Rehabilitation effects on compensatory gait mechanics in people with arthritis and strength impairment



Recent studies suggest that moderate exercise and other physical activity are safe and effective for older adults with osteoarthritis (OA) (1–3). These and similar exercise interventions traditionally presumed that eliminating or at least improving the impairment (e.g., pain, weakness, limited range of motion) improves function. Studies have shown this is not, in general, a valid assumption (4–6). In addition, functional performance measures such as walking speed, cadence, and stride length do not provide a mechanistic explanation for the change in gait function (2, 3, 7, 8). Understanding how function is affected by pathology and impairment may be critical in designing effective interventions for preventing disability due to OA and other musculoskeletal diseases of the elderly.

Current disability concepts recognize the importance of understanding mobility impairments for preventing disability, and furthermore suggest that emphasis be placed on determining whether interventions should be aimed at the impairment level or at the functional limitations level (9–11). Although several studies document functional improvements in people with arthritis following exercise-based intervention (12–14), no studies have specifically compared outcomes of impairment level intervention to functional limitation level intervention for older people with lower extremity arthritis. Without validated analysis techniques for understanding how lower extremity dysfunction is compensated, it is difficult to determine the value of therapeutic intervention techniques.

To address these issues, we implemented an approach for analyzing lower extremity joint compensations using the concepts of mechanical power and energy analysis, and we have applied the technique to samples of healthy and disabled elders (15–17). As described by Winter (18), joint mechanical power and energy terms reflect the underlying neuromuscular control mechanisms of human movement, and thus are potentially useful for quantifying neuromuscular adaptations and compensations for impairment. In this article, we apply the concept to a pilot study comparing a strength training (impairment level) intervention with a functional training (functional limitation level) intervention in elderly patients with lower extremity weakness and joint degeneration. Our objective was 2-fold: To measure the impairment level and functional level benefits of a strength training (ST) intervention and a functional training (FT) intervention for disabled elders; and to determine if the mechanical power analysis approach can provide a mechanistic explanation of change in function following the interventions. We briefly discuss the rationale behind the second objective; however, a more detailed review has been published elsewhere (19).

A compensation may be defined, in physiologic terms, as a substitution process whereby the function of healthy body systems fulfill the role(s) of diseased or defective body systems. The neuromuscular system is highly redundant, offering numerous possible solutions for generating the required extremity kinematics (20). This flexibility in neuromuscular patterning potentially allows one to ambulate effectively with impairments. Several studies suggest that the hip is used to compensate for age-related weakness in knee extensor and/or ankle plantar flexor muscles of otherwise healthy elders (21, 22). Elders with lower extremity impairments secondary to pathology also exhibit hip compensatory gait patterns, though the precise neuromuscular compensations and adaptations may differ from those due to aging and strength loss alone (15, 16, 23). The results of these studies suggest that elders with lower extremity musculoskeletal pathologies, such as OA, compensate for movement restrictions and limitations by decreasing ankle and knee energy expenditures, and increasing energy expenditures at the hip and low back.

In the present study, we examined disabled elders randomized into either an ST or an FT intervention (24). We then compared outcomes of lower extremity strength and gait stability, and also compared groups based on their joint mechanical energy expenditures (MEE) at the ankle, knee, and hip. We hypothesized that 1) both groups would increase in strength and gait stability relative to baseline, and 2) strength increases would be greater in the ST group compared with the FT group, and gait stability improvements would be greater in the FT group compared with the ST group. Furthermore, based on our prior studies, we hypothesized that 3) the FT group would decrease compensations (increased ankle and knee MEE and reduced hip MEE) and the ST group would increase compensations (decreased ankle and knee MEE and increased hip MEE).

Subjects and Methods


Fifteen elderly individuals (62–85 years old) participated in the study after signing informed consent according to institutional guidelines on human research. All subjects reported at least 1 lower extremity impairment for which they were referred to physical therapy services at the Massachusetts General Hospital outpatient clinic; each had at least 1 functional limitation on the short form 36 9-item physical function inventory (excluding the vigorous activity item). Eligibility criteria included age 60 years or older, cognitively intact, able to ambulate independently for at least 15 feet, and receiving permission from the treating physical therapist and/or primary care physician. Excluded were subjects with terminal illness, progressive neurologic disease, major loss of vision (legally blind), acute pain, and persons not ambulatory. Subjects were recruited through weekly screening of the outpatient physical therapy appointments for subjects' 60 years or older.


Patients were randomly assigned to one of two 6-week intervention programs (24). An ST intervention exercise program was based on resisted proprioceptive neuromuscular facilitation exercise patterns using a series of graded resistance elastic bands as described previously (25, 26). The exercises included upper extremity patterns and emphasized lower extremity movement patterns. The band use required both concentric and eccentric muscle contractions, as the subjects were instructed to control extremity movements while countering and returning with band resistance. These exercises were performed in a variety of sitting and standing positions. The level of band resistance was progressed when the subject could perform all exercises correctly and without fatigue for 10 repetitions.

The second intervention program was a novel movement control FT program. This program consisted of exercises simulating activities of daily living (such as gait, rising from a chair, reaching, stepping, and squatting down) performed at 3 different speeds (self selected, fast, and slow) with progressive levels of difficulty. When subjects completed 1 task level correctly and without fatigue the next level was introduced.

All subjects were followed weekly for guided progression and monitoring of proper performance of the exercises. Both interventions contained 10 main exercises and were individualized to the needs of each subject following a planned algorithm, to promote consistency in application of the interventions. Each exercise program included the same 5-minute warm up and cool down stretching exercises and both interventions were designed with 4 levels of progression and 4 additional advanced levels. Subjects were encouraged to exercise 3–5 times a week and completed daily exercise logs.

Lower extremity strength testing.

Bilateral lower extremity muscle strength testing, using the hand-held Compufet dynamometer (model #5025, Hoggan Health Industries, Draper, UT), was performed during the first and final intervention sessions, which occurred within 2 weeks of the formal baseline and postintervention gait analysis trials. Muscle strength testing used a “make test” isometric hold protocol documented previously (26). Knee extension and hip abduction muscle strength testing were performed while sitting. Knee flexion muscle strength was tested in a prone position. Ankle plantar-flexor and dorsi-flexor muscle strength was tested in long sitting (sitting with the knees in full extension). One practice trial was performed and the average of 2 recorded trials was used for data analysis.

Gait analyses.

Subjects walked along a 10-meter walkway at their preferred pace while full-body kinematics (segment motions) and force plate data (foot-floor reaction forces) were captured at 150 Hz with an integrated 4-camera optoelectric Selspot II (Selective Electronics, Partille, Sweden) and Kistler force plate (Kistler Instruments, Winterthur, Switzerland) system. Full-body analyses were performed by tracking arrays of infrared light emitting diodes attached securely to 11 body segments (right and left feet, shanks, thighs, arms, pelvis, trunk, and head). Six degrees of freedom kinematics of the 11 body segments were then computed, as described elsewhere (27). Segmental kinematics, anatomy, and inertial data were then combined with force plate data to compute net joint moments and powers using an inverse dynamic analysis, as previously described (28).

Gait stability.

Gait stability measures included maximum and average center of gravity (CG) velocity, double support duration, maximum moment arm, and base of support, and are described in detail elsewhere (29). Briefly, average anterior-posterior CG velocity was determined from change in CG displacement relative to change in time over a complete gait cycle (time elapsed between consecutive ipsilateral heel strikes), and maximum CG velocity was the peak anterior-posterior CG velocity. Double support duration was obtained from the second of the 2 periods during which the body is supported by both limbs, expressed as percent gait cycle. Maximum moment arm was the peak horizontal distance between the whole body CG and center of pressure (which defines the equivalent point of application of foot-floor reaction force) during single limb stance; the maximum moment arm is an indicator of how far an individual allows their center of gravity to displace from their center of pressure (30). Base of support was taken as the mediolateral distance between ankle centers at heel strike of gait (30). All gait variables were analyzed based on the average of the 2 gait trials.

Joint mechanical energy.

Kinetic analyses were performed using methods described in detail elsewhere (17, 28). Briefly, net joint power was calculated from the scalar dot product of net joint torque and joint angular velocity during stance phase of free speed gait trials for subjects at baseline, and at 6 weeks following intervention. Mechanical energy expenditure (the amount of energy expended, or work done, over a specific time interval) was then computed by integrating power over intervals of time, as described previously (9, 10). Sagittal plane ankle MEE, knee MEE, and hip MEE were documented for all subjects' gait trials and percent change was computed from pre- and postintervention data. Percent changes in ankle, knee, and hip MEE were then compared between FT and ST intervention groups.

Independent sample t-tests comparing groups and paired sample t-tests comparing pre- and postintervention measurements were performed with SPSS for Windows (SPSS Inc., Chicago, IL).


According to the weekly exercise logs, all subjects were adherent with the exercise requirement of at least 3–5 times per week. The average ± SD number of days per week that subjects performed the exercise program was 4.99 ± 1.07 for the ST group and 5.39 ± 1.27 for the FT group, a difference that was not significant (P = 0.290). The average maximum exercise levels (on the 8-level scale) attained were 3.60 ± 1.52 for the ST group and 4.38 ± 0.74 for the FT group, a difference that was also not significant (P = 0.099). There were no significant differences in weight or height between the two groups (Table 1); however, the FT group was older than the ST group (P = 0.018).

Table 1. Subject characteristics*
Treatment groupHeight, metersWeight, kgAge, yearsSexPresenting Diagnosis
  • *

    Data presented as mean (SD). For diagnostic classification, the presenting complaints from the patient's “reason for physical therapy visit” were used. ST = strength training; O = orthopedic/rheumatic, chiefly osteoarthritis; N = neurologic, including mild stroke, peripheral neuropathy; A = all others, including diffuse balance problems, cardiovascular, general weakness due to medical illnesses (e.g., diabetes mellitus), etc; FT = functional training.

ST1.65 (0.12)72.27 (11.69)70.40 (6.49)M = 2; F = 4O = 4; N = 2; A = 0
FT1.67 (0.82)73.84 (9.51)78.09 (4.60)M = 3; F = 6O = 5; N = 2; A = 2
Total1.66 (0.10)73.60 (10.06)75.02 (6.51)M = 5; F =10O = 9; N = 4; A = 2


The mean combined strength (sum of the 5 muscle groups) for all subjects showed a gain of 19% relative to baseline values (P = 0.003). Of all the muscle groups tested, the greatest strength gains were for hip abductors (P = 0.020) and ankle dorsi flexors (P = 0.006). Knee flexor strength also increased significantly (P = 0.038), as did ankle plantar flexor strength (P = 0.039) across all subjects. There were modest, though not statistically significant, gains in knee extensor strength across all subjects (P = 0.052). Although the FT group improved overall by 25.6% and the ST group improved overall by 15.6% after the 6 week intervention, there were no statistically significant differences in strength gains between the FT and ST groups (Table 2).

Table 2. Percent differences in strength between baseline and postexercise values for the ST and FT exercise groups, and significance of the between group difference*
Muscle groupFT group mean change (SD)ST group mean change (SD)P
  • *

    FT = functional training; ST = strength training.

Hip abductors16.42 (22.82)13.75 (12.85)0.40
Knee flexors9.51 (14.97)19.77 (41.15)0.25
Knee extensors33.62 (49.39)3.75 (30.42)0.11
Plantar flexors20.23 (32.49)12.56 (14.93)0.30
Dorsi flexors48.17 (66.62)28.31 (34.52)0.26

Gait stability.

Gait stability measures (Table 3) indicated significant improvements in maximum (P = 0.002) and average (P = 0.001) CG velocity across all subjects following the 6-week intervention. There also were significant improvements across all subjects in double support time (P = 0.002) and maximum moment arm (P = 0.029), but not for base of support at initial contact (P = 0.121). The FT group had significantly greater improvements in maximum and average CG velocity (P = 0.024 and P = 0.023, respectively) and had significantly (P = 0.037) greater improvement in double support time (a decrease in percent cycle time) compared with the ST group after the intervention.

Table 3. Change in gait stability measures between baseline and postexercise values for the ST and FT exercise groups, and significance of the between group difference*
Stability measureFT group mean change (SD)ST group mean change (SD)P
  • *

    ST = strength training; FT = functional training; CG = center of gravity; s = second.

Maximum CG velocity (cm/s)19.42 (16.91)5.80 (5.02)0.024
Average CG velocity (cm/s)17.83 (15.11)5.45 (4.98)0.023
Double support duration (% cycle)−2.48 (2.10)−0.63 (1.19)0.037
Maximum moment arm (cm)4.01 (6.62)0.01 (1.01)0.086
Initial base of support (cm)−1.03 (4.34)−1.78 (4.41)0.325

Joint mechanical energy.

Both groups improved (i.e., became more like healthy values reported earlier [15]) ankle kinetics during gait: ankle plantar flexor MEE increased by 56% for the functional training group and by 9% for the strength training group. Only the FT group, however, improved knee kinetics: knee extensor MEE increased by 108% for the FT group but decreased by 27% for the ST group. The FT group also improved hip kinetics: hip extensor MEE decreased by 10% for the FT group, but increased 97% in the ST group. The differences in percent change between groups were significant for the ankle (P = 0.024) and knee (P = 0.038), and borderline for the hip (P = 0.053) (Figure 1).

Figure 1.

Results of the mechanical energy analysis pre- and postintervention. The data are percent changes in joint mechanical energy expenditures (MEE; or work) from baseline. The strength training group had a small increase in ankle plantar flexor MEE, a small decrease in knee extensor MEE, and a large increase in hip extensor MEE. The functional training group had large increases in ankle plantar flexor and knee extensor MEE, and a small decrease in hip extensor MEE. The between-group differences in these changes were significant for the ankle and knee, and borderline for the hip, but suggest a profound difference in each group's response to the intervention that tends to favor the functional training intervention.


Osteoarthritis causes impairments, functional limitations, and disability that significantly impacts quality of life in older people. As understanding of OA has developed, so have attitudes toward treatment practices (31). Prior to the last decade, the majority of treatment approaches considered useful for OA were pharmacologic and focused mainly on the alleviation of pain, assuming that improved function would follow. Changing attitudes toward the management of OA, as indicated in several reviews (32–36), emphasize the paradigm shift to nonpharmacologic interventions, with physical activity, and specifically exercise, being among the most accepted forms of current treatment options (37–40).

As with pharmacologic interventions, exercise interventions are varied, and include physical therapy (12, 41), fitness walking (7, 42), aerobic exercise (1, 43, 44), resistance or isokinetic training (1, 8, 43–49), cycle ergometry (3), physical training (2), continuous passive motion (50), and aquatic therapy (51). In addition to the above studies, which included only patients with knee or hip OA, other randomized controlled studies have examined the effects of exercise in older adults with functional limitations (24–26, 52–55), many of which included patients with OA. Similar to studies that included only OA patients, these studies showed improvements in outcome measures (impairment, function, and/or disability) following exercise intervention. Indeed, there is an abundance of evidence supporting physical activity and exercise as a useful treatment for patients with OA and other ailments; however, it is still unclear what benefits these interventions have on functional ability and what modifications are needed with specific problems (25, 45).

Several studies have measured function during activities of daily living, and generally support the hypothesis that impairment and pain reduction improve physical function (1–3, 7, 8, 12, 41, 45, 46). No study to date, however, has shown how these positive changes take place and why functional improvements may occur for some patients and not others, despite similar gains in impairment and/or disability measures. Furthermore, the measures acquired by studies that have examined functional outcomes have mainly been limited to time-distance variables. These include maximum walking distance or 6-minute walk test (1, 3, 7, 41), walking time (generally over 50 ft) (12), walking velocity (2, 3, 8), cadence and/or stride length (7, 8), and chair rise and/or stair ascent/descent time (1, 12, 46). Locomotor dynamics and coordination, which can reveal compensatory mechanics caused by impairments (15, 16), have rarely been examined in a randomized controlled design for elders with OA or other functional limitations (24).

Functional gains measured in the present study are consistent with gains reported by other studies on patients with OA. Peterson et al (7) examined the effects of a walking program in a large sample of elders with knee OA (n = 102) and found a 15% increase in walking distance, a 9.1% increase in stride length at preferred walking speed, and a 17% increase in stride length at fast walking speed. Rogind et al (2) used a physical training intervention, consisting of mobility and coordination exercises, in patients with knee OA and found significant increases in walking velocity in the treatment group (13%). This improvement, however, was not statistically different from the control group, despite modest improvements in strength (2). Fisher et al (45), however, reported no change in gait variables following progressive exercise rehabilitation in patients with knee OA, despite improved strength and pain scores, and suggested gait training as a possible treatment approach to “reprogram” locomotor patterns.

The present study showed increases in strength of lower extremity muscle groups and increases in gait stability in both intervention groups following the 6-week program. This is consistent with our first hypothesis, and simply illustrates the benefits of any physical activity program for disabled elderly individuals. Although the FT group did improve lower extremity strength marginally more than the ST group (25.6% versus 15.6%), there was no statistically significant difference between groups. However, the FT group had significant improvements in gait speed and double support time compared with the ST group. This supports the second portion of our second hypothesis that the specificity of training would reflect greater functional benefits by the FT group, but apparently the ST group did not receive greater strength benefits, compared with the FT group. This leads to the question: If both groups increased similarly in strength, how did the FT group improve so much in function?

As previously discussed, strength is a basic requirement of movement. However, strength alone does not account for a person's ability to perform a task. Coordination, balance, posture, and mobility also are inherent to performance of functional activities. In this regard, we may expect an eventual plateau in performance if only a strength component is trained and improved upon. Buchner et al (56) reported a curvilinear effect of strength to functional performance in elderly individuals; after some threshold strength level is achieved, further strength does not improve function. The FT intervention we designed and implemented emphasizes variation of speed and progressive levels of difficulty for each functional task. In this regard, the subjects were required to exaggerate speed control during task performance, encouraging the experience of controlling movement during activities of daily living. This functional training program did not promote repetition of the person's current performance of the functional task, but rather progression to peak performance of each activity. Therefore, improvements noted among the FT group include strength but may also include changes in coordination and movement control.

The mechanical energy analysis suggests the FT group made better gains in functional performance through improved coordination of muscle power. Such improvements may be particularly beneficial for patients with arthritis, whose compensatory gait styles may be promoting further damage to joint tissues. Our prior studies and those of others suggest there is a trade off between distal (ankle) work and proximal (hip and low-back) work of the lower body during gait. Although strength may be a factor, it is not the only factor (16). Pain, skeletal malalignment, range of motion restrictions, and stiffness probably contribute to the presence of this compensatory strategy. Decreased energy transferred into the foot during push-off of the stance limb may require a compensatory pull-off action by the hip muscles (19, 23). Because the compensatory hip mechanics appear in early-mid stance, they might be in response to the contralateral limb when at terminal stance. Although both groups increased strength and walking speed, the FT group appeared to increase gait speed more by increasing ankle and knee power output, while those in the strengthening group increased gait speed by increasing hip power output. Thus, it appears only the FT group modified their gait to a more “healthy,” coordinated pattern, while the ST group simply exaggerated their learned compensatory style.

This study has some limitations that affect the generalizability of the results. The biomechanical analysis does not take into consideration the actual muscle forces, but only the net joint moments caused by muscles acting to control the observed movements, which furthermore cannot differentiate between contributions of uniarticular and biarticular muscles. The study sample size was small, and there was a small, yet significant, age discrepancy between groups. In addition, due to the small sample sizes, it was not feasible to statistically control for potentially confounding variables, such as baseline differences in gait speed, height, and weight (although the MEE data are normalized to weight).

Our preliminary findings do suggest, however, that mechanical energy analysis offers a very promising avenue of investigation that could have significant implications for exercise and activity in the management of rheumatic diseases. The recommendation of this study is clear: a patient-specific intervention program that targets the patient's functional limitations appears to confer greater benefits than targeting the strength impairment alone. However, several questions remain unanswered:

  • 1Are the findings of this study representative of an arthritis population? Patients in the present study had a mixture of impairments in addition to OA that make conclusions difficult to generalize. This question may be answered by replicating the present study with a homogenous sample of patients, i.e., those with only unilateral knee OA.
  • 2Are specific OA symptoms and signs important for studying compensatory movement strategies in patients with OA? Varying degrees of pain, varus/valgus malalignment, and range of motion restrictions may influence the degree and type of compensatory movement strategy (57). This question may be answered by studying compensatory movements in OA patients stratified by symptoms.
  • 3How are the compensatory movement strategies related to muscle function? The present mechanical energy analysis is based on Newtonian inverse dynamics and therefore does not directly address the role of specific muscles of the lower extremity and lumbopelvis region. This question can be addressed by extending the mechanical energy analysis to include muscle power flow using more sophisticated joint models, and perhaps incorporating dynamic electromyography.

Future studies, addressing the above questions, should enable strategies to be designed for implementing patient-specific functional training programs for patients with arthritis. A better understanding of how arthritis patients' symptoms and signs relate to body segment and muscle coordination will allow rehabilitation professionals to more accurately target functional limitations, reducing long-term disability associated with arthritis.