Ottawa Panel Evidence-Based Clinical Practice Guidelines for Therapeutic Exercises and Manual Therapy in the Management of Osteoarthritis

Target Population

Studies of adult patients (>18 years of age) with classical or definite OA as defined by Klippel et al³⁸ were included in our literature search. Patients with OA that affected peripheral joints were eligible to participate. Patients at different stages of the disease participated in the included clinical trials; some trials involved patients with both chronic and acute conditions. All stages of the disease were included in our analysis. The recommendations state the disease stage for which the intervention is most appropriate. If, however, the trial on which the recommendation was based did not mention disease stage, neither does our recommendation (Appendix 2). Most trials involved patients with chronic OA (>12 year' duration).

Various exclusion criteria were established:

• Studies of patients with OA involving spinal problems (excluded due to the numerous associated signs and symptoms and because the Philadelphia Panel guidelines for low back pain³⁹ and neck pain⁴⁰ were recently developed by the same methodologists);

• Studies of patients who recently had surgery;

• Patients with other rheumatologic or musculoskeletal problems (eg, fractures, tendinitis, or bursitis), clinically important medical problems, or psychiatric conditions that could hamper rehabilitation or reduce functional status;

• Studies of subjects without known pathology or impairments; and

• Studies of subjects with mixed arthritic conditions such as the sample in a study by D'Lima et al.⁴¹

Table 1 lists the complete inclusion and exclusion criteria.

Study Inclusion/Exclusion Criteria

Generally, comparisons of 2 active interventions (head-to-head studies) were excluded for the same reasons explained in the previous publication on the Ottawa Panel EBCPGs on RA.¹⁶ Examples of head-to-head studies include dynamic exercises versus isometric exercises,⁴² individual versus group exercises,⁴³ home exercises versus aquatics,⁴⁴ walking versus patient education,⁴⁵ sham electrical stimulation versus patient education combined with TE,⁴⁶ aerobics (walking) versus strengthening exercises,^45,47 and walking versus jogging in water.⁴⁸ Some studies had several comparative groups, and only some of the group comparisons were eligible to be included.

Other excluded interventions comprised surgery, drug, or psychosocial (nonphysical) interventions. For instance, the RCTs on exercises after a total hip replacement for severe hip OA were excluded; RCTs with frequent use of continuous passive motion (CPM) following a total knee arthroplasty for severe knee OA^49–61 also were excluded. However, practitioners can refer to a recent meta-analysis on the efficacy of CPM combined with physical therapy versus physical therapy alone (n=799) following a total knee arthroplasty for knee OA⁶² to find further recommendations on these postsurgery interventions (grade A for flexion deformity and time to achieve 90 degrees of flexion and grade C+ for active knee flexion range of motion [ROM], pain related to analgesic use, and number of patients needing postoperative manual therapy). Postsurgery intervention studies usually allowed samples with varying proportions of patients with OA and RA. Most of the RCTs on efficacy of postsurgery interventions such as CPM recruited subjects with mixed arthritic conditions, which is the reason they are excluded in this article.

Subjects who received placebo, were untreated, or received routine conventional therapeutic approaches were acceptable control groups. If concurrent interventions (eg, electroanalgesia and medication) were provided to the experimental and control groups, these interventions were included. However, interventions where the patient acts as his or her own control were not included. A priori, we did not include or exclude studies based on the quality of their methods. However, we did consider quality when grading our recommendations.

The categories of interventions selected were approved by the Ottawa Panel according to the study's description of the intervention. Category selection also was influenced by previous work performed by the Ottawa Methods Group¹⁵ and by the Ottawa Panel on TE for patients with RA.¹⁶

Therapeutic Exercises

EBCPGs Related to Strengthening Exercises

Lower-extremity (LE) strengthening versus control, level 1 (3 RCTs, n=103): grade A for pain getting up and down from floor and functional status (clinically important benefit); grade C+ for pain during walking, pain while climbing and descending stairs, arthritis activity, functional tasks, and quadriceps femoris muscle peak torque (clinically important benefit); grade C for stiffness, mobility, quadriceps femoris muscle force, muscle activation, and quality of life (no benefit). Patients with a diagnosis of OA of the knee.

Lower-extremity isometric strengthening versus control, level 1 (1 RCT, n=67): grade A for pain getting down to and up from floor (clinically important benefit); grade C+ for pain getting down and up stairs and timed functional tasks (clinical benefit); grade C for stiffness and functional status (no benefit). Patients with a diagnosis of OA of the knee.

Isotonic resistance training versus isotonic combined with isokinetic (Kinetron*) resistance training for quadriceps femoris and hamstring muscles, level 1 (1 RCT, n=15): grade C for quadriceps femoris muscle peak torque (no benefit). Patients with a primary diagnosis of OA of the knee.

Isotonic combined with isokinetic (Kinetron*) resistance training for quadriceps femoris and hamstring muscles versus control, level 1 (1 RCT, n=18): grade C for muscle force (no benefit). Patients with primary diagnosis of OA of the knee.

Eccentric resistance training (Cybex*) for quadriceps femoris and hamstring muscles versus control, level 1 (1 RCT, n=17): grade C for muscle force (no benefit). Patients with primary diagnosis of OA of the knee.

Concentric resistance training for quadriceps femoris and hamstring muscles versus control, level 1 (1 RCT, n=15): grade A for pain at rest and during activities (clinically important benefit); grade C for global functional status (no benefit). Patients with knee OA bilaterally and grade II or III OA.

Concentric-eccentric resistance training for quadriceps femoris and hamstring muscles versus control, level 1 (1 RCT, n=14): grade A for pain at rest and during specific functional activities: 15-m walk and stair climbing/descending time (clinically important benefit). Patients with knee OA bilaterally and grade II or III OA.

Home strengthening program for quadriceps femoris muscle versus control, level 1 (1 controlled clinical trial [CCT], n=53): grade A for pain, functional status, energy level, and ROM in flexion (clinically important benefit); grade C for physical mobility, muscle force, swelling, and exercise (no benefit). Patients with OA of the knee.

General LE exercise program (including muscle force resistance, flexibility, and mobility/coordination) versus control, level 1 (8 RCTs, n=876): grade A for pain at night and ability on stairs (clinically important benefit); grade C for knee flexion ROM, muscle force, knee joint position sense, kinesthesia, stance, gait, functional status, quality of life, muscle activation, stiffness, and physical activity (no benefit). Patients with a diagnosis of OA.

Progression versus no-progression LE strengthening exercises, level 1 (1 RCT, n=179): grade A for pain at rest and ROM (clinically important benefit); grade C for stiffness and functional status (no benefit). Patients with radiographic evidence of OA in the tibiofemoral compartment.

Hand strengthening versus control, level 1 (1 RCT, n=40): grade A for pain and grip force (clinically important benefit). Patients who met the American College of Rheumatology criteria for hand OA.¹⁵⁴

EBCPGs Related to General Physical Activity, Including Fitness and Aerobic Exercises

Whole-body functional exercise versus control, level 1 (4 RCTs, n=864): grade A for pain and functional status (mobility, walking, work, disability in activities of daily living [ADL]) (clinically important benefit); grade C for knee flexion ROM, quadriceps femoris muscle force, hamstring muscle force, gait, stair climbing time, climbing self-efficacy, and quality of life (no benefit). Patients with OA of the knee.

Walking program versus control, level 1 (6 RCTs, n=711): grade A for pain, functional status, stride length, disability transferring from bed, disability bathing, aerobic capacity, exercise endurance, energy level, physical activity, and sleep (clinically important benefit); grade C+ for disability in ADL (clinical benefit); grade C for walking speed, disability toileting, disability dressing and stairs, morning stiffness, and quality of life (no benefit). Patients with OA.

Jogging in water versus control, level 1 (1 RCT, n=79): grade A for physical activity (clinically important benefit); grade C for walking time, morning stiffness, pain, grip force, trunk ROM, functional status, and exercise endurance (no benefit). Patients with current symptoms of chronic pain and stiffness in involved weight-bearing joints.

Water exercises versus control, level 1 (1 RCT, n=30): grade C for hip and shoulder abduction torque and ROM (no benefit). Patients with OA or RA diagnosed by a rheumatologist or an orthopedic physician.

Yoga versus control, level 1 (1 RCT, n=30): grade A for pain during activity and ROM (clinically important benefit); grade C for tenderness, grip force, swelling, and hand function (no benefit). Patients with OA of the distal interphalangeal or proximal interphalangeal joints of the fingers.

EBCPGs Related to the Combination of Exercises

Manual therapy combined with exercise versus control, level 1 (1 RCT, n=83): grade A for pain (clinically important benefit); grade C for 6-minute walk distance (no benefit). Patients with a diagnosis of OA.

Summary of Trials

Twenty-nine trials (n=2,486 patients) evaluated different types of TE for managing OA-affected joints of the upper extremities and LEs. Most of the trials compared these exercises with a control, but the trials examined different kinds of TE. The strengthening exercise trials were as follows: LE strengthening (n=345),^42,70,71,79 LE isometric strengthening (n=102),⁴² isotonic resistance training versus isotonic combined with isokinetic (Kinetron) resistance training for the knee (n=32),⁷⁰ isotonic combined with isokinetic (Kinetron) resistance training for the knee (n=32),⁷⁰ eccentric resistance training (Cybex) for the knee (n=32),⁷⁰ concentric resistance training for the knee (n=23),⁶⁷ concentric-eccentric resistance training for the knee (n=23),⁶⁷ home program strengthening for the knee (n=81),⁴⁷ general LE exercise program (including muscle force, flexibility, and mobility/coordination) (n=490),^{64,65,68,77,78,82,83} progression in LE strengthening exercises versus no progression (n=179),⁷⁵ and home program hand strengthening (n=40).⁸⁰

Several RCTs examined general physical activities, including fitness and aerobic exercises, such as whole-body functional exercises (n=864),^{45,63,72,73,76} walking program (n=1,089),^{45,47,48,69,73,74,76} jogging in water (n=115),⁴⁸ water exercise (n=30),⁸¹ and yoga (n=30).⁶⁶ One trial was related to the combination of manual therapy and exercises (n=83).³⁰

Twenty-three included trials were RCTs^{42,45,47,48,63–67,69–76,78,80–83} and 3 trials were CCTs^47,68,77 (Appendix 2). We used the Jadad scale to decide whether a study was an RCT or a CCT.¹⁵

The trials examined 2 basic types of exercises. The first type involved strengthening exercises, such as resistance isometric, stretching, eccentric, and concentric exercises; these exercises were specific to different muscles. The other type focused on whole-body functional strengthening programs and included aerobic conditioning and general fitness. Program duration, treatment schedule for exercise intervention, and length of exercise session varied from 4 weeks⁶⁴ to 18 months^45,73,76 for program duration, from once a week⁶⁶ to 10 times a day⁸⁰ (depending on the phase of the program) for treatment schedule, and from 5 minutes to longer per exercise session (length of exercise session increased depending on tolerance)⁷⁴ (Appendix 2).

Strengthening Exercises

Lower-extremity strengthening versus control (4 RCTs, n=345),^42,70,71,79 showed clinical benefits for pain during walking, pain ascending and descending the stairs, quadriceps femoris muscle peak torque, and timed functional tasks (Tab. 3). Statistically significant differences were found for pain (Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC]; Fig. 1a), pain while getting up from the floor (weighted mean differ-ence [WMD]=−2.36, 95% confidence interval [CI]=−4.22 to −0.50; Fig. 1a), and functional status (Tab. 3). However, other outcomes were not statistically significant. Outcomes were measured at the end of intervention (4 months for Topp et al⁴² and 8 weeks for Kreindler et al⁷⁰ and Schilke et al⁷⁹) or at follow-up (6 weeks for Kreindler et al⁷⁰) (Figs. 1a–f).

For LE isometric strengthening versus control (1 RCT, n=102),⁴² clinical benefits were found for pain getting down and up from the floor, pain while going up and down stairs, and timed functional tasks but not for stiffness and functional status (Tab. 4). Statistically significant differences were found for pain while getting down to the floor (WMD=−2.05, 95% CI=−3.62 to −0.48) and up from the floor (WMD=−2.14, 95% CI=−4.01 to −0.27). Stiffness, functional limitation, pain, time to get down to the floor and to get up, and time to go up and down the stairs were not considered to be clinically important benefits at 4 months (Figs. 2a–d).

One trial (RCT, n=32)⁷⁰ showed no statistically significant difference or clinically important benefit for quadriceps femoris muscle peak torque in patients with OA either at the end of a 6-week intervention or at a 6-week follow-up. This trial compared isokinetic resistance training versus isotonic and isokinetic (Kinetron) resistance training for the knee (Fig. 3), resistance training and Kinetron versus control (Fig. 4), and eccentric resistance training (Cybex) for the knee versus control (Fig. 5).

Statistically significant differences favored concentric exercises over control (1 RCT, n=23)⁶⁷ for pain (WMD=−17.7, 95% CI=−22.79 to −12.61) and functional status (WMD=−10.85, 95% CI=−21.34 to −0.36) at 8 weeks (Figs. 6 and 7). A clinically important benefit was observed for pain (Tab. 5) but not for functional status.

For concentric-eccentric versus control (1 RCT, n=23),⁶⁷ clinically important benefits and statistically significant differences were observed for pain (WMD=−11.40, 95% CI=−17.95 to −4.85) (Tab. 6, Fig. 8) and functional status (WMD=−12.15, 95% CI=−22.67 to −1.63) (Tab. 6) at 8 weeks. Results for 15-m walk, stair-climbing time, and stair-descending time also were significant, as were results for pain at night, pain sitting, pain rising from a chair, pain standing, and pain climbing stairs (Tab. 6).

One CCT examined home program strengthening for knee versus control (n=81)⁴⁷ and showed clinically important benefits for pain, functional status, energy level (Tab. 7), and ROM (results not shown) but not for physical mobility (Tab. 7). Statistically significant data were found for WOMAC–pain (WMD=3.00, 95% CI=1.58 to 4.42), visual analog scale (VAS)–pain (WMD=3.30, 95% CI=2.62 to 3.98), WOMAC physical function index (WMD=9.90, 95% CI=8.08 to 11.72), Nottingham Health Profile (NHP)–pain (WMD=10.60, 95% CI=8.90 to 12.30), NHP–energy (WMD=15.90, 95% CI=14.87 to 16.93), NHP–physical mobility (WMD=7.10, 95% CI=4.14 to 10.06), NHP–sleep (WMD=3.40, 95% CI=0.89 to 5.91) (Figs. 9a–c, all at follow-up of 6 months), swelling (WMD=12.5, 95% CI=5.51 to 19.49), and ROM (WMD=19.5°, 95% CI=5.69° to 33.31°) (results not shown). However, no statistically significant difference was observed for muscle force or exercise tolerance (results not shown). No clinically important effects were found for muscle force, swelling, or exercise tolerance.

Several trials examined general LE exercise programs (including muscle force, flexibility, and mobility/coordination) versus control (7 RCTs, n=690).^{64,65,68,77,78,82,83} Important benefits were demonstrated for pain intensity and ability to step down but not for ROM, muscle force, gait, functional status, quality of life (Tabs. 8 and 9), knee joint position at 6 weeks (Fig. 10f), or muscle activation at 6 weeks (Fig. 10b). Statistically significant differences were found for the following outcomes:

• ROM in knee flexion, most affected knee at 10–12 weeks (WMD=10.00°, 95% CI=5.91° to 14.09°) (Fig. 10a);

• ROM in knee flexion, least affected knee at 10–12 weeks (WMD=10.00°, 95% CI=7.75° to 12.25°) (Fig. 10a);

• ROM in knee flexion, least affected knee at 12-month follow-up (WMD=12.00°, 95% CI=7.06° to 16.94°) (results not shown);

• isometric quadriceps femoris muscle force at 6 weeks (WMD=73 N, 95% CI=25.75 to 120.25 N) (Fig. 10b);

• quadriceps femoris muscle voluntary activation at 6 weeks (WMD=14.0%, 95% CI=5.87% to 22.13%) (Fig. 10b);

• aggregate functional performance time at 6 weeks (WMD=−8.47, 95% CI=−16.79 to −0.15) (Fig. 10d);

• functional status at 6 weeks (WMD=−3.50, 95% CI=−4.94 to −2.06) (Fig. 10d);

• mean change in physical function score at 3-month follow-up (WMD=−3.54, 95% CI=−6.04 to −1.04) (Fig. 10d);

• mean change in WOMAC-pain at 8 weeks (WMD=−12.10, 95% CI=−14.24 to −9.96) (Fig. 10e);

• mean change in pain at 10 to 12 weeks (WMD=−17.10, 95% CI=−29.99 to −4.21) (Fig. 10e);

• mean change in global pain score at 6-month follow-up (WMD=−1.87, 95% CI=−2.76 to −0.98) (Fig. 10e);

• mean change in pain (VAS), walking at 10 to 12 weeks (WMD=−7.07, 95% CI=−13.90 to −0.24) (Fig. 10e);

• mean change in pain (VAS), stairs (WMD=−10.42, 95% CI=−18.58 to −2.26) (results not shown);

• pain at night at 12-month follow-up (WMD=−4.00, 95% CI=−5.94 to −2.06) (Fig. 10e);

• pain at rest at 10 to 12 weeks (WMD=−1.50, 95% CI=−2.80 to −0.20) (Fig. 10e);

• pain on weight bearing at 10 to 12 weeks (WMD=−2.00, 95% CI=−3.02 to −0.98) (Fig. 10e);

• mean change in isometric knee extensor muscle force at 8 weeks (WMD=13.20 N, 95% CI=11.96 to 14.44 N) (Fig. 10g);

• mean change in isometric knee flexor muscle force at 8 weeks (WMD=9.00 N, 95% CI=8.04 to 9.96 N) (Fig. 10g);

• mean change in fast speed at 8 weeks (WMD=6.70 cm/s, 95% CI=6.34 to 7.06 cm/s) (Fig. 10h);

• mean change in fast cadence at 8 weeks (WMD=1.60 steps/min, 95% CI=1.40 to 1.80 steps/min) (Fig. 10h);

• mean change in fast stride length at 8 weeks (WMD=4.30 cm, 95% CI=3.99 to 4.61 cm) (Fig. 10h);

• quality of life measured with Medical Outcomes Study 36-Item Short-Form Health Survey questionnaire (SF–36) at 8 weeks (WMD=3.10, 95% CI=2.76 to 3.44) (Fig. 10i);

• mean change in WOMAC-function at 8 weeks (WMD=−7.80, 95% CI=−8.48 to −7.12) (Fig. 10k);

• improvement in self-reported disability at 24-week follow-up (WMD=−1.10, 95% CI=−1.91 to −0.29) (Fig. 10k);

• mean change in left quadriceps femoris muscle force at 6-week follow-up (WMD=10.86%, 95% CI=3.15% to 18.57%) (results not shown); and

• mean change in SF-36-physical function at 8 weeks (results not shown).

No statistically significant data were found for the remaining outcomes: ROM in knee extension and flexion (Fig. 10a); improvement in hip and knee ROM at 24-week follow-up (Fig. 10a); improvement in knee or hip muscle force at 24-week follow-up (Fig. 10c); Health Status Survey (HSS) score (Fig. 10d); pain, pain during walking, and pain at night at 10 to 12 weeks (Fig. 10e); knee joint position sense at 6 weeks (Fig. 10f); peak torque of the knee extensors and flexors at 10 to 12 weeks (Fig. 10g); step frequency and stride length at 10 to 12 weeks (Fig. 10h); stance at 10 to 12 weeks or at 12-month follow-up for most affected and least affected LEs (Fig. 10j); walking speed and stair-climbing time at 10 to 12 weeks or at 12-month follow-up (Fig. 10l); Algofunctional Index–pain at 10 to 12 weeks or at 12-month follow-up (Fig. 10m); improvement in physical activity at 10 to 12 weeks (Fig. 10n); and ability to step down (Fig. 10o).

For progression versus no-progression LE strengthening exercises (one RCT, n=179),⁷⁵ clinical benefits (Tab. 10) and statistically significant differences were found for ROM in knee flexion (WMD=13°, 95% CI=11.55° to 14.45°) and pain at rest (WMD=−23 mm, 95% CI=−24.03 to −21.97). No important differences were found for WOMAC–stiffness, WOMAC–pain, or pain after walk test. Outcomes were measured after 8 weeks (Figs. 11a–c).

Hand strengthening versus control (one RCT, n=40)⁸⁰ showed clinically important benefits for pain and grip force at 3 months (Tabs. 11 and 12, Figs. 12a–b). Statistically significant differences were found for pain (WMD=7.43, 95% CI=1.78 to 31.04) and change in grip force in the right hand (WMD=0.11, 95% CI=0.09 to 0.13) and the left hand (WMD=0.10, 95% CI=0.09 to 0.11) (Fig. 12b).

General Physical Activities, Including Fitness and Aerobic Exercises

For whole-body functional exercise versus control (5 RCTs, n=864),^{45,63,72,73,76} clinically important benefits were found for pain and functional status (mobility, walking, work, and disability on ADL). Statistically significant differences were found for numerous outcomes:

• pain frequency in transfer at 9 months (WMD=0.88, 95% CI=0.49 to 1.27) (Fig. 13a);

• pain intensity in transfer at 3 months (WMD=−0.94, 95% CI=−1.33 to −0.55), at 9 months (WMD=−0.46, 95% CI=−0.84 to −0.08), and at 18 months (WMD=−0.37, 95% CI=−0.70 to −0.04) (Fig. 13a);

• pain (WMD=−0.80, 95% CI=−1.29 to −0.31) (Fig. 13b);

• functional status measured with the Arthritis Impact Measurement Scales (AIMS) (WMD=5.49, 95% CI=3.92 to 7.06) (results not shown);

• functional status measured with the Arthritis Impact Measurement Scales 2 (AIMS2): arthritis pain (WMD=−0.85, 95% CI=−1.52 to −0.18) (Fig. 13b);

• functional status measured with AIMS2: mobility level at 12 weeks (WMD=−0.50, 95% CI=−0.93 to −0.07) favoring the control group (Fig. 13c);

• functional status measured with AIMS2: walking and bending at 12 weeks (WMD=−1.25, 95% CI=−2.08 to −0.42) (Fig. 13c);

• functional status measured with AIMS2: level of tension at 12 weeks (WMD=2.58, 95% CI=1.88 to 3.28) (Fig. 13c);

• hamstring muscle and low back flexibility at 12 weeks (WMD=3.63 in, 95% CI=2.04 to 5.22 in) (Fig. 13d);

• 5-minute walk test at 12 weeks (WMD=42.19 m, 95% CI=14.19 to 70.19 m) (Fig. 13e);

• hamstring muscle isometric torque at 30 degrees: most affected LE at 12 weeks (WMD=8.85 N·m, 95% CI=1.91 to 15.79 N·m) (Fig. 13f);

• hamstring muscle isometric torque at 30 degrees: least affected LE at 12 weeks (WMD=10.51 N·m, 95% CI=3.24 to 17.78 N·m) (Fig. 13f);

• quadriceps femoris muscle isometric torque at 60 degrees: most affected LE at 12 weeks (WMD=19.80 N·m, 95% CI=4.75 to 34.85 N·m) (Fig. 13f);

• quadriceps femoris muscle isometric torque at 60 degrees: least affected LE at 12 weeks (WMD=17.03 N·m, 95% CI=1.08 to 32.98 N·m) (Fig. 13f);

• hamstring muscle isometric torque at 60 degrees: most affected LE at 12 weeks (WMD=8.02 N·m, 95% CI=0.88 to 15.16 N·m) (Fig. 13f);

• hamstring muscle isometric torque at 60 degrees: least affected LE at 12 weeks (WMD=10.99 N·m, 95% CI=3.68 to 18.30 N·m) (Fig. 13g);

• hamstring muscle isokinetic torque at 30°/s: most affected LE at 12 weeks (WMD=10.98 N·m, 95% CI=0.38 to 21.58 N·m) (Fig. 13g);

• hamstring muscle isokinetic torque at 30°/s: least affected LE at 12 weeks (WMD=10.51 N·m, 95% CI=3.24 to 17.78 N·m) (Fig. 13g);

• hamstring muscle isokinetic torque at 90°/s: most affected LE at 12 weeks (WMD=9.73 N·m, 95% CI=−1.40 to 20.86 N·m) (Fig. 13g);

• cadence at 3 months (WMD=0.87 steps/min, 95% CI=0.33 to 1.41 steps/min) (Fig. 13h), at 9 months (WMD=0.94 steps/min, 95% CI=0.37 to 1.51 steps/min), and at 18 months (WMD=2.08 steps/min, 95% CI=1.56 to 2.60 steps/min) (results not shown);

• stride length at 3 months (WMD=2.17 cm, 95% CI=1.18 to 3.16 cm), at 9 months (WMD=2.84 cm, 95% CI=1.77 to 3.91 cm), and at 18 months (WMD=6.49 cm, 95% CI=5.49 to 7.49 cm) (Fig. 13i);

• walking speed at 3 months (WMD=3.77 cm/s, 95% CI=2.60 to 4.94 cm/s), at 9 months (WMD=4.37 cm/s, 95% CI=3.12 to 5.62 cm/s), and at 18 months (WMD=7.79 cm/s, 95% CI=6.60 to 8.98 cm/s) (Fig. 13j);

• stance time at 3 months (WMD=−0.01 s, 95% CI=−0.01 to −0.01 s) and at 18 months (WMD=−0.02 s, 95% CI=−0.02 to −0.02 s) (Fig. 13k);

• percentage of swing at 3 months (WMD=0.39, 95% CI=0.18 to 0.60), at 9 months (WMD=−0.36, 95% CI=−0.59 to −0.13), and 18 months (WMD=0.54, 95% CI=0.32 to 0.76) (Fig. 13k);

• stair-climbing time at 18 months (WMD=−1.92 s, 95% CI=−2.01 to −1.83 s) (Fig. 13l);

• climbing self-efficacy score at 18 months (WMD=9.32, 95% CI=8.86 to 9.78) (Fig. 13l);

• quality of life (WMD=3.10, 95% CI=2.97 to 3.23) (results not shown); and

• disability in bathing (WMD=0.41, 95% CI=0.18 to 0.91) (results not shown).

No clinically important benefits were found for quadriceps femoris and hamstring muscle force at 12 weeks (Figs. 13a–b), knee flexor ROM, gait, or quality of life (results not shown). No statistical data were found for pain intensity and frequency in ambulation at 3, 9, and 18 months (Fig. 13a); pain frequency in transfer at 9 and 18 months (Fig. 13a); AIMS2 hand and finger functional status, arm functional status, self-care tasks, household tasks, social activity, support from friends, work, or mood at 12 weeks (Fig. 13c); quadriceps femoris muscle isometric and isokinetic torque at 30 degrees (Figs. 13f–g), quadriceps femoris muscle isokinetic torque at 90 degrees (Fig. 13g), or hamstring muscle isokinetic torque at 90 degrees (Fig. 13g), all at 12 weeks; stance time at 9 months (Fig. 13k); or incidence of disability in ADL, disability in transferring from a bed to a chair, disability in toileting, disability in dressing and eating, or quality of life measured by HSS score (results not shown).

Six RCTs and 1 CCT examined walking versus control (n=1,089),^{45,47,48,69,73,74,76} and trials discovered clinical benefits for pain, functional status, stride length, disability transferring from bed, disability bathing, disability in ADL, energy level, medication use, aerobic capacity, and quality of life (Tabs. 13 and 14). No clinical benefits were found for walking speed (Tab. 13), pain in ambulation (results not shown), disability toileting, or disability dressing (Fig. 13e, both at 18-month follow-up). Statistically significant results were shown for the following outcomes:

• pain frequency in ambulation at 3 months (WMD=−0.56, 95% CI=−1.07 to −0.05) and in transfer (WMD=−0.42, 95% CI=−0.77 to −0.07) (results not shown);

• pain intensity in transfer at 3 months (WMD=−0.55, 95% CI=−1.02 to −0.08), at 9 months (WMD=−0.46, 95% CI=−0.84 to −0.08), and at 18 months (WMD=−0.41, 95% CI=−0.76 to −0.06) (results not shown);

• NHP-physical mobility, –pain, –energy, and –sleep (Tab. 13);

• WOMAC-physical function and –pain (Tab. 13);

• VAS-pain (Tab. 13);

• walking speed at 3 months (WMD=3.69 cm/s, 95% CI=2.47 to 4.91 cm/s), at 9 months (WMD=10.29 cm/s, 95% CI=9.05 to 11.53 cm/s), and at 18 months (WMD=10.29 cm/s, 95% CI=9.07 to 11.51 cm/s) (Tab. 13);

• cadence at 3 months (WMD=3.56 steps/min, 95% CI=3.00 to 4.12 steps/min), at 9 months (WMD=3.69 steps/min, 95% CI=3.12 to 4.26 steps/min), and at 18 months (WMD=3.77 steps/min, 95% CI=3.21 to 4.33 steps/min) (Tab. 13);

• stance time at 3 months (WMD=−0.04 s, 95% CI=0.04 to −0.04 s), at 9 months (WMD=−0.03 s, 95% CI=−0.03 to −0.03 s), and at 18 months (WMD=−0.03 s, 95% CI=−0.03 to −0.03 s) (results not shown);

• percentage of swing at 3 months (WMD=0.86, 95% CI=0.64 to 1.08) and at 18 months (WMD=0.54, 95% CI=0.32 to 0.76) (Tab. 13);

• stride length at 9 months (WMD=7.53 cm, 95% CI=6.48 to 8.58 cm) and at 18 months (WMD=7.54 cm, 95% CI=6.52 to 8.56 cm) (Tab. 13);

• climbing self-efficacy score at 18-month follow-up (WMD=8.00, 95% CI=7.55 to 8.45) (Fig. 14a);

• 6-minute walk test at 8 weeks (WMD=111.50 m, 95% CI=76.24 to 146.77 m) (Fig. 14b);

• stair-climbing time at 18-month follow-up (WMD=−1.41, 95% CI=−1.51 to −1.31) (Fig. 14c);

• general health status at 18-month follow-up (WMD=3.59, 95% CI=3.46 to 3.72) (Fig. 14d);

• incidence of disability at 18-month follow-up (WMD=0.52, 95% CI=0.28 to 0.96) (Fig. 14e);

• disability in transferring from a bed to a chair at 18-month follow-up (WMD=0.42, 95% CI=0.22 to 0.79) (Fig. 14e);

• disability in bathing (WMD=0.38, 95% CI=0.17 to 0.84) (Tab. 14) and in dressing at 18-month follow-up (WMD=0.28, 95% CI=0.10 to 0.83) (Fig. 14e);

• fast speed at 8 weeks (WMD=15.00 m/min, 95% CI=6.53 to 23.47 m/min) (Fig. 14f);

• fast stride at 8 weeks (WMD=0.20 m, 95% CI=0.03 to 0.37 m) (Fig. 14f);

• free speed at 8 weeks (WMD=7.00 m/min, 95% CI=0.34 to 13.66 m/min) (Fig. 14g);

• free stride at 8 weeks (WMD=0.20 m, 95% CI=0.08 to 0.32 m) (Fig. 14g);

• AIMS-pain at 8 weeks (WMD=−1.00, 95% CI=−1.79 to −0.21) (Fig. 14i);

• AIMS-physical activity at 8 weeks (WMD=−2.22, 95% CI=−3.25 to −1.19) (Tab. 13) and at 12 weeks (WMD=−1.30, 95% CI=−2.48 to −0.12) (Tab. 14, Fig. 14j);

• AIMS-medication use (WMD=2.74, 95% CI=1.93 to 3.55) (Tab. 13);

• 15.2-m (50-ft) walking time at 8 weeks (WMD=−2.00 s, 95% CI=−2.97 to −1.03 s) and at 12 weeks (WMD=−0.90 s, 95% CI=−1.71 to −0.09 s) (Fig. 14p for 12 weeks only);

• exercise heart rate at 12 weeks (WMD=16.00 bpm, 95% CI=3.94 to 28.06 bpm) (Fig. 14s);

• aerobic capacity at 12 weeks (WMD=5.10 mL/kg min⁻¹, 95% CI=2.88 to 7.32 mL/kg min⁻¹) (Fig. 14s); and

• exercise endurance at 12 weeks (WMD=3.30 min, 95% CI=1.12 to 5.48 min) (Fig. 14).

No statistically significant data were found for the following: pain intensity in ambulation (results not shown), pain frequency in ambulation at 9 and 18 months (results not shown), pain frequency in transfer at 9 and 18 months (results not shown), stride length (Tab. 13), disability in toileting and in eating at 18-month follow-up (Fig. 14e), fast cadence at 8 weeks (Fig. 14f), free cadence at 8 weeks (Fig. 14g), NHP-physical mobility and WOMAC-physical function at 6-month follow-up (Fig. 14h), AIMS-pain at 12 weeks and 9-month follow-up (Tab. 13, Fig. 14i), AIMS-physical activity at 9-month follow-up (Fig. 14j), AIMS-arthritis impact at 8 weeks (Fig. 14k), grip force at 12 weeks and 9-month follow-up (Fig. 14m), NHP-pain at 6-week follow-up (Fig. 14n), morning stiffness at 12 weeks and 9-month follow-up (Fig. 14o), 15.2-m (50-ft) walking time at 9-month follow-up (Fig. 14p), resting systolic blood pressure and resting diastolic blood pressure at 12 weeks and 9-month follow-up and exercise heart rate at 9-month follow-up (Fig. 14q), trunk flexibility at 12 weeks (Fig. 14r), and aerobic capacity and exercise endurance at 9-month follow-up (Fig. 14s).

Physical activity and aerobic capacity yielded clinically important benefits favoring jogging in water versus control (one RCT, n=115)⁴⁸ (Tab. 15). However, no clinical benefits were shown for functional status (AIMS-physical activity) at 12 weeks (Tab. 15), pain at 12 weeks and 9-month follow-up (results not shown), morning stiffness at 12 weeks and 9-month follow-up (results not shown), trunk ROM at 12 weeks and 9-month follow-up (results not shown), exercise heart rate (Tab. 15), or exercise endurance at 12 weeks (Tab. 15). Statistically significant differences were found for AIMS-physical activity at 12 weeks (WMD=−1.20, 95% CI=−2.29 to −0.11) (Fig. 15a), 15.2-m (50-ft) walking time at 12 weeks (WMD=−1.10 s, 95% CI=−2.12 to −0.08 s) (Fig. 15e), exercise heart rate at 12 weeks (WMD=13.00 bpm, 95% CI=1.32 to 24.68 bpm) (Fig. 15f), exercise endurance at 12 weeks (WMD=2.80 min, 95% CI=0.23 to 5.37 min) (Fig. 15h), and aerobic capacity (WMD=5.90 mL/kg min¹, 95% CI=3.30 to 8.50 mL/kg min¹) (results not shown). AIMS-pain, morning stiffness, grip force, trunk flexibility, and resting blood pressure offered no statistically significant differences (results not shown for last). One RCT that compared water exercises with control (n=30)⁸¹ yielded no statistically significant differences and no clinical benefits for torque or ROM at 6 weeks (Figs. 16a–b).

For yoga versus control (one RCT, n=30),⁶⁶ clinically important benefits were found for ROM and pain during activity at 6 weeks (Figs. 17b–c but not for tenderness, swelling, hand functional status, or grip force at 6 weeks (Figs. 17a, d, e, and f). Statistically significant data were found for mean change in tenderness of right hand (WMD=1.80, 95% CI=0.99 to 2.61) and left hand (WMD=1.73, 95% CI=0.63 to 2.83), mean change in pain during activity (WMD= −3.29, 95% CI=−5.30 to −1.28), and mean change in ROM of right hand (WMD=10.02, 95% CI=6.50 to 13.54), all at 6 weeks (Figs. 17a–c). No statistical data were found for mean change in hand pain at rest, mean change in ROM of left hand, mean change in circumference of the hands, mean change in hand functional status, or mean change in grip force of both hands (Figs. 17b–f).

Manual Therapy Combined With Therapeutic Exercises

One RCT (n=83)³⁰ was found on manual therapy combined with exercises, and this RCT compared the intervention with a control. Important clinical benefits were demonstrated for pain but not functional status (Tab. 16). Statistically significant data were found for all the outcomes: 6-minute walk test at 4 weeks (WMD=81.90 m, 95% CI=22.85 to 140.95 m) and at 8 weeks (WMD=77.70 m, 95% CI=18.59 to 136.81 m) (Fig. 18a) and pain at 4 weeks (WMD=−416.00, 95% CI=−618.15 to −213.85) and at 8 weeks (WMD=−471.90, 95% CI=−732.81 to −210.99) (Fig. 18b).

Strength of Published Evidence Compared With Other Guidelines

Good evidence (level I, RCT) shows that various kinds of exercises and manual therapy are useful for patients with OA, with different outcomes occurring depending on the intervention: strengthening exercises relieve pain at rest and during functional activities and improve knee ROM, quadriceps femoris muscle peak torque, grip force, level of energy, and functional status; general physical activities, including fitness and aerobic exercises, relieve pain during functional activities and improve stride length, functional status, and aerobic capacity; and manual therapy combined with TE relieves pain.

Three sources have considered the strength of evidence: The Philadelphia Panel,³⁶ American Pain Society,³³ and Ontario Program for Optimal Therapeutics.³⁵ All 3 sources reported good-quality evidence for TE, including strengthening exercises and general physical activities (Appendix 1). To our knowledge, the scientific literature offers no guidelines on manual therapy for patients with OA.

Clinical Recommendations Compared With Other Guidelines

The Ottawa Panel concluded that good evidence exists supporting the inclusion of all of the following main categories of interventions in the management of patients with OA: strengthening exercises (grade A for pain at rest and during functional activities, ROM, grip force, level of energy, and functional status; grade C+ for quadriceps femoris muscle peak torque, specific functional activities, and timed functional activities); general physical activities, including fitness and aerobic exercises (grade A for pain during functional activities, stride length, functional status, energy level, aerobic capacity, and medication use; grade C+ for disability in ADL); and manual therapy combined with exercises (grade A for pain). The recommendations related to strengthening exercises and general physical activities generally concur with all other existing guidelines.^31–36

Practitioner' Response to Ottawa Panel Guidelines

The 5 practitioners who reviewed our guidelines agreed with the recommendations. Four practitioners found the recommendations to be clear; 1 practitioner was uncertain which type of exercise was effective. The Ottawa Panel explained that, depending on the specific outcome, interventions with grade A, B, or C+ are beneficial. Guideline summaries (see evidence-based clinical practice guidelines in Appendix 4) were rewritten for better comprehension. A decision aid was created to clarify the application of the guidelines. This aid can be found on the University of Ottawa Web site (www.health.uottawa.ca/rehabguidelines).