Effects of a modified shoe on knee load in people with and those without knee osteoarthritis

Authors

  • Kim L. Bennell,

    Corresponding author
    1. University of Melbourne, Parkville, Victoria, Australia
    • Centre for Health, Exercise, and Sports Medicine, Department of Physiotherapy, University of Melbourne, 200 Berkeley Street, Parkville, Victoria 3010, Australia
    Search for more papers by this author
    • Drs. Bennell and Hinman and Mr. Wrigley will receive royalties from Asics Oceania Pty., Ltd. if the shoe is commercialized.

  • Crystal O. Kean,

    1. University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Tim V. Wrigley,

    1. University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
    • Drs. Bennell and Hinman and Mr. Wrigley will receive royalties from Asics Oceania Pty., Ltd. if the shoe is commercialized.

  • Rana S. Hinman

    1. University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
    • Drs. Bennell and Hinman and Mr. Wrigley will receive royalties from Asics Oceania Pty., Ltd. if the shoe is commercialized.


Abstract

Objective

To evaluate the effects of a modified shoe that incorporates both lateral wedging and a variable-stiffness sole on knee joint loading in 3 populations: individuals with symptomatic and radiographic knee osteoarthritis (OA), asymptomatic overweight individuals, and asymptomatic healthy weight individuals.

Methods

Ninety participants (30 per group) underwent a 3-dimensional gait analysis across 3 test conditions: modified shoes, standard control shoes, and barefoot. For each condition, the first peak knee adduction moment (KAM) and knee flexion moment (KFM) (both expressed as Nm/[body weight × height]%) as well as the KAM impulse (expressed as Nm.s/[body weight × height]%) were determined.

Results

The modified shoes significantly reduced the peak KAM as compared to the control shoes in both the OA (P = 0.002) and the overweight (P = 0.03) groups. In the OA group, there was no significant difference in peak KAM when walking in the modified shoe as compared to walking barefoot. In the overweight and the healthy weight groups, the peak KAM when walking in the modified shoe was significantly higher than that when walking barefoot (P < 0.001). Irrespective of group, the KAM impulse was significantly reduced when walking in the modified shoe as compared to the control shoe (P < 0.001) and was significantly higher during both shoe conditions as compared to walking barefoot (P < 0.001). There was no change in the KFM between walking conditions for any group.

Conclusion

The findings illustrate that a shoe incorporating both a lateral wedge and a variable-stiffness sole can significantly reduce medial knee joint load. Further research examining the effects of these shoes on pain, function, and structural changes in the joint is warranted.

Knee osteoarthritis (OA), which predominantly involves the medial tibiofemoral joint compartment, is a major public health problem that is common in the elderly and in those who are overweight (1, 2). Individuals with OA experience significant knee pain and physical dysfunction that affects quality of life. Increased joint loading is believed to be central to the development and progression of symptomatic and structural disease (3). There is no cure for OA, and it is typically progressive in nature. Thus, nonpharmacologic treatments that reduce symptoms and facilitate self management over the long term are urgently needed. Given that increased knee load plays a role in the disease pathogenesis (3), interventions that target mechanical load are potentially beneficial for managing symptoms and minimizing structural deterioration.

Since direct measurement of joint loads in vivo is not feasible in humans, 3-dimensional gait analysis is a valuable tool for inferring dynamic loading conditions. The peak external knee adduction moment (KAM) is a reliable (4) and valid (5, 6) indicator of medial knee load. The KAM impulse (area under the KAM-time curve), reflecting both the magnitude and duration of knee loading throughout the gait cycle, may provide additional information (7). Importantly, both KAM parameters are significantly associated with structural deterioration at the knee joint (8–10) and are amenable to change with conservative interventions (11–13). Thus, the KAM is a potential target when developing new interventions aimed at maximizing knee joint health. When evaluating interventions to reduce the KAM (and thus potentially also reduce joint loading), it is important to ensure that sagittal plane moments, in particular the knee flexion moment (KFM), do not concurrently increase, as this may counteract the beneficial effects of the intervention on the knee load (14).

The foot is a key component of the lower limb closed kinetic chain, and thus, foot position and motion influence loading at the knee joint. Because of this biomechanical link, shoes can potentially increase or decrease knee loading, depending on their design features. Research shows that the KAM is increased with many currently available off-the-shelf shoes, particularly those that are more rigid and/or have high heels (12, 15–18). Given this, there has been recent interest in the study of off-the-shelf flat, flexible shoes (19) and in specially designed shoes with modifications intended to reduce KAM (12). These modifications have included grooved soles (20), variable-stiffness soles (21, 22), and laterally wedged soles (21), which were postulated to change the center of pressure at the foot and, thus, lower the KAM.

Since the development of novel footwear designs for knee OA is a very recent research and treatment focus, the published literature in this area is limited. Studies have shown that these types of modified shoes can reduce the KAM (20–22), but their effects on the KFM have not been investigated. While modified shoes have been developed primarily for people with knee OA, these shoes could also be beneficial for overweight individuals who are at risk of OA (1). Furthermore, if modified shoes are to be commercialized and made available to the general public, an understanding of their effects in healthy individuals without knee pain is also important. Given that walking mechanics differ between people with and those without knee OA and in the presence of obesity (23–26), it is possible that modified shoes could have greater unloading effects in some populations compared to others.

Our team has developed a novel recreational-style shoe with special features aimed at reducing medial knee loading in people with knee OA or those at risk of developing the disease (e.g., overweight/obese). The purpose of this study was to investigate the immediate biomechanical effects of this shoe on parameters of the KAM during walking in people with knee OA, in asymptomatic overweight individuals, and in asymptomatic healthy weight individuals. A secondary aim was to evaluate the effects of the shoe on the KFM and pain while walking.

PATIENTS AND METHODS

Participants.

Three groups of 30 participants each were recruited: symptomatic and radiographic knee OA patients, asymptomatic and overweight (body mass index [BMI] ≥25) subjects, and asymptomatic and healthy weight (BMI <25) subjects. Weight categories were based on World Health Organization definitions (27). Participants were recruited from the community via notices in local and major newspapers, the University Staff News, Facebook advertisements, community club newsletters, and posters in medical, radiology, or physical therapy waiting rooms.

All participants were required to be ≥40 years of age. For the OA group, inclusion criteria were established medial tibiofemoral knee OA identified on standing, weight-bearing radiographs (fulfilling the American College of Rheumatology classification criteria [28, 29]), average knee pain during the previous week >3 on an 11-point scale (0 = no pain and 10 = maximal pain), and pain on most days during the last month. For the asymptomatic overweight and healthy weight groups, inclusion criteria were no history of traumatic knee injury and no knee pain within the last 12 months.

Exclusion criteria for all groups were a BMI >36 (because of difficulty with placement of skin markers for gait analysis in very obese individuals), history of a lower limb joint replacement, high tibial osteotomy, spine or hip surgery, knee surgery or intraarticular corticosteroid injection within the previous 6 months, current or past (within 4 weeks) use of oral corticosteroids, any medical condition that may affect walking, any systemic arthritic condition, any major medical condition (e.g., cancer or major heart trouble), inability to walk unassisted, or the inability to understand written or spoken English.

For OA group participants in whom both knees were eligible for study, the most symptomatic knee was selected for study. For asymptomatic group participants, the study knee was selected at random.

The study was approved by the University of Melbourne Human Research Ethics Committee. All participants provided written informed consent.

Descriptive information.

Demographic information regarding age, height, and body mass were collected for each participant. The BMI was calculated by dividing the subject's body mass (in kg) by the subject's height squared (in m2). To provide an indication of static knee alignment, tibial orientation with respect to the vertical (0°) was measured with a gravity inclinometer (Acuangle; Isomed), which was attached to a set of calipers. Measurements were made with the subject in a standing position. This method was previously validated against the mechanical axis obtained from a long-leg radiograph and was found to be reliable in our laboratory (intraclass correlation coefficient 0.94) (30). Negative scores indicate varus alignment and positive scores indicate valgus.

A semiflexed weight-bearing posteroanterior knee radiograph was obtained on all OA patients to determine the Kellgren/Lawrence grade (29), the medial/lateral tibiofemoral osteophyte grade (range 0–3), and the medial/lateral joint space narrowing grade (range 0–3) (31). Participants with OA also completed the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). The WOMAC is a disease-specific self-report questionnaire consisting of 24 questions measuring 3 subscales: pain (range of scores 0–20), stiffness (range of scores 0–8), and physical function (range of scores 0–68) (32). Lower scores indicate less pain/stiffness and better physical function.

Shoes.

The modified shoes (Gel Melbourne OA shoe, which was fabricated especially for this study by Asics Oceania) were recreational-style walking shoes with 2 specific built-in design features aimed at reducing KAM: 1) a triple-density compression-molded ethylene vinyl acetate midsole (where the lateral midsole is stiffer than the medial) extending more than two-thirds of the shoe length, and 2) a thin, firmer lateral wedge, with an angulation of 4–6°, extending from the center of the shoe to the lateral edge and attached to the “sock-liner” (minimal insole).

The control shoes (Asics Oceania) were a standard recreational walking shoe similar in appearance to the modified shoe but without the variable-stiffness sole or lateral wedge. Photographs of the modified shoe and the control shoe are shown in Figure 1.

Figure 1.

Women's (top) and men's (bottom) shoes used in the study. The external appearance of both the modified and control shoes was the same.

Gait analysis.

Prior to the gait analysis, participants completed 5 walking trials over a 10-meter walkway while wearing their own regular walking shoes. Subjects were instructed to walk at their self-selected speed. During these trials, the time to walk between 2 photoelectric timing gates set 4-meters apart was recorded and was used to control speed during the gait analysis.

Participants underwent 3-dimensional gait analysis under 3 test conditions: wearing the modified shoes, wearing the control shoes, and barefoot. The tests were performed in random order. The study participants, but not the clinicians, were blinded with regard to shoe type. Passive reflective markers were secured to the skin/shoe according to the Vicon Plug-In-Gait marker set. This included markers placed bilaterally over the anterior and posterior superior iliac spine, lateral thigh, lateral femoral condyle, lateral shank, lateral malleolus, and on the shoe itself, over the posterior calcaneus and the base of the second metatarsal. Additional markers placed over the medial knees and ankles were included during the initial static standing trial to determine relative positioning of joint centers.

Kinematic data were collected at 120 Hz using a 12-camera Vicon MX motion analysis system, and kinetic data were collected at 1,200 Hz using 3 floor-mounted force plates (Advanced Mechanical Technology) in synchrony with the cameras. For each condition, 5 trials with clean force plate strikes by the study limb and within 5% of initial walking speed were collected. Joint moments were calculated using inverse dynamics (Vicon Plug-In Gait, version 2). The moment values (expressed in Nm) were then normalized by dividing by body weight times height (33–35) and expressed as a percentage. The first peak KAM and KFM (Nm/[body weight × height]%) and positive KAM impulse (Nm.s/[body weight × height]%) (representing the positive area under the KAM–time curve) were determined for the walk trials and were then averaged for the test condition.

Shoe comfort and pain ratings.

Immediately following each shoe walking condition, participants were asked to rate shoe comfort using an 11-point numerical rating scale (NRS) of 0–10, where 0 = extremely uncomfortable and 10 = extremely comfortable. Participants with OA were also asked to rate knee pain experienced during each of the 3 walking conditions using an 11-point numeric rating scale, where 0 = no pain and 10 = worst pain possible.

Sample size calculation.

We based the sample size calculation on detecting a statistically significant interaction effect between walking condition and study group at an alpha level of 0.05. Based on this, 10 patients per cell were required (total of 9 cells for a 3 × 3 analysis) to show a large interaction effect (0.4) with 80% power and using a 2-tailed hypothesis. Accordingly, we required ∼90 participants in total (30 per group).

Statistical analysis.

One-way analysis of variance (ANOVA; for age, BMI, tibial alignment, and walking speed) and chi-square test (for sex) were used to compare group demographic and clinical characteristics. The gait variables were compared using mixed 3 × 3 ANOVAs, with walking condition and group as the independent variables. Group main effects are not reported, since the differences in loading parameters across groups without reference to the footwear condition were not of interest in this study. When a significant main effect of walking condition or significant walking condition × group interaction was noted, Tukey's post hoc test was used to further examine within-group differences. To account for the potential influence of age, alignment, and walking speed, a secondary analysis adjusted for these covariates (analysis of covariance) was performed. Furthermore, we also reanalyzed the results, removing data for any participant with predominantly lateral radiographic disease as compared to medial.

A mixed 2 × 3 ANOVA was used to examine differences in shoe comfort between walking conditions (modified and control shoe only) and between study groups. A group main effect is not reported, since the differences in shoe comfort across groups were not of interest. For the OA group only, knee pain ratings were compared between walking conditions using one-way repeated-measures ANOVA with Tukey's post hoc test. Statistical significance was set at an alpha level of 0.05. All analyses were completed using Statistica v7.0 software (StatSoft).

RESULTS

Demographic and clinical data are summarized in Table 1. The OA group was significantly older and walked slower than both of the asymptomatic groups, and they had a more varus tibial alignment than the overweight group. Both the OA group and the overweight group had a significantly higher BMI than the healthy weight group. All participants in the OA group had medial tibiofemoral OA on radiographs, with medial osteophytes and/or medial joint space narrowing. Two had worse disease in the lateral compartment, whereas 1 had OA of equal severity in the medial and lateral compartments. The OA group reported mild-to-moderate levels of pain and physical dysfunction, with a similar number of participants demonstrating mild, moderate, and severe radiographic disease severity.

Table 1. Characteristics of the study subjects, by study group*
 OA (n = 30)Overweight (n = 30)Healthy weight (n = 30)P
  • *

    Osteoarthritis (OA) severity was assessed radiographically and was graded according to the Kellgren/Lawrence (K/L) scale. Lower scores on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) indicate less pain/stiffness and better function. BMI = body mass index.

  • For the comparison of between-group differences, by analysis of variance.

  • P < 0.05 versus the OA group.

  • §

    P < 0.05 versus the healthy weight group.

Age, mean ± SD years63.3 ± 9.755.3 ± 8.957.3 ± 10.20.004
Sex, no. (%)    
 Female17 (57)16 (53)21 (70)0.38
 Male13 (43)14 (47)9 (30)
BMI, mean ± SD kg/m228.6 ± 3.6§28.7 ± 2.6§21.8 ± 1.8<0.001
Tibial alignment, mean ± SD degrees−2.5 ± 3.9−0.1 ± 2.5−1.7 ± 1.70.006
Walking speed, mean ± SD meters/second1.29 ± 0.251.49 ± 0.201.51 ± 0.240.001
K/L grade of OA severity, no. (%)    
 Grade 211 (37)
 Grade 39 (30)
 Grade 410 (33)
WOMAC domain scores, mean ± SD    
 Pain (range 0–20)7.6 ± 3.1
 Stiffness (range 0–8)4.1 ± 1.6
 Physical function (range 0–68)26.0 ± 9.2

The mean peak KAM, KAM impulse, and peak KFM values for each condition/group (and all participants combined) are presented in Table 2. For the peak KAM, the walking condition × group interaction was significant (P = 0.02). Tukey's post hoc tests revealed that the modified shoes significantly reduced peak KAM as compared to the control shoes, both in the OA (P = 0.002) and the overweight (P = 0.03) groups. There was no difference in peak KAM between the 2 shoe conditions for the healthy weight group (P = 0.32). The reduction in peak KAM when walking in the modified shoe compared to the control shoe corresponded to a mean ± SD percentage change of −7.9 ± 8.9% for the OA group (range −42.9% to 6.4%), −4.7 ± 5.4% for the overweight group (range −12.5% to 8.6%), and −3.5 ± 8.2% for the healthy weight group (range −15.9% to 21.9%).

Table 2. Peak KAM, KAM impulse, and KFM for each walking condition, by study group*
 Modified shoesControl shoesBarefoot
  • *

    Values are the mean ± SD. KAM = knee adduction moment; KFM = knee flexion moment; OA = osteoarthritis; BW = body weight; Ht = height.

  • P < 0.05 versus the control shoes group.

  • P < 0.001 versus barefoot group.

  • §

    P < 0.001 versus modified shoes group.

OA group (n = 30)   
 Peak KAM, Nm/(BW × Ht)%3.73 ± 1.304.02 ± 1.353.54 ± 1.27
 KAM impulse, Nm.s/(BW × Ht)%1.17 ± 0.471.26 ± 0.481.10 ± 0.47
 Peak KFM, Nm/(BW × Ht)%3.62 ± 1.823.53 ± 1.863.50 ± 1.94
Overweight group (n = 30)   
 Peak KAM, Nm/(BW × Ht)%4.68 ± 1.284.92 ± 1.374.14 ± 1.14§
 KAM impulse, Nm.s/(BW × Ht)%0.97 ± 0.211.04 ± 0.210.90 ± 0.20
 Peak KFM, Nm/(BW × Ht)%5.22 ± 1.835.02 ± 1.915.21 ± 2.01
Healthy weight group (n = 30)   
 Peak KAM, Nm/(BW × Ht)%4.36 ± 1.094.54 ± 1.143.93 ± 0.98§
 KAM impulse, Nm.s/(BW × Ht)%0.98 ± 0.221.06 ± 0.230.94 ± 0.22
 Peak KFM, Nm/(BW × Ht)%4.94 ± 1.924.83 ± 1.754.94 ± 1.97
All groups combined (n = 90)   
 Peak KAM, Nm/(BW × Ht)%4.26 ± 1.284.48 ± 1.333.87 ± 1.15
 KAM impulse, Nm.s/(BW × Ht)%1.04 ± 0.331.12 ± 0.340.97 ± 0.33§
 Peak KFM, Nm/(BW × Ht)%4.64 ± 2.014.48 ± 1.934.57 ± 2.10

Individual percentage changes in the first peak KAM while wearing the modified shoes compared to the control shoes in each group are plotted in Figure 2. For the OA group, there was no significant difference in peak KAM when walking in the modified shoes as compared to barefoot (P = 0.19). For the overweight and the healthy weight groups, the peak KAM when walking in the modified shoe was significantly higher than that when walking barefoot (P < 0.001). The peak KAM for all 3 groups when walking in the control shoe was significantly higher than that when walking barefoot (P < 0.001). These results were unchanged when covariates were included in the analysis or when the 3 participants with predominantly lateral radiographic disease were excluded from the analysis.

Figure 2.

Individual changes in the first peak knee adduction moment (KAM) when walking in the modified shoe, reported as the percentage change from the control shoe condition. A, Osteoarthritis group. B, Overweight group. C, Healthy weight group.

For the KAM impulse, there was a significant main effect for walking condition (P < 0.001), but there was no significant walking condition × group interaction (P = 0.30). Tukey's post hoc tests indicated that, irrespective of group, the KAM impulse was significantly reduced when walking in the modified shoe as compared to the control shoe (P < 0.001) and was significantly higher when walking under both shoe conditions as compared to walking barefoot (P < 0.001). The reduction in KAM impulse when walking in the modified shoe compared to the control shoe corresponded to a mean ± SD percentage change of −8.7 ± 11.2% for the OA group (range −55.3% to 10.1%), −7.3 ± 5.6% for the overweight group (range −20.4% to 2.9%), and −7.8 ± 6.6% for the healthy weight group (range −23.0% to 6.3%). These results were unchanged after adjusting for covariates and after the 3 individuals with predominantly lateral radiographic disease were excluded.

Individual percentage changes in the KAM impulse with the modified shoes compared to the control shoes for all 3 groups are plotted in Figure 3. For the KFM, there was no main effect of walking condition (P = 0.06), nor was the walking condition × group interaction significant (P = 0.60). These results were unchanged when covariates were included in the analysis or when the three participants with predominant lateral radiographic disease were excluded.

Figure 3.

Individual changes in the knee adduction moment (KAM) impulse when walking in the modified shoe, reported as the percentage change from the control shoe condition. A, Osteoarthritis group. B, Overweight group. C, Healthy weight group.

Comparison of shoe comfort revealed a significant main effect for walking condition (P < 0.001), but no walking condition × group interaction (P = 0.18). Across groups, the control shoe was rated as significantly more comfortable than the modified shoe, but the mean difference was minimal (mean ± SD NRS comfort score for the control shoe 7.4 ± 1.8 and for the modified shoe 6.7 ± 2.1, with a mean difference of 0.7 [95% confidence interval 0.4–1.1], P < 0.001). For the OA group, knee pain ratings did not differ significantly between walking conditions (P > 0.05). The mean ± SD knee pain was 3.1 ± 2.0 when walking in the modified shoe, 3.5 ± 2.3 when walking in the control shoe, and 3.3 ± 2.2 when walking barefoot.

DISCUSSION

Given that many people require supportive recreational-style shoes for everyday comfort as well as for physical activity such as walking, there is an obvious need for recreational shoes that reduce knee load rather than increasing it, as many currently available shoes do (15, 16). We evaluated a modified shoe in 3 experimental groups: people with symptomatic and radiographic knee OA, asymptomatic overweight individuals, and asymptomatic healthy weight individuals. Understanding the biomechanical effects in a range of groups is important for a shoe that could be commercialized and thus available to a wide market of consumers, not all of whom may necessarily have knee OA. The results of this study showed that a modified shoe incorporating 2 specific design features significantly reduced peak KAM in the OA and overweight groups and improved KAM impulse in all 3 groups during walking.

For the peak KAM, a significant unloading effect of the modified shoe was evident in both the OA and the overweight groups, but it was not evident in the healthy weight group. Indeed, the greatest percentage reduction in knee load with the modified shoe was observed in the OA group, resulting in peak KAM levels approximating those of barefoot walking. We also found that the modified shoe significantly reduced the KAM impulse as compared to control shoes by an average of 7–9% across all 3 groups. While the peak KAM and KAM impulse are somewhat correlated (36), they reflect different aspects of loading and thereby provide different information about the effects of the shoe (7, 37). Since the KAM impulse is determined by the amplitude of the moment integrated over the interval of the stance time, a lower KAM impulse with the modified shoe indicates less cumulative load on the joint across each stance phase.

The mechanisms underpinning the reductions in KAM with the modified shoe are most likely due to a reduction in the frontal plane knee–ground reaction force lever arm as has been noted with a variable-stiffness sole shoe (38) and with lateral wedge insoles (36). However, explanations for the differential group response in peak KAM to the modified shoe are unclear. It is likely that the shoe response varies within individuals according to a complex interaction of factors, including knee alignment, walking patterns, and foot biomechanics. Further detailed evaluation of the mechanism behind the KAM reduction as well as the biomechanical effects at the hip and ankle joints with the modified shoes may provide an explanation, but this is beyond the scope of our study.

We additionally investigated the KFM, since it also contributes to the medial joint compressive load and can be differentially affected by interventions under some circumstances (14). For example, in a single case study involving a patient with an instrumented knee replacement, reductions in the first peak KAM with 2 different gait modifications did not correspond to reductions in medial joint contact force (14). This lack of correlation was due to a concomitant increase in the KFM counteracting the beneficial reduction in the KAM. However, another study found that variable-stiffness shoes reduced both the peak KAM by 13.3% and the medial joint contact force by 12.3%, with a change in one positively correlating with a change in the other (6). Given these findings in a similarly modified shoe, together with the fact that our shoe reduced the KAM but did not increase the KFM, we are optimistic of an overall medial joint unloading effect.

Our results are consistent with the reduction in peak KAM reported for other types of modified shoes. In a large study of 79 individuals with medial knee OA, Erhart et al (22) found average reductions in the peak KAM with variable-stiffness sole shoes ranging from 2.4% at slow speed to 4.5% at normal speed and 6.2% at fast speed (22). Our 7.9% reduction at a similar normal speed is larger than the values reported in that study and may have been accentuated had we also tested at faster walking speeds. The larger effect found in our study may be due to the presence of the dual-design shoe features. Conversely, our nonsignificant peak KAM reduction of 3.5% in the healthy weight group was less than that reported in 14 healthy (albeit younger) individuals in a study where a variable-stiffness sole shoe reduced the KAM by 7.2% (21).

Another type of modified shoe described in the literature, the “mobility shoe,” reduced the peak KAM by 8% compared to self-chosen walking shoes and by 12% compared to a recreational control shoe in people with knee OA (20). That shoe is vastly different in design from our modified shoe, being extremely lightweight, with a thin sole containing specialized grooves to allow greater flexibility during walking. Similarly, while not strictly “modified,” an off-the-shelf minimalist and flexible shoe (Moleca) was found to give peak KAM and KAM impulse values comparable to those of barefoot walking in elderly women with and without knee OA (39). A commercial rocker-sole shoe reduced the peak KAM in overweight males, although concomitant increases in knee muscle activation could actually lead to higher knee compressive loads (but this was not specifically measured) (40).

Similar to the findings of other studies of modified footwear (20–22), our study showed diverse, sometimes reversed (albeit rarely), effects with the modified shoe among all 3 study groups, suggesting that individual characteristics mediate the effectiveness of the shoes. A comparison of the characteristics of the study participants (such as age, sex, body mass, alignment, and Kellgren/Lawrence grade) did not provide any meaningful insight into why some individuals experience an increase in peak KAM or KAM impulse when wearing the modified shoes. A small number of participants in the OA and overweight groups (4 of 30 in each group [13%]) responded to the modified shoe with an increase in peak KAM as compared to the control shoe, whereas a larger proportion of the healthy weight group showed an increase with the modified shoe (10 of 30 [33%]). Only 10% of the healthy weight group showed an increase in the KAM impulse with the modified shoes, similar to the proportions seen in the 2 other groups. This suggests that the majority of people experience reduced load across the entire stance phase with the modified shoes. However, it may be prudent to limit recommendations for the shoe to people who have knee OA and to asymptomatic overweight individuals.

While our results demonstrate reduction in knee load with the modified shoe, its symptomatic effects need to be evaluated. Although in the OA group, we found no immediate reductions in knee pain during walking, this may partly relate to the relatively low pain levels experienced during testing (around 3 units out of 10) and/or the short duration of shoe use. Participants generally rated the shoes as comfortable, although a better indication of shoe comfort would be gained following a longer period of use. In the only randomized controlled trial to date, a shoe with a variable-stiffness sole and confirmed load-reducing ability (6) worn for at least 4 hours per day significantly improved pain and function compared to a control shoe at 6 months (41), but not at 12 months (42), in people with medial knee OA. This nonsignificant finding in the longer term may relate to the large dropout rate and limited statistical power to detect a difference. Importantly, despite changes in hip and ankle biomechanics with this shoe (22), the limited adverse events did not appear to differ from those reported with the control shoe (42).

No previous trials have investigated the structural effects of load-modifying footwear. Previous research has shown that a higher peak KAM predicts the onset of knee symptoms (9) and that a 1 unit Nm/(body weight × height)% greater peak KAM translates into a 6-fold increase in the risk of radiographic disease progression in people with established knee OA (10). While wearing the modified shoes, the OA and overweight groups achieved a mean reduction of 0.29 and 0.24 Nm/(body weight × height)% in the peak KAM, respectively, as compared to control shoes. This one-fourth unit of knee load reduction accumulated over thousands of steps per day could possibly elicit a protective structural effect; however, future research into the effects of prolonged wearing of the modified shoe and its effects on structural changes are warranted. The modified shoes also reduced the KAM impulse, and we have recently shown that this parameter, rather than peak KAM, was independently associated with greater loss of medial tibial cartilage volume over 12 months in people with medial knee OA (8). It should be noted that no conservative load-reducing treatment has as yet proven successful at slowing disease progression (43, 44); however, further research is needed.

The strengths of this study include the large sample size with adequate statistical power to detect clinically relevant effects, as well as the inclusion of different study groups. The order of walking condition was randomized, speed was controlled across walking conditions within each group, and participants were blinded as to shoe status. A limitation is that the immediate biomechanical effects were evaluated following a few minutes of shoe acclimation. However, a study of a shoe with a variable-stiffness sole found that the immediate reduction in KAM at baseline was similar to that seen after 12 months of daily shoe use, suggesting that immediate effects may be indicative of those seen in the longer term (42). Although all participants in the overweight and healthy weight groups were asymptomatic, no radiographic evaluation was performed, and we therefore cannot conclude that there is no radiographic OA. Our OA group was also significantly older and walked slower than our asymptomatic groups; however, results were unchanged when these differences were statistically controlled.

From a clinical perspective, modified shoes are neither a burdensome nor time-consuming treatment option and could be used as regular everyday/walking shoes. Compared to other mechanical remedies (such as knee bracing), modified shoes are readily and easily applied and are likely to be more comfortable. Given their favorable effects on knee load, modified shoes have the potential to offer symptomatic and structural benefits for people with OA and may also be of value for overweight individuals at risk of knee OA. Further research is needed to confirm these potential load-modifying benefits.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Bennell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Bennell, Kean, Wrigley, Hinman.

Acquisition of data. Kean.

Analysis and interpretation of data. Bennell, Kean, Wrigley, Hinman.

ROLE OF THE STUDY SPONSOR

Asics Oceania Pty., Ltd. had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Asics Oceania Pty., Ltd.

Ancillary