology, Cambridge, MA) was used to compute the spatial position and orientation of body segments from the LED array information, as described previously (12, 15). Segmental velocities and accelerations were computed using a Lagrangian 5-point numerical differentiation (16), and inertial parameters were determined from regression equations using the subject's anthropomorphic measurements (17, 18). Kinematic, force plate, and inertial parameters were then used to compute knee joint forces and moments of force using a Newton-Euler inverse dynamic approach (19). In this study, the 3D kinetic components were referenced in proximal segment coordinates. Using this convention, moments in the frontal plane produce rotary movements about the x-axis, adducting or abducting the joint. Therefore, the adduction moment at the knee moves the tibia inward towards the midline of the body relative to the femur.

Study participants performed all 4 locomotor activities barefoot, using a standardized protocol (10). Standing: Subjects stood with feet 30 cm apart and eyes open. Feet were not permitted to move from the starting position until the 7-second task was completed. Chair rise: Subjects arose from a chair adjusted to their knee height. Feet were 10 cm apart. As in the standing trial, feet were not permitted to move during the task. Walking: Subjects walked at their self-selected pace along a 10-meter walkway and were required to cleanly strike 1 of the 2 force plates for at least 1 gait trial. Stair descent: Subjects descended a 4-step modular staircase with the second and third steps positioned over the force plates. Step height was 19 cm and no railing support was provided.

Study participants had at least 1 practice trial for each task prior to data collection and performed each of the 4 activities at least twice during data acquisition. Some subjects were not able to descend stairs or rise from a chair unassisted, so for these subjects data in those activities were not available for analyses.

Determination of new chronic knee pain.

Study participants were contacted 3–4 years following their motion analysis study. A single interviewer administered a standardized questionnaire by telephone. Neither participants nor interviewer were aware of results from the locomotor analysis.

The survey inquired whether participants had knee pain during the year of the motion analysis and at 3–4 years followup. We used standard prompts to remind participants of the year of the motion analysis study, including using milestone life events and events in other family members to reorient them to the year of study (20). This aided with the recall of knee symptoms during the period of time of the motion analysis evaluation (20). The survey included 2 specific questions on the frequency of knee pain and particular activities that precipitated knee pain and were posed for each time period: 1) Do you have pain in or around your knee joint on most days of the month? and 2) Do you have knee pain with walking 2 blocks or when using stairs? These 2 questions were selected because recent work has shown that an affirmative response to both questions was better correlated with the presence of symptomatic knee OA than any other specific knee-pain questions (21). We used a validated questionnaire (22) to evaluate and exclude those elders who might have had rheumatoid arthritis or other forms of inflammatory arthritis either at baseline or followup.

To determine who had developed new chronic knee pain since the baseline motion analysis, we excluded all those who reported any knee pain (positive response to either of the 2 knee pain questions) at the time of the baseline motion analysis. We also excluded those with a history of total joint replacement of either the knee or the hip at baseline. Among those who reported no baseline knee pain, we then examined how many had since developed new chronic knee pain (affirmative responses to both of the above questions at followup). We excluded from analyses those who responded affirmatively to 1 but not both of the knee pain questions for the followup period. Those who screened positive for inflammatory arthritis either at baseline or followup were also excluded.

Statistical analysis.

We averaged the adduction moment from repeated trials of each activity. Left-sided and right-sided peak adduction moments at the knee during each of the activities were then averaged to generate more stable kinetic data. In general, this was always possible for standing, chair rise, and stair descent in which subject's foot placement could be controlled (to ensure force plate measures for dynamic analysis). For walking, when information on force plate measures for dynamic loading was obtained from only 1 side of the body during a trial, it was often possible to acquire joint moments for both sides of the body during successive walking trials, as subjects performed each task at least twice. Adduction moment was normalized to the subject's body weight and height to achieve a unitless parameter.

To compare the adduction moment during different activities of those with new chronic knee pain with those without knee pain, we performed a multivariate analysis of variance. Statistical analyses were performed using SAS software, release 6.12 (SAS Institute, Cary, NC).

RESULTS

The mean ± SD age of the 132 participants at the time of the baseline motion analysis in 1995–1996 was 74.9 ± 6.6 years with a range of 61–90 years. Of 132 participants (100 women, 32 men), we could not interview 14 subjects in 1999, of whom 4 had died and 10 could not be contacted. Among the 118 we surveyed (89% of the original participants; 93 women and 25 men), 37 were excluded due to the presence of any knee pain at baseline, total hip or knee replacement at baseline, or the presence of an inflammatory arthritis at either baseline or followup. Only 1 subject could not recall her baseline symptoms, so was also excluded (Figure 1). There were therefore 80 subjects remaining for our study who were characterized as having no knee pain at baseline (mean age 75.2 years; 62 women and 18 men). Of these 80, 7 developed new chronic knee pain (5 women and 2 men), and 66 (51 women and 15 men) had no knee pain at followup (Figure 1). Those with new chronic knee pain reported symptoms as being present for 3 months or longer. The remaining 7 subjects reported knee pain inconsistently (responding affirmatively to 1 but not both questions) and were excluded from analyses.

Figure 1.

Eligible study participants.

Of those who were destined to develop new chronic knee pain, body mass index was slightly higher (28.0 kg/m2) than those without knee pain (25.6 kg/m2; Table 1). Subjects with new chronic knee pain were also slightly older (76.3 versus 75.2 years; Table 1). There was no correlation between mean peak adduction moment in any activity with either age (r = 0.03, –0.11, 0.08, and 0.13 for standing, chair rise, walking, and stair descent, respectively; P > 0.3) or body mass index (r = 0.13, 0.21, –0.03, and 0.10 for standing, chair rise, walking, and stair descent, respectively; P ≥ 0.1).

Table 1. Baseline characteristics of subjects
Baseline characteristicsNo knee pain (n = 66)New chronic knee pain (n = 7)
Age, mean ± SD years75.2 ± 6.576.3 ± 10.1
Body mass index, mean ± SD kg/m225.6 ± 3.628.0 ± 3.0
Women, %7771

When we compared the mean peak adduction moment at baseline of the 2 groups, we found that baseline mean peak adduction moments were higher for all activities in those destined to develop new chronic knee pain compared with those without pain (F = 6.38, P = 0.01; Table 2). Mean peak adduction moments ranged anywhere from 8% to 39% higher depending on the activity and showed the greatest difference between the 2 groups with standing and descending stairs (Table 2).

Table 2. Baseline knee adduction moment of subjects*
ActivityBaseline mean peak adduction moment (% BW-Ht)
No knee pain (n = 66)New chronic knee pain (n = 7)Difference, %
  • *

    Data are given as mean ± SD. % BW-Ht = Percent body weight-height.

  • P = 0.01 for difference in adduction moment for all activities between 2 groups tested by multivariate analysis of variance.

  • n = 64 for no knee pain group because not all subjects could arise from a chair unassisted.

  • §

    n = 54 for no knee pain group and n = 6 for new knee pain group because not all subjects could descend stairs unassisted.

Standing0.42 ± 0.290.59 ± 0.43+35
Chair rise1.15 ± 0.541.31 ± 0.70+8
Walking3.05 ± 1.253.55 ± 1.56+13
Stair descent§4.42 ± 1.696.12 ± 1.88+39

DISCUSSION

In this followup study of elders who underwent a baseline motion analysis, our data suggest that among those without knee pain, higher peak adduction moment across the knee with activities, especially standing and descending stairs, is associated with the development of future chronic knee pain. These data are the first predictive evidence, of which we are aware, that greater peak adduction moment in asymptomatic individuals may cause future chronic knee pain. They also provide the first evidence that a biomechanical parameter may affect development of future knee symptoms.

It remains unknown how a higher knee adduction moment with activity may mediate the development of future chronic knee pain, and whether it affects soft tissue structures surrounding the knee joint or subchondral bone. One possible hypothesis is that higher knee adduction moments may mediate knee pain by excessively loading subchondral bone of the medial compartment. There is growing evidence that subchondral bone and its turnover may play a causal role in the pathogenesis of osteoarthritis as well as its related symptoms, especially at the knee (23–26). We speculate that high adduction moments, particularly with certain activities such as descending stairs, could lead to increased loading on the medial compartment of the knee, producing trauma and contusion to the subchondral bone, with subsequent pain. Thus, high adduction moments at the knee could be a potential mechanism for development of chronic knee pain. It may also be that individuals with higher adduction moments at the knee have other biomechanical stressors that could affect symptoms at the knee, such as varus alignment, varus-valgus instability, or other structural change at the medial tibiofemoral joint, such as early asymptomatic medial knee OA.

It is important to note that because we did not have knee radiographs at baseline nor followup, we therefore studied chronic knee pain and not necessarily knee OA. To be characterized as knee pain positive, subjects had to have both frequent pain and pain with activities (21). We also used a validated questionnaire to exclude those subjects whose knee pains may have been attributed to inflammatory joint disease (22), as those diseases may have other mechanisms for pain. Furthermore, at least 5 of the 7 subjects who developed new chronic knee pain in our study reported that a physician had diagnosed them as having osteoarthritis of their knees. Our number of subjects with new chronic knee pain is small, but it is consistent with epidemiologic data on the incidence of new knee pain attributable to knee OA over 3–4 years (0.5–1% per year) (27). Nonetheless, even though the prevalence of radiographic knee OA is high among elders (28), we do not know whether the development of chronic knee pain among our sample of subjects correlates with the presence of radiographic disease or disease severity.

Our study limitations include the following. The characterization of new chronic knee pain depends on a subject's ability to recall whether they had knee pain 3–4 years prior to the survey. Accurate recollection of events 3–4 years earlier is difficult, especially for elders. We tried to maximize accuracy by using prompts that have been reported to substantially improve the accuracy of recollection (20). With the exception of 1 interviewed subject, study participants did not have difficulty with their recall of past symptoms, readily recalling the time period of their participation in the Strong for Life study. Nevertheless, the major bias that might have occurred is that persons characterized as having new chronic knee pain at followup actually did have knee pain at the time of the baseline motion analysis. However, other investigators have shown that persons with knee pain alter their locomotion kinematics and kinetics to decrease their adduction moments (not increase them) (29, 30). Given this alteration in locomotion that occurs as a result of knee pain, had those reporting new chronic knee pain at followup actually had knee pain at baseline, we would have expected our results to be biased toward the null because the baseline mean peak adduction moment would be expected to be lower among those with new chronic knee pain, not higher, than those without knee pain.

Lastly, the calculation of internal forces of the joint is difficult in living humans, and the assumption that increased net joint torque translates into increased articular force can be questioned; depending on the instantaneous alignment of bones, the complex arrangement of soft tissue may instead absorb the loads rather than transmit them to articular surfaces. The net joint torque calculated from Newton-Euler inverse dynamic equations are those torques that represent the joint motions we observe in subjects' behaviors. Inverse dynamics, however, neglect muscle co-contraction (31–33) and forces developed in joint capsule, ligaments, aponeurosis, fascia, and other passive connective tissues (34, 35) because these forces do not cause motion. Indeed, instrumented hip studies have demonstrated greater joint reaction forces than determined from the inverse dynamic estimation, which tend to agree with data from muscle model studies (36–38). However, one could argue that calculation of muscle and joint reaction forces from more complicated joint models is equally prone to error because of the additional assumptions required for the muscle model (37, 39). In fact, the largest contribution to error in predicting muscle forces arises from uncertainties that propagate through the inverse dynamic solution (40), thus questioning the rationale of going beyond the calculation of net joint toques. Nevertheless, the net torque obtained from inverse dynamic equations is used extensively in the literature because it is easily estimated and probably yields rank-ordered kinetic data that are representative of the ranking in joint articular forces among different locomotor activities (41), particularly since the activities we studied were low impact (42).

Our results do suggest that biomechanical characteristics of knee loading during activities may influence the development of future knee pain. Prospective studies would be valuable to confirm our findings and to further explore the cutoff above which adduction moments during activities have deleterious effects. Our results also suggest a potential opportunity for prevention of chronic knee pain. It may be possible to lower an individual's risk of developing chronic knee pain, and perhaps osteoarthritis, if the adduction moment can be decreased during activities, such as with wedged insoles in shoes (43, 44). Additional work in this area is necessary.

In summary, subjects who developed chronic knee pain had higher adduction moments with locomotor activities 3–4 years earlier, particularly during standing and descending stairs. This study is the first to evaluate the knee adduction moment and its relation to the development of future chronic knee pain in a group who are at greatest risk of disability from chronic knee pain, elders.

Ancillary