What Is Known
- Resistance exercise training can promote pain relief in persons with knee osteoarthritis, but effects of this exercise on physical function performance are mixed.
What Is New
- This study tested whether muscle actions type (eccentric vs. concentric) provides different benefits on knee pain and functional ability among older adults with knee osteoarthritis. The two muscle action types increased leg strength but did not differentially improve functional performance. Only concentric exercise reduced ambulatory pain. Clinically, any resistance exercise can be used to manage knee osteoarthritis pain but concentric actions may provide additional pain relief during walking.
Physical activity is a key strategy to manage pain, improve physical function, and increase mobility among individuals with knee osteoarthritis (OA). The evidence-based consensus guidelines provided by the Osteoarthritis Research Society International (OARSI) for nonsurgical management of knee OA include resistance exercise training (RT) as a core treatment.1 Meta-analyses show that RT in general can have a positive effect on pain relief and physical function and that greater improvements in pain tend to be associated with higher functional levels.2 Systematic reviews report that nonweight-bearing RT is more effective for pain relief compared with weight-bearing RT or aerobic exercise.3
For the last few years, federal research programs have focused on determining the mechanisms underlying the health benefits of exercise and physical activity mediation of disease states. What is known from the mechanistic perspective is that RT exercises involve muscle actions that are concentric and eccentric. Eccentric actions are essential in daily activities, such as stair descent, squatting, or sitting into a chair. Concentric actions are vital in stair ascent, standing up, and rising from a chair. Although RT is widely used to manage knee OA, the role of RT muscle actions on OA functional pain symptoms4 and translation to gains in physical functions are not known. This information could advance the knowledge of potential mechanical mechanisms underlying changes in OA symptoms and functional gains and could guide the development of personalized RT programs. Eccentric exercise actions are characterized low energy cost, high force production, hypertrophic impact, and favorable effect on fall risk and physical function and mobility.5,6 Eccentric resistance training may also increase volitional drive and reduce corticospinal inhibition to muscle more than concentric training in OA.7 In theory, these neural output features may improve functionality in activities, such as stair climb or ambulation. Most resistance exercise interventions for knee OA therefore have overloaded muscle and progressed concentric actions but understimulated muscle during the eccentric actions.8 Two promising studies have shown reductions in knee OA functional pain and accentuated improvements in physical function when eccentric muscle actions were added to manual RT or to isokinetic machine training programs.9,10 These RT training methods, however, were not designed to enable direct comparison of the two muscle action types on OA outcomes.
Therefore, the purpose of this study was to address this evidence gap by directly comparing the effects of concentrically focused resistance training (CNCRT) and eccentrically focused resistance training (ECCRT) on functional pain and physical functional performance among individuals with knee OA. We hypothesized that ECCRT would elicit greater improvements in functional pain and several performance measures compared with CNCRT.
This was a secondary analysis of a randomized, controlled, single-blinded parallel study. This study adhered to the Consolidated Standards of Reporting Trials 2010 guidelines for reporting parallel group randomized trials and reports the required information accordingly (see Supplemental Checklist, Supplemental Digital Content 1, http://links.lww.com/PHM/A997). The registration number for this study on ClinicalTrials.gov was NCT01245283.
Participants and Screening
Older adults with knee OA were recruited from the Gainesville area using the UF Orthopedics Clinics, the Clinical Trials Register, and a mailing list provided by the UF Claude Pepper Aging Center from November 2010 to December 2012. The inclusion criteria were as follows: adults aged 60–85 yrs; presence of tibiofemoral knee OA for 6 mos or longer (defined using American College of Rheumatology criteria); bilateral standing anterior-posterior radiograph demonstrating Kellgren and Lawrence OA grade 2 or 3 of 4; free from other musculoskeletal limitations that would impede participation in exercise; and free of abnormal cardiovascular responses during the screening graded maximal walk test. The exclusion criteria were as follows: surgery to either knee within the last 12 mos; lumbar radiculopathy; vascular claudication; knee pain due to isolated patellofemoral syndrome or chondromalacia; received corticosteroid or hyaluronic acid injections within last 3 mos; and have added new over-the-counter or prescription pain medication within 2 mos of study participation. Eligibility criteria were reviewed by the study coordinator and the study physician to ensure that the appropriate participants were enrolled. This study was approved by the University of Florida Institutional Review Board. All procedures on human subjects were performed in accordance with the Helsinki Declaration of 1975, as revised in 1983. All participants provided written informed consent to participate. Figure 1 depicts the Consolidated Standards of Reporting Trials study flow diagram.
After determining eligibility, all enrolled participants completed a progressive walking symptom-limited Naughton exercise test in an academic research laboratory for screening to continue in the program. The testing sessions followed the American College of Sports Medicine guidelines with electrocardiogram heart monitoring and blood pressure measures. Open-circuit spirometry was used to determine the rate of oxygen use and carbon dioxide production using a metabolic cart (VIASYS; CareFusion Corp, San Diego, CA). All tests were supervised and reviewed by the study physician. Sample size was determined based on knee pain subscore improvements (on the Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC]) with resistance exercise as described in our previous analysis.11
Once cleared for further participation, participants completed two additional laboratory visits at baseline for functional performance testing and quality of life measures. At month 4, participants returned to the laboratory three times for repeat aerobic fitness testing and physical functional testing.
Randomization and Blinding
Participants were randomly and equally assigned to one of three study groups in blocks of two: a CNCRT, an ECCRT, and nonexercise, wait-list control group (CON). A computer-generated list and hidden sequencing of the individual assignment were used for randomization and provided to the participants by one coordinator. This was a single-blinded design as the PI, coordinators, and exercise physiologists who collected measures did not know the group allocation of each participant. Group assignments were placed in opaque, numbered sealed envelopes and each new enrolled participant opened an envelope to receive the group assignment. Training was supervised by another co-investigator and performed by exercise physiologists.
Resistance Exercise Interventions and Control Group
The two resistance training groups trained on MedX clinical resistance exercise machines using the general guidelines described by the American College of Sports Medicine. All participants received an informational packet of healthy behaviors (Centers for Disease Control Physical Activity for Everyone and Nutrition for Everyone; American Heart Association Physical Activity in Daily Life). Participants in the CNCRT group performed two resistance exercise sessions per week. One set of each of the following exercises was completed during each session: leg press, knee flexion, knee extension, calf press, chest press, seated row, shoulder press, and biceps curl. For each set, 12 repetitions were performed at a resistance load of 60% of the one-repetition maximum (1RM) for that exercise.11 The effort of performing the exercise set was subjectively rated using a 6- to 20-point rating of perceived exertion scale. As the participant adapted, the effort felt less, and the resistance load was raised for the set to keep the rating of perceived exertion value at approximately 17–18 of 20 points for each exercise over the study duration.
Participants in the ECCRT group also trained on modified MedX machines. Enhanced eccentric training continually performs the eccentric muscle action with the equivalent of the 1RM and to sequentially reduce the load to 60% of 1RM for the concentric muscle action. Rating of perceived effort was used in this group, and progressive loading was as described for CNCRT. The repetition structure on the eccentric exercise machine and comparative concentric exercise machine was adjusted to equalize the work performed between the study groups.
Participants randomized to the CON group continued performing their normal activities for 4 mos. Offers were made to each individual in the group to complete either the CNCRT or ECCRT program after the control period. Weekly telephone contact was made to help encourage adherence to the healthy behaviors guidelines and to provide attention to this group.
Functional Pain Ratings
Pain symptoms were collected during each trial of each functional assessment using the 11-point Numerical Pain Rating scale (NRSpain; where 0 = no pain and 10 = worst imaginable pain). Functional pain score was the primary outcome of the study. The NRSpain has good psychometric properties, similar to the WOMAC function scale, is valid in knee OA and has moderate to large responsiveness with treatments.12 A 1-point reduction in pain or reduction by 15% represents a minimal clinically important difference, with reductions of 30%–36% considered meaningful reduction in pain severity.
Physical Function Assessments
Four physical function performance test secondary measures were administered at baseline and month 4. These tests are part of the recommended set of performance-based measures reflecting typical activities important to persons with knee or hip OA by the Osteoarthritis Research Society International.13 These included the chair rise test, stair climb test, walking gait test, and 6-min walking test. Strength was assessed using the 1RM method. One-repetition maximum values were determined using the following method: for each machine exercise, a warm up of five repetitions at a low weight was followed by three repetitions at a higher weight. Single lifts were performed at progressively higher loads until the exercise could not be performed or performed with good form. Rest periods between each lift for each exercise were 60 secs.
Chair Rise Time and Stair Climb Time
Chair rise time was measured as the time required for the participant to move from a sitting to a full standing position. Participants sat in an armless, straight-backed chair (42.5-cm seat height, 45-cm seat depth) and had their arms folded in front of their chests. On a standard countdown cue, participants rose from the chair as quickly as possible. This transitional activity is common in daily life and may reflect muscle strength and dynamic balance. Stair climb time was measured by having the participants walk up one standard flight of stairs as quickly as possible (12 stairs; 18 cm high, 30.5 cm deep). One handrail was permitted, but the use of the legs alone was encouraged. The chair rise test and stair climb tests are used in older adults with knee OA and are reliable (intraclass coefficient values range = 0.94–0.96).14 Both tests were repeated three times, and the trial times were averaged for data analysis.
Gait Parameters at Self-Selected Speed
Participants walked on an 8-meter-long gait mat at a self-selected speed (GaitRite; CIRSystems Inc, Havertown, PA). The participants performed three acclimation trials to reduce the learning effect. The temporal-spatial parameters collected included velocity, cadence (steps per minute), step length, step width, and single support times (when only 1 foot was making contact with ground). The coefficient of variation CV (percent) of specific measures reflect relative gait variability, and coefficient of variation = within person standard deviation of the measure / mean over the gait cycle.15 Increased variability of gait parameters is linked to decreased gait stability, complexity, and increased risk of falling.16 Given that the variability of most gait parameters described by the coefficient of variation is speed dependent, the most reliable and reproducible gait patterns occur near the self-selected walking speeds.15 Walks were repeated three times, and the scores were averaged and reported. These were secondary measures.
Six-Minute Walk Test
Each participant performed a 6-min walk test in the laboratory’s 24-meter hallway according to Osteoarthritis Research Society International standards. Distance was marked at 3-meter intervals. Participants were instructed as follows: “Cover as much ground as fast as you can, but do not push yourself to a point of overexertion or beyond what you think is safe for you. You may rest or sit if needed in the test.” Regular, standardized encouragement was provided by the coordinators flanking each side of the ends of the hallway (“Keep up the good work, you are doing really well”). The NRSpain scale was used to capture the knee pain severity scores at rest, at 1-min intervals during the test and to 5-min posttest. No assistive devices were used for this test. The measures captured here were secondary outcomes.
All participants were provided a StepWatch step activity monitor (SAM; Cyma, Seattle, WA) by the study team to wear during all nonsleeping hours as an estimate of physical activity level and return after use. This is an accurate, dual-axis accelerometer devised to count steps in individuals with disabilities, abnormal or slow gaits, or lower limb prostheses. This watch has coefficients of 3.0%–51.3% with higher variation for short periods of 1–2 days.17 Thus, a 7-day tracking period was completed by each participant to optimize reliability of the StepWatch output.
Statistics were conducted in IBM SPSS Version 25.0 (Armonk, NY). Differences in baseline categorical measures across concentric (CNCRT), eccentric (ECCRT), and control (CON) groups were assessed using χ2 tests. Differences in baseline continuous measures across study groups were assessed with analysis of variance, using the Tukey-Kramer test for pairwise comparisons, which also adjusted for multiple comparisons using the Bonferroni method. Before analyses, nonnormal measures were log transformed. Baseline values for all outcome variables were conducted using a one-way analysis of variance with a Tukey post hoc test. For the functional and pain outcomes, two analytical approaches were used: per-protocol analysis (inclusion of participants who did not have any violations to the study protocol) and the modified intent-to-treat concept (inclusion of all participations who were randomized to the study minus those who were deemed ineligible after randomization).18 For both approaches, data were analyzed using general linear models. These models included time (pretraining, posttraining) and study group (CON, CNCRT, ECCRT) as main effects, with an interaction model between time and group. Covariates in the models included age, duration of pain symptoms, and body mass index. A significant time × group interaction would indicate that change in outcome from pretraining to posttraining differed among groups. Changes in functional pain scores are reported in raw numbers and as a percent change from pretraining to posttraining. Partial η2 was used to estimate effect size, with values of 0.02 (small), 0.13 (medium), and 0.26 (large). For the intent-to-treat analysis, if participants began but did not finish the study, the last data observations were carried forward.19 An α level was considered statistically significant if less than 0.05. Given that there were no major differences in the overall results based on approach type, we present the per-protocol outcomes in the results and the modified intent-to-treat results as supplementary tables.
Baseline characteristics of the three study groups are shown in Table 1. The study groups were well matched with respect to several demographics. However, the ECCRT group had 16.6%–27.7% fewer participants with obesity compared with the two other groups (P = 0.005). This same group had an average of 5-yr longer duration of knee pain symptoms than the other groups (P = 0.009).
Exercise-Induced Strength Improvements
There were significant group × time interactions for all muscle 1RM strength measures (all P < 0.05), which demonstrated efficacy of the training programs. For the leg musculature, the mean relative strength gains for the CNCRT, ECCRT, and CON from pretraining to posttraining were as follows: leg press (33.5%, 32.8%, and −2.2%, respectively), knee extension (29.1%, 20.2%, and −7.4%, respectively), and knee flexion (20.8%, 19.1%, and −0.5%, respectively).
Temporal-spatial measures and variability of gait during walking at a self-selected speed at baseline and month 4 are presented in Table 2 (per-protocol analysis). There were no baseline group differences in any gait measure. There were no significant group × time interactions for any gait variable (partial η2 range = 0.005–0.092, small effect). However, main effects existed for group, where the CNCRT group had slower velocities, cadence, step length, wider steps, and less single leg support time than the remaining groups (P < 0.05). The intent-to-treat analysis did detect any significant interactions or main effects of group or time (Supplemental Table 1, Supplemental Digital Content 2, http://links.lww.com/PHM/A998).
Physical Function Performance Measures
Baseline 6-min walk test distance was lowest in the CNCRT group compared with the other two groups. The per-protocol analysis results of the functional tests of chair rise, stair climb, and the 6-min walk are shown in Table 3. There were no significant group × time interactions for any of the timed scores of the performance tests (partial η2 range = 0.040–0.078, small effect). The intent-to-treat analysis also did not detect significant interactions for these outcomes, but a significant main effect for group was found, where the CNCRT group demonstrated shorter walking distances during the 6-min walk test (Supplemental Table 2, Supplemental Digital Content 3, http://links.lww.com/PHM/A999; P = 0.05).
Functional Pain Scores
Baseline chair rise pain scores were lowest in the CON group compared with the other two groups. Pain scores during the physical function tests are shown in Figures 2 and 3, respectively. Intent-to-treat analyses revealed that mean chair rise NRSpain scores increased for the CON and decreased for the CNCRT and ECCRT groups by month 4 (partial η2 = 0.044; Fig. 2A; P < 0.05, small effect). Mean stair climb NRSpain scores increased for the CON and decreased for the CNCRT and ECCRT groups by month 4 (partial η2 = 0.044; Fig. 2B, small effect). Intent-to-treat analysis revealed a significantly greater reduction in chair pain scores in the CNCRT compared with the CON and ECCRT groups (Supplementary Table 2, Supplemental Digital Content 3, http://links.lww.com/PHM/A999; P = 0.008).
Figures 3A–C provide the per-protocol NRSpain scores before, during, and after the 6-min walk test. By month 4, the CNCRT group demonstrated significant reductions in pain severity at minutes 2, 4, 5, 6, and at minute 1 of recovery that ranged from 1.1 to 1.4 points on the NRSpain scale (B; all P < 0.05). The ECCRT group averaged 0.1- to 0.71-point NRSpain reductions at these same time intervals. Intent-to-treat analyses revealed the same pain responses among the three groups at the same time points as the per-protocol analysis.
The StepWatch results from baseline to month 4 are presented in Table 4. Baseline differences existed in the daily steps per day, with the CNCRT group walking fewer steps than the remaining groups. No significant group × time interactions were found for average daily steps taken or the daily minutes of walking at high to low intensities using per-protocol or intent-to-treat approaches (partial η2 range = 0.002–0.033; small effect). We did find a consistent main effect of group for daily moderate-intensity walking time, where the CNCRT group accumulated fewer minutes of walking at moderate intensities during the day compared with the remaining two groups at both time points (Supplementary Table 3, Supplemental Digital Content 4, http://links.lww.com/PHM/A1000; P < 0.05).
The key findings of this study were that ECCRT did not produce greater changes in functional test scores, gait performance, or functional pain for stair climb or chair rise than CNCRT. However, CNCRT group experienced lower pain severity than the other two groups during the 6-min walk test after training. Thus, resistance training muscle action type did not seem to be a strong mechanism modulating knee pain change or consistent translation to better physical function outcomes tested here.
Comparative evidence of eccentric versus concentric muscle actions on physical function tasks and functional pain outcomes in knee OA is very limited. The existing data are discrepant, potentially because of heterogeneity among interventions. Some investigations that used concentric-focused leg exercise (including leg extension, leg press, hip adduction/abduction, lunge or straight leg raise with multiple sets)20,21 produced improvements in WOMAC pain subscores,16,20–22 and functional ability (walking velocity, stair climb).16,22 One study of home functional strengthening exercise (step-ups, squats, and ankle weight resistance for knee flexion-extension, hip flexion-extension), produced changes in WOMAC pain subscores and faster stair climb times and faster time to complete chair stands.23 Topp et al.16 subjected individuals to either dynamic or isometric leg exercise with Therabands. After 4 mos, Topp et al.16 reported that functional pain was reduced by 42%–58.5% with both exercise types and that pain during stair climb and descent was reduced by 28.2%–42.5% compared with controls. In another study, Foroughi et al.24 reported that 6 mos of RT on pneumatic machines reduced pain more than sham exercise (32.1% vs. 21.4%), but this group difference was not significant over time. We found large reductions in functional pain with both resistance training types but additional protection against pain during the 6-min walk test with CNCRT. These collective data provide support that regular strengthening, irrespective of muscle action type, can reduce pain with load-bearing movements.
Compared with other published regimens, our training did not emulate movements of daily living irrespective of muscle action type.16,23 Thus, leg strength gains achieved by ECCRT or CNCRT did not directly translate to gait performance. Sitting while exercising may not sufficiently challenge the neuromotor system to translate to gains in functional outcomes and pain. Investigators used a combination of seated machines (leg press), squats, and walking with dumbbells over stable and unstable surfaces.25 Our data show that ECCRT of CNCRT did not differentially change temporal spatial parameters of gait and gait variability. Machine-based resistance training, irrespective of muscle action type, may not facilitate improvements in motor exploration and reduction in variance during walking. Normal walking variability contains some random and periodic components,26 which may reflect pain avoidance strategies or pathology-related changes to the joint. Recent evidence suggests that inclusion of resistance exercise performed on a unstable surface (performed on foam pads and BoSUballs) may increase motor adaptability, motor output, and better responses to environmental conditions during walking.25 Our seated machine intervention was more similarly aligned with the pneumatic machines used by Foroughi et al.24 This investigation did not detect group by time interactions for pain or functional scores despite leg muscle strength gains. Potentially, higher speeds of more functional resistive movement at lower percentage of 1RM may be more functionally useful for the knee OA population.27 For example, after 8 wks of resistance training (sit-to-stand, squat, calf raises using vest weight starting at 20% 1RM), participants demonstrated significant improvements in timed-up-and-go, sit-to-stand repetitions accomplished in 30 secs and stair climb power than in controls.27
Community ambulation did not change over time among any group in this study, indicating that regular resistance training irrespective of muscle contraction type did not facilitate more walking during the day. Our findings have been corroborated by Brisson et al.,28 who recently showed that steps per day did not change over a 3-yr period among individuals with mild-to-moderate knee OA.28 Other strategies to increase daily step count are needed to keep this population moving throughout the day. These could involve inclusion of daily step cues via text messages and e-mails in digital exercise programs or physical activity trackers.
Limitations and Future Directions
This study consisted of a self-selected sample of volunteers, which could contribute selection bias; many of these participants were not candidates for joint replacement but enrolled because they felt that exercise could help with symptom management. Despite rigorous control and measurement, the generalizability of the findings to all knee OA may not be possible. We provided per-protocol and intent-to-treat analyses to reduce this bias. This study was previously powered to detect differences in subjective average resting pain not for all the variables in this analysis, which may explain the small training intervention effects observed in most of the functional variables. Importantly, there was considerable interindividual variation in responsiveness to the interventions, and eight of the control participants maintained or improved their strength. The 4-month changes in leg extension strength ranged from −1.6% loss to +140% in the CNCRT and from −59% to +47% in the ECCRT, which could contribute to variation to changes in functional performance. Thus, discrepant conclusions may exist based on participant responsiveness. We may have missed potential functional and pain benefits with ECCRT as this study did not include functional performance tasks that specifically stressed eccentric capacity of the knee. Inclusion of activities, such as stair descent, multiple chair rise, or lowering oneself to the floor, may be useful in future investigation to assess ECCRT translation to daily function. Future resistance training protocols should consider this stability element in the design. In addition, our participants with varying severity of knee OA my not have had substantial gait variability at the onset of the study.
Both resistance exercise types conferred benefit to functional pain during chair rise and stair climb. Concentrically focused resistance training reduced severity of ambulatory pain and persistence of pain upon walking cessation. However, the ultimate selection of which type to use is dependent on several factors including equipment availability, pain at the knee or other joints, and the comfort level of the patient. The typical concentrically focused program may improve comfort during walking. The most important message is that patients with knee OA should participate in resistance training to help decrease pain during physical activity. Future research should examine mechanisms of responsiveness to ECCRT and CNCRT and protocols that include load-bearing movements emphasizing different muscle actions.
1. McAlindon TE, Bannuru RR, Sullivan MC, et al.: OARSI guidelines for the non-surgical management of knee osteoarthritis. Osteoarthr Cartil
2. Jansen MJ, Viechtbauer W, Lenssen AF, et al.: Strength training alone, exercise therapy alone, and exercise therapy with passive manual mobilisation each reduce pain and disability in people with knee osteoarthritis: a systematic review. J Physiother
3. Tanaka R, Ozawa J, Kito N, et al.: Efficacy of strengthening or aerobic exercise on pain relief in people with knee osteoarthritis: a systematic review and meta-analysis of randomized controlled trials. Clin Rehabil
4. Vincent HK, Montero C, Conrad BP, et al.: “Functional pain,” functional outcomes, and quality of life after hyaluronic acid intra-articular injection for knee osteoarthritis. PM R
5. LaStayo P, Marcus R, Dibble L, et al.: Eccentric versus traditional resistance exercise for older adult fallers in the community: a randomized trial within a multi-component fall reduction program. BMC Geriatr
6. LaStayo PC, Marcus RL, Dibble LE, et al.: Eccentric exercise versus usual-care with older cancer survivors: the impact on muscle and mobility—an exploratory pilot study. BMC Geriatr
7. Li Y, Su Y, Chen S, et al.: The effects of resistance exercise in patients with knee osteoarthritis: a systematic review and meta-analysis. Clin Rehabil
8. Reeves ND, Maganaris CN, Longo S, et al.: Differential adaptations to eccentric versus conventional resistance training in older humans. Exp Physiol
9. Gür H, Cakin N, Akova B, et al.: Concentric versus combined concentric-eccentric isokinetic training: effects on functional capacity and symptoms in patients with osteoarthrosis of the knee. Arch Phys Med Rehabil
10. Katsura Y, Takeda N, Hara T, et al.: Comparison between eccentric and concentric resistance exercise training without equipment for changes in muscle strength and functional fitness of older adults. Eur J Appl Physiol
11. Vincent KR, Vasilopoulos T, Montero C, et al.: Eccentric and concentric resistance exercise comparison for knee osteoarthritis. Med Sci Sports Exerc
12. Ornetti P, Dougados M, Paternotte S, et al.: Validation of a numerical rating scale to assess functional impairment in hip and knee osteoarthritis: comparison with the WOMAC function scale. Ann Rheum Dis
13. Dobson F, Hinman RS, Roos EM, et al.: OARSI recommended performance-based tests to assess physical function
in people diagnosed with hip or knee osteoarthritis. Osteoarthr Cartil
14. Lin YC, Davey RC, Cochrane T: Tests for physical function
of the elderly with knee and hip osteoarthritis. Scand J Med Sci Sports
15. Kiss R: Variability of gait characterized by normalized deviation. Acta Bioeng Biomech
16. Topp R, Woolley S, Hornyak J, et al.: The effect of dynamic versus isometric resistance training on pain and functioning among adults with osteoarthritis of the knee. Arch Phys Med Rehabil
17. Mudge S, Taylor D, Chang O, et al.: Test-retest reliability of the StepWatch activity monitor outputs in healthy adults. J Phys Act Health
18. Gupta SK: Intention-to-treat concept: a review. Perspect Clin Res
19. Sainani KL: Making sense of intention-to-treat. PM R
20. DeVita P, Aaboe J, Bartholdy C, et al.: Quadriceps-strengthening exercise and quadriceps and knee biomechanics during walking in knee osteoarthritis: a two-centre randomized controlled trial. Clin Biomech Bristol Avon
21. Hall M, Hinman RS, Wrigley TV, et al.: Knee extensor strength gains mediate symptom improvement in knee osteoarthritis: secondary analysis of a randomised controlled trial. Osteoarthr Cartil
22. Jorge RTB, Souza MC, Chiari A, et al.: Progressive resistance exercise in women with osteoarthritis of the knee: a randomized controlled trial. Clin Rehabil
23. Baker KR, Nelson ME, Felson DT, et al.: The efficacy of home based progressive strength training in older adults with knee osteoarthritis: a randomized controlled trial. J Rheumatol
24. Foroughi N, Smith RM, Lange AK, et al.: Lower limb muscle strengthening does not change frontal plane moments in women with knee osteoarthritis: a randomized controlled trial. Clin Biomech Bristol Avon
25. Eckardt N, Rosenblatt NJ: Instability resistance training decreases motor noise during challenging walking tasks in older adults: a 10-week double-blinded RCT. Front Aging Neurosci
26. Alkjaer T, Raffalt PC, Dalsgaard H, et al.: Gait variability and motor control in people with knee osteoarthritis. Gait Posture
27. Pazit L, Jeremy D, Nancy B, et al.: Safety and feasibility of high speed resistance training with and without balance exercises for knee osteoarthritis: a pilot randomised controlled trial. Phys Ther Sport Off J Assoc Chart Physiother Sports Med
28. Brisson NM, Gatti AA, Maly MR: Pain is not associated with steps per day in persons with mild-to-moderate, symptomatic knee osteoarthritis - a mixed models analysis of multiple measurements over 3 years. Arthritis Care Res