Exercise for osteoarthritis of the knee

 

Characteristics of included studies [ordered by study ID]

Baker 2001

---

Methods	Moderate bias risk Unblinded assessor Intention-to-treat analysis Nutrition education control
Participants	46 volunteers, knee OA 74% female Mean age 69 years ACR criteria
Interventions	Home muscle strengthening program (+12 visits) Control: 7x home visits, nutrition education
Outcomes	At 16 weeks: Pain (WOMAC) Function (WOMAC)
Notes	Very closely monitored intensive strengthening program with 12 home visits over the 16 weeks (ankle weights, squats, etc)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Generated by independent statistician
Allocation concealment?	Yes	Central allocation
Blinding? All outcomes	No	Outcomes assessor unblinded
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Bautch 1997

---

Methods	High bias risk Unblinded assessor Efficacy analysis Education control
Participants	34 patients/volunteers, knee OA Mean age 68 years ACR criteria
Interventions	Individual program (12 weeks) providing 36 sessions ROM/ walking and education classes Control: 12 weekly education classes
Outcomes	At 12 weeks: Pain (VASx2) Function (AIMS)
Notes	Allocation groups very incomparable base pain/BMI/Xray with the active treatment allocation demonstrating more severe disease Low-intensity walking.
Risk of bias
Item	Authors' judgement	Description
Blinding? All outcomes	No

---

Bennell 2005

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Sham US control
Participants	140 community volunteers, knee OA 68% female mean age 68 years ACR criteria Pain>3
Interventions	Individual program (8 sessions over 12 weeks) Control: 8x sham Ultrasound
Outcomes	At 12 weeks: VAS pain WOMAC function
Notes	Novel intervention with little attention to knee strengthening. Taping, knee massage, thoracic mobs and hip mm strengthening
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer-generated random numbers
Allocation concealment?	Yes	Central allocation
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Deyle 2000

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Subtherapeutic US control
Participants	83 military care patients, knee OA 60% female Mean age 61 years ACR criteria
Interventions	Individual program (4 weeks) consisting of 8 sessions of manual therapy/strengthening exercises/aerobic exercise Control: Ultrasound (subtherapeutic)
Outcomes	At 8 weeks (delayed): Pain+Function (WOMAC)
Notes	30 minutes manual therapy + 30-45 minutes exercises
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers generator
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	No	Imbalance in missing data between allocation groups

---

Ettinger 1997a

---

Methods	Low bias risk Blinded assessor Intention-to-treat Education control
Participants	293 volunteers, knee OA 69% female Mean age 69 years ACR criteria
Interventions	Class based program. 36 sessions of aerobic walking Control: 3x monthy education classes then monthly telephone calls
Outcomes	Mean score of 3,9,18 months (delayed): Pain (FASTx6) Function (FASTx23)
Notes	Large classes (10-15 participants) Long follow-up after cessation of classes at 12 weeks.
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Web-based
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis, missing data balanced between allocation groups

---

Etttinger 1997b

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Education control
Participants	295 volunteers, knee OA 71% female Mean age 69 years ACR criteria
Interventions	Class-based program, 36 sessions muscle strengthening (cuff weights) Control: 3x monthy education classes then monthly telephone calls
Outcomes	Mean score of 3,9,18 months (delayed): Pain (FASTx6)
Notes	Large classes (10-15 participants) Long follow-up after cessation of classes (12 weeks)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Web-based
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis, missing data balanced between treatment groups

---

Foley 2003

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Waiting list
Participants	*Hip and knee OA 70 mostly clinic patients Mean age 70 years Radiographic criteria
Interventions	Class based program (6 weeks), 18 sessions of muscle strengthening, range of motion Control: Fortnightly telephone call
Outcomes	At 6 weeks: WOMAC pain WOMAC function
Notes	Separate analysis per knee OA only or hip OA, gym based group versus controls. About 40% on orthopaedic waiting list
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Central allocation
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Small numbers lost to follow up, balanced between allocation groups

---

Fransen 2001

---

Methods	Moderate bias risk Unblinded assessor Intention-to-treat analysis Waiting list
Participants	126 patients, knee OA 70% female Mean age 66 ACR criteria
Interventions	Individual or class-based allocation (8 weeks), 16 sessions with muscle strengthening and aerobic components Control: waiting list
Outcomes	At 8 weeks: WOMAC pain WOMAC function
Notes
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Yes	Sealed opaque envelopes, sequentially numbered
Blinding? All outcomes	Unclear	'mostly blinded' outcomes assessment
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Fransen 2007

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Waiting list
Participants	*Hip and knee OA 97 community volunteers 75% female Mean age 70 years ACR criteria
Interventions	Class based program (12 weeks), 24x Tai Chi classes Control: Waiting list
Outcomes	At 12 weeks: WOMAC pain WOMAC function
Notes	Disaggregated analysis (hip or knee OA) according to identified signal (most painful) joint
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Yes	Central allocation by administrator
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis
Free of selective reporting?	Yes	Trial registered NCT00123994

---

Gur 2002

---

Methods	Moderate bias risk Unblinded assessor Intention to treat (no dropouts) Control: no treatment
Participants	23 volunteers, knee OA Gender? Mean age: 56 years Radiographic, bilateral K&L 2-3 Sedentary past 10 years Cardiovascular clearance
Interventions	Individual program (8 weeks), 24 sessions of strengthening extensors/flexors (Cybex) Control: No treatment, but 2 additional testing sessions during during 8 week period
Outcomes	At 8 weeks: Pain (VAS - 7 items) Fx (VAS - 5 items)
Notes	No medications allowed Young sample High intensity, maximal effort
Risk of bias
Item	Authors' judgement	Description
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	No
Incomplete outcome data addressed? All outcomes	Yes	No dropouts

---

Hay 2006

---

Methods	Low risk of bias Blinded outcomes assessment Intention to treat analysis Control: No intervention
Participants	217 patients referred from general practice presenting with persistent knee pain and aged 55 years and over.
Interventions	Exercise advice and access to 3-6 sessions with physiotherapist over a 10 week period. Control: Advice/education leaflets with one followup telephone call.
Outcomes	At 3 months: WOMAC pain and physical function
Notes	Proportion with knee OA unknown
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers generator
Allocation concealment?	Unclear	Small blocks of 6 per practice
Blinding? All outcomes	Yes	Outcomes assessor blinded
Incomplete outcome data addressed? All outcomes	Yes	Loss to followup minimal and balanced

---

Hopman-Rock 2000

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Waiting list control
Participants	*Hip and knee 91 volunteers with OA knee 80% female Mean age 65 years
Interventions	Class-based program (6 weeks), 6 sessions of education + exercise
Outcomes	At 6 weeks VAS pain (2) IRGL mobility
Notes	Only 6 treatment occasions. Separate analysis for OA knee provided Only 28 volunteers reported hip OA
Risk of bias
Item	Authors' judgement	Description
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	Yes	Blinded outcomes assessment

---

Huang 2003

---

Methods	High bias risk Quasi-randomization Uncertain blinding Efficacy analysis Control: ROM/hot packs
Participants	132 patients, bilateral knee OA
Interventions	Individual program (8 weeks), 24 sessions of muscle strengthening (KinCom) extensor/flexor + hotpack/ROM Control: hotpack/ROM
Outcomes	At 8 wks/1 year VAS pain Lequesne funcion
Notes	Combined the three muscle strengthening groups for meta-analysis
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	No	Sequential numbers I-IV (representing treatment allocation)

---

Huang 2005a

---

Methods	Moderate bias risk Quasi-randomization Blinded assessor Efficacy analysis Active control
Participants	70 patients, bilateral moderate knee OA Lequesne score <7 Mean age 65 years 80% female
Interventions	Individual program (8 weeks), 24 sessions of muscle strengthening (Kin-Com) + hotpack/ROM Control: hotpack/ROM
Outcomes	At 8 weeks, 1 year VAS pain Lequesne fx
Notes	Analysed group 1 (exercise only) versus group 4 (control)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	No	Sequential numbers I-IV (representing treatment allocation)
Blinding? All outcomes	Yes

---

Hughes 2004

---

Methods	High bias risk Unblinded assessor Efficacy analysis Control: Arthritis Helpbook and list of accessible community exercise classes Waiting list (2 years)
Participants	*Hip and knee (combined) 150 community volunteers ACR criteria Mean age 74 years 83% female
Interventions	Class-based program (8 weeks), 24 sessions of muscle strenthening plus aerobic walking (1 hour) plus education/discussion (30 minutes). Control: Arthritis Help book and list of available community exercise programs
Outcomes	8 weeks, 6 months WOMAC pain WOMAC function
Notes	Large loss to followup at 2 months in controls (40%) Only simple exercise equipment used. Proportion of participants with knee versus hip OA unknown
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Incomplete outcome data addressed? All outcomes	No	Imbalance in missing data

---

Keefe 2004

---

Methods	High bias risk Unblinded assessor Efficacy analysis Control: No intervention
Participants	34 volunteers and patients. Persistent knee pain Married Mean age 59 years 50% female
Interventions	Class-based program (12 weeks), 36 aerobic sessions, 24 strengthening sessions
Outcomes	At 12 weeks: Pain: AIMS pain subscales
Notes	Analysed group 3 (exercise only) versus standard care (no spouse intervention groups)
Risk of bias
Item	Authors' judgement	Description
Incomplete outcome data addressed? All outcomes	Yes	Missing data minimal and balanced between allocation groups

---

Kovar 1992

---

Methods	High bias risk Unblinded assessor Efficacy analysis Control: Weekly telephone call
Participants	103 patients, knee OA 84% female Mean age 69 years Pain and + Xray
Interventions	Class-based program (8 weeks), 24 sessions of fitness walking/ stretch/education
Outcomes	At 8 weeks: Pain (AIMS) Function (AIMS) Global (AIMS)
Notes	Large classes (20-30 participants) Weekly telephone calls regarding ADL function for controls
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	No
Incomplete outcome data addressed? All outcomes	Yes	Missing data balanced between allocation groups

---

Maurer 1999

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Control: 4 education classes
Participants	113 patients, knee OA 42% female Mean age 64 years ACR criteria
Interventions	Individual program (8 weeks), 24 sessions of unilateral quadriceps strengthening only
Outcomes	At 8 weeks: Pain (WOMAC) Function (WOMAC)
Notes	Low % females Only unilateral exercise but many? with bilateral symptoms
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers generator
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Missing data balanced between groups, not study related

---

Messier 2004

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Control: Education classes/telephone calls
Participants	158 obese community volunteers 70% female Mean age 69 years Pain, disability and Xrays
Interventions	Class-based program (16 weeks), 48 sessions of strengthening and aerobic walking then telephone monitored home program (with weights) Control: Healthy lifestyle - 3 monthly education meetings re weight loss and exercise with followup telephone monitoring (about 8 calls)
Outcomes	At 6 and 18 months: (delayed) WOMAC pain WOMAC function
Notes	Analysis of exercise only versus healthy lifestyle control group Physical function assessed at 18 months, sd of baseline used. Very obese sample
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Web-based
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Mikesky 2006

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis ROM control
Participants	37 community volunteers in OA/pain strata 60% female Mean age 69 years Pain and + Xray
Interventions	Mostly home program (12 months) 0-12 months, 45 clinic sessions: strengthening KinCom 12-30 months, home program: strengthening (Theraband).
Outcomes	30 months (delayed): WOMAC pain WOMAC function
Notes	Analysis only of participants with knee OA/pain. Twice-weekly clinic-based classes in the first 12 weeks
Risk of bias
Item	Authors' judgement	Description
Blinding? All outcomes	Yes

---

Minor 1989

---

Methods	High bias risk Uncertain blinding Efficacy analysis ROM controls
Participants	80 patients/ volunteers, knee OA 80% female Mean age 64 years Pain and +Xray
Interventions	Class-based program (12 weeks), 36 sessions of aerobic walking Control: 36x ROM/relaxation
Outcomes	At 12 weeks: Pain (AIMS)
Notes	Large classes(max 12 participants) Aim of treatment was to increase aerobic capacity without exacerbation of symptoms
Risk of bias
Item	Authors' judgement	Description
Blinding? All outcomes	Yes

---

O'Reilly 1999

---

Methods	Moderate bias risk Unblinded assessor Intention-to-treat analysis Control: Lifestyle advice
Participants	180 volunteers, knee OA 66% female Mean age 62 years Knee pain past week
Interventions	Home program (+4 home visits), quads/hamstrings strengthening + lifestyle advice
Outcomes	At 6 months: Pain (WOMAC) Function (WOMAC)
Notes	Community sample with mostly mild radiographic/symptomatic disease (only 41% > KL grade I)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Unclear	Not reported if sealed envelopes were opaque with sequential numbers for audit trail
Blinding? All outcomes	No
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treatment analysis

---

Peloquin 1999

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Control: Education classes
Participants	137 volunteers, knee OA 70% female Mean age 66 years + Xray (< Grade 4)
Interventions	Class-based program (12 weeks), 36 sessions of aerobic and strength/stretching exercise Control: 12x education classes
Outcomes	At 12 weeks: Pain (AIMS) Function (AIMS)
Notes	Excluded people with severe disease: >10 ° varum, KL Grade 4, >15 ° of flexion deformity
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	No	Imbalance in missing data between allocation groups

---

Petrella 2000

---

Methods	Moderate bias risk Blinded assessor Intention-to-treat analysis Control: Sham exercise
Participants	179 patients/ volunteers, knee OA 58% female Mean age 74 years Xray (Gd 1-3) Pain in one knee most days
Interventions	Home program (unmonitored) of resistive quads/ ROM + NSAIDs Control: ROM + NSAIDs
Outcomes	At 8 weeks: Pain (WOMAC) Function (WOMAC)
Notes	Unmonitored low-intensity home program
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer-generated random numbers
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Unclear	Only n? participants with >75% program completion assessed

---

Quilty 2003

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Control: no co-intervention
Participants	87 community volunteers with patello-femoral pain Mean age 67 years
Interventions	9 physiotherapy sessions over 10 weeks
Outcomes	At 5 and 12 months: VAS pain WOMAC function
Notes	Zelen randomisation Treatment directed at patello-femoral joint
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Sealed opaque envelopes, sequentially numbered for audit trail
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Rogind 1998

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Control: no intervention
Participants	25 patients, knee OA 92% female Mean age 72 years ACR criteria + Xray (>KL Gd 2)
Interventions	Class-based program (12 weeks), 24 sessions complex mix of exercises Control: no intervention
Outcomes	At 12 weeks: Pain (VASx3) Function (AFI x10)
Notes	Moderate to severe disease Very complex exercise program (including venous, truncal muscles, balance)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Yes	No reported audit trail
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Unclear	Missing data minimal and balanced between the allocation groups

---

Schilke 1996

---

Methods	High bias risk Uncertain blinding Efficacy analysis Control: no intervention
Participants	20 patients, knee OA 85% female Mean age 66 years Rheumatology clinic attendees
Interventions	Individual program (8 weeks), 24 sessions of strengthening bilateral knee extensors and flexors
Outcomes	At 8 weeks: Pain (OASI) Function (OASI)
Notes	All training on Cybex. Intensive, maximal effort
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Unclear	No reported audit trail
Incomplete outcome data addressed? All outcomes	Yes	No missing data

---

Song 2003

---

Methods	Moderate bias risk Blinded assessor Efficacy analysis Controls: weekly telephone call
Participants	72 sedentary female patients, knee OA (confirmed by email) Mean age 65 years Clinical and radiographic criteria
Interventions	Class-based program (12 weeks), 16 sessions of Tai Chi classes
Outcomes	At 12 weeks: Pain and Function: Korean WOMAC
Notes	About 40% dropout
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Yes	Central allocation
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	No	43% missing data, efficacy analysis

---

Talbot 2003

---

Methods	High bias risk Unblinded assessor Efficacy analysis Control: Arthritis Self-Management Program classes
Participants	34 patients, knee OA Mean age 70 years 78% female ACR criteria OA knees
Interventions	Home program (12 weeks), 12 ASMP classes plus home-based pedometer walking program Control: 12 weekly ASMP classes
Outcomes	At 12 and 24 weeks Pain (McGill Pain questionnaire)
Notes	Evaluating the addition of a home based pedometer monitored walking program to the Arthritis Self-Management Program (ASMP)
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers table
Allocation concealment?	Unclear	B - Unclear
Blinding? All outcomes	No
Incomplete outcome data addressed? All outcomes	Yes	Minimal missing data, balanced between allocation groups

---

Thomas 2002

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Control: monthly telephone calls
Participants	786 patients, knee pain 65% female Mean age 62 years
Interventions	Home program (plus 8 visits), daily muscle strength training, bilateral, with Theraband Control: Short (2 minute) monthly telephone call
Outcomes	At 24 months: Pain (WOMAC) Function (WOMAC)
Notes	Patients with knee pain, may not all be OA
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Computer generated
Allocation concealment?	Yes	Central administration, sequential list audit trail
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

Thorstensson 2005

---

Methods	Moderate risk of bias Uncertain blinding outcomes assessment Efficacy analysis Waiting list control group
Participants	65 patients (identified by radiologists/orthopedic surgeons) with radiographic knee OA (Kellgren Lawrence grade 3 or more) and long standing knee pain. Aged between 35 and 65.
Interventions	Clinic-based classes, twice a week for 6 weeks, intensive muscle strengthening program. Control: waiting list for 6 months
Outcomes	At 6 weeks: KOOS pain KOOS ADL
Notes	Younger sample and more severe radiographic disease than most RCTs evaluating exercise for OA.
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Unclear	No sequence generation
Allocation concealment?	Unclear	No audit trail
Blinding? All outcomes	Unclear	Not reported
Incomplete outcome data addressed? All outcomes	Yes	Acceptable and balanced loss to follow-up

---

Topp 2002

---

Methods	High bias risk Uncertain blinding Efficacy analysis? Control: No intervention
Participants	102 volunteers 74% female Mean age 63 years ACR clinical criteria
Interventions	Class-based program (16 weeks), 16 sessions of muscle strengthening (dynamic or isometric) with Theraband
Outcomes	At 16 weeks: Pain (WOMAC) Function (WOMAC)
Notes	Clinic-based classes 1x per week Home program 2x per week

---

van Baar 1998

---

Methods	Low bias risk Blinded assessor Intention-to-treat analysis Control: GP education
Participants	113 patients, knee OA 79% female Mean age 68 years ACR criteria
Interventions	Individual program (12 weeks), 17 sessions of physiotherapy + GP education
Outcomes	At 12 weeks: Pain (VASx1) Function IRGL
Notes	Recruited hip and knee OA patients. Separate results provided for knee OA. Mostly early disease as approx. 50% sample had symptom duration <1 year
Risk of bias
Item	Authors' judgement	Description
Adequate sequence generation?	Yes	Random numbers talbe
Allocation concealment?	Yes	Sealed opaque envelopes, sequential numbering for audit trail
Blinding? All outcomes	Yes
Incomplete outcome data addressed? All outcomes	Yes	Intention-to-treat analysis

---

 ACR: American College of Rheumatology
WOMAC: Western Ontario and McMasters Universities Osteoarthritis Index
ROM: Range of motion
US: Ultrasound
KL: Kellgren and Lawrence
VAS: Visual analogue scale
AFI: Arthritis Function Index
GP: General practitioner
IRGL: Influence of Rheumatic Disease on Health and Lifestyle scale.
ASMP: Arthritis Self Management Program

 

Characteristics of excluded studies [ordered by study ID]

---

Study	Reason for exclusion
Atamaz 2006	Physical therapy did not include an exercise program (IR, short wave diathermy, interferential)
Borjesson 1996	Patients scheduled for joint replacement surgery.
Callaghan 1995	Unable to ascertain effect size as only provided with median % improvements without baseline scores and with extremely wide confidence intervals due to small sample.
Chamberlain 1982	No appropriate control. Assessed added benefit of SWD to exercises.
Cheing 2002	No control group. Control group was an extremely effective sham TENS.
Cheing 2004	Secondary analysis of Cheing 2002. Only gait and muscule strength evaluated.
Deyle 2005	No non-exercise control group
Dias 2003	Unable to extract change (sd) or post treatment (sd) scores from the published manuscript. Unusually, the published manuscript only provided median/test statistic/degrees of freedom data.
Diracoglu 2005	No non-exercise control group
Durmus 2007	No non-exercise control group.
Eungpinichpong 1997	Only abstract. No outcome scores provided and no full report found.
Evcik 2002	Not a randomized trial. Patients were 'separated' into three groups.
Eyigor 2004	No non-exercise control group.
Forster 2007	No non-exercise control group.
Green 1993	No appropriate control. Assessed added benefit of hydrotherapy to home exercises.
Haslam 2001	Advice and exercise was the control group. The evaluated treatment was acupuncture.
Hinman 2007	No land-based exercise group.
Hoeksma 2004	No non-exercise control. Manual therapy versus exercise.
Huang 2005b	Earlier version of Huang 2005 (1).
Hurley 1998	Not even quasi-randomized.
Jan 1991	Not even quasi-randomized.
Kreindler 1989	No pain/function/patient global outcome assessment. Only outcome is muscle strength.
Kuptniratsaikul 2002	Cluster random sampling.
Lankhorst 1982	No control group in analysis of results. No pain/function/patient global outcome assessment.
Lim 2002	No non-exercise control group.
Lin 2004	Water exercise program.
Lin 2007	No self-reported pain, physical function or patients global assessment.
Mangione 1999	No appropriate control group. Both allocations on stationary cycling, high versus low intensity.
McCarthy 2004	No non-exercise control group.
Messier 1997	Secondary analysis Ettinger 1997. Gait assessment.
Messier 2000 (1)	No appropriate control group. Assessed added benefit of dietary therapy to an exercise program.
Messier 2000 (2)	Secondary analysis Ettinger 1997. Balance assessment.
Moss 2007	No exercise group - patient passive for mobisations.
Nicklas 2004	Secondary analysis Messier 2004. Outcomes limited to markers of chronic inflammation.
Ozdincler 2005	No non-exercise control group.
Penninx 2001	Secondary analysis Ettinger 1997.
Penninx 2002	Secondary analysis Ettinger 1997.
Peterson 1993	Secondary analysis Kovar 1992. Gait assessment.
Quirk 1985	No appropriate control group. Assessed added benefit of interferential therapy or SWD to exercises.
Ravaud 2004	Cluster randomised trial.
Rejeski 1998	Secondary analysis Ettinger 1997.
Sen 2004	No non-exercise control group.
Stitik 2007	Not randomised or quasi randomised - sequentially assigned. In addition, all participants received hyaluronan (5 or 3 weekly injections).
Sullivan 1998	Secondary analysis (1 year follow up) Kovar 1992.
Sylvester 1989	No appropriate control. Hydrotherapy compared with exercises plus SWD (N=14).
Toda 2001	Not randomized.
Tuzun 2004	No non-exercise control group.
van Baar 2001	Secondary analysis van Baar 1998 (follow-up study).
Van Gool 2005	Secondary analysis ADAPT study.
Wang 2006	No land-based exercise group.
Williamson 2007	Patients awaiting knee replacement surgery.
Wyatt 2001	No non-exercise control.
Yip 2007	Evaluated treatment was the Arthritis Self-management program, with exercise an added component. Unable to attribute treatment effect to education or home exercise program.

---

 

Comparison 1. All included studies

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pain	32	3616	Std. Mean Difference (IV, Random, 95% CI)	-0.40 [-0.50, -0.30]
2 Physical function	31	3719	Std. Mean Difference (IV, Random, 95% CI)	-0.37 [-0.49, -0.25]

---

 

Comparison 2. Treatment content

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Simple quadriceps strengthening

3

319

Std. Mean Difference (IV, Random, 95% CI)

-0.29 [-0.51, -0.06]

---

1.2 Lower limb muscle strengthening

9

1383

Std. Mean Difference (IV, Random, 95% CI)

-0.53 [-0.79, -0.27]

---

1.3 Strengthening and aerobic

9

998

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.61, -0.19]

---

1.4 Walking program

4

351

Std. Mean Difference (IV, Random, 95% CI)

-0.48 [-0.83, -0.13]

---

1.5 Other

7

565

Std. Mean Difference (IV, Random, 95% CI)

-0.32 [-0.49, -0.15]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Simple quadriceps strengthening

4

498

Std. Mean Difference (IV, Random, 95% CI)

-0.24 [-0.42, -0.06]

---

2.2 Lower limb muscle strengthening

9

1383

Std. Mean Difference (IV, Random, 95% CI)

-0.58 [-0.88, -0.27]

---

2.3 Strengthening and aerobic

8

967

Std. Mean Difference (IV, Random, 95% CI)

-0.42 [-0.65, -0.18]

---

2.4 Walking program

3

317

Std. Mean Difference (IV, Random, 95% CI)

-0.35 [-0.58, -0.11]

---

2.5 Other

7

554

Std. Mean Difference (IV, Random, 95% CI)

-0.22 [-0.48, 0.05]

---

 

Comparison 3. Treatment delivery mode

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3717

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Individual treatments

10

849

Std. Mean Difference (IV, Random, 95% CI)

-0.55 [-0.81, -0.29]

---

1.2 Class-based programs

17

1608

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.51, -0.24]

---

1.3 Home program

6

1260

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.39, -0.16]

---

 2 Physical Function

31

3820

Std. Mean Difference (IV, Random, 95% CI)

-0.36 [-0.48, -0.25]

---

2.1 Individual treatments

10

849

Std. Mean Difference (IV, Random, 95% CI)

-0.52 [-0.86, -0.19]

---

2.2 Class-based programs

16

1563

Std. Mean Difference (IV, Random, 95% CI)

-0.35 [-0.50, -0.19]

---

2.3 Home program

6

1408

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.38, -0.17]

---

 

Comparison 4. Number of contact occasions

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Less than 12 occasions

9

1594

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.40, -0.16]

---

1.2 12 or more occasions

23

2022

Std. Mean Difference (IV, Random, 95% CI)

-0.46 [-0.60, -0.32]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Less than 12 occasions

9

1731

Std. Mean Difference (IV, Random, 95% CI)

-0.23 [-0.37, -0.09]

---

2.2 12 or more occasions

22

1988

Std. Mean Difference (IV, Random, 95% CI)

-0.45 [-0.62, -0.29]

---

 

Comparison 5. Blinding of outcomes assessment

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Blinded assessment

18

2559

Std. Mean Difference (IV, Random, 95% CI)

-0.33 [-0.43, -0.22]

---

1.2 Uncertain/unblinded assessment

14

1057

Std. Mean Difference (IV, Random, 95% CI)

-0.53 [-0.73, -0.33]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Blinded assessment

19

2730

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.39, -0.17]

---

2.2 Uncertain/unblinded assessment

12

989

Std. Mean Difference (IV, Random, 95% CI)

-0.55 [-0.83, -0.28]

---

 

Comparison 6. Statistical analysis method

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Intention to treat

14

2394

Std. Mean Difference (IV, Random, 95% CI)

-0.36 [-0.51, -0.21]

---

1.2 Efficacy

18

1222

Std. Mean Difference (IV, Random, 95% CI)

-0.45 [-0.58, -0.32]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Intention to treat

15

2576

Std. Mean Difference (IV, Random, 95% CI)

-0.30 [-0.45, -0.16]

---

2.2 Efficacy

16

1143

Std. Mean Difference (IV, Random, 95% CI)

-0.43 [-0.64, -0.23]

---

 

Comparison 7. Allocation concealment

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3801

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Adequate allocation concealment

14

2599

Std. Mean Difference (IV, Random, 95% CI)

-0.33 [-0.44, -0.23]

---

1.2 Unclear allocation concealment

18

1202

Std. Mean Difference (IV, Random, 95% CI)

-0.49 [-0.66, -0.32]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Adequate allocation concealment

15

2596

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.38, -0.18]

---

2.2 Unclear allocation concealment

16

1123

Std. Mean Difference (IV, Random, 95% CI)

-0.48 [-0.73, -0.23]

---

 

Comparison 8. Estimated risk of bias

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 Low risk of bias

10

2021

Std. Mean Difference (IV, Random, 95% CI)

-0.28 [-0.42, -0.15]

---

1.2 Moderate risk of bias

13

972

Std. Mean Difference (IV, Random, 95% CI)

-0.48 [-0.67, -0.29]

---

1.3 High risk of bias

9

623

Std. Mean Difference (IV, Random, 95% CI)

-0.51 [-0.72, -0.30]

---

 2 Physical Function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 Low risk of bias

10

2024

Std. Mean Difference (IV, Random, 95% CI)

-0.25 [-0.38, -0.13]

---

2.2 Moderate risk of bias

14

1140

Std. Mean Difference (IV, Random, 95% CI)

-0.39 [-0.60, -0.18]

---

2.3 High risk of bias

7

555

Std. Mean Difference (IV, Random, 95% CI)

-0.55 [-0.90, -0.21]

---

 

Comparison 9. Sample size

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Pain

32

3616

Std. Mean Difference (IV, Random, 95% CI)

-0.40 [-0.50, -0.30]

---

1.1 n>50 in each allocation

9

2120

Std. Mean Difference (IV, Random, 95% CI)

-0.29 [-0.42, -0.16]

---

1.2 25<n<50 in any allocation

12

1099

Std. Mean Difference (IV, Random, 95% CI)

-0.43 [-0.57, -0.28]

---

1.3 n<25 in any allocation

11

397

Std. Mean Difference (IV, Random, 95% CI)

-0.63 [-0.92, -0.34]

---

 2 Physical function

31

3719

Std. Mean Difference (IV, Random, 95% CI)

-0.37 [-0.49, -0.25]

---

2.1 n>50 in each allocation

10

2302

Std. Mean Difference (IV, Random, 95% CI)

-0.24 [-0.33, -0.16]

---

2.2 25< n<50 in any allocation

12

1088

Std. Mean Difference (IV, Random, 95% CI)

-0.43 [-0.66, -0.20]

---

2.3 n>25 in any allocation

9

329

Std. Mean Difference (IV, Random, 95% CI)

-0.63 [-1.04, -0.22]

---

 

Comparison 10. Clinical Relevance

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 WOMAC Pain	14	2051	Std. Mean Difference (IV, Random, 95% CI)	-0.35 [-0.48, -0.21]
2 VAS Pain	8	669	Std. Mean Difference (IV, Random, 95% CI)	-0.43 [-0.65, -0.21]
3 WOMAC Function	17	2460	Std. Mean Difference (IV, Random, 95% CI)	-0.28 [-0.39, -0.17]

---

 

Table 1. Clinical Relevance: WOMAC Pain

---

Outcome (scale)	# patients (#trials)	Control baseline m*	Wt absolute change	Relative % change	NNT(B)	Statistical signific	Quality of evidence
Improvement in knee pain (WOMAC 0-20)	1869 (13)	7.4*	5.2% (1.0 fewer points on a 0-20 scale)	14.1% (I)	7	Statistically significant	Platinum
95% CI		*Thomas 2002	(3.6%, 6.8%)	(10.3%, 18.5%)

---

 

Table 2. Clinical Relevance: VAS Pain

---

Outcome (scale)	# patients (#trials)	Control baseline m*	Wt absolute change	Relative % change	NNT(B)	Statist significance	Quality of evidence
Improvement in knee pain (VAS 0-10)	669 (8)	5.2*	7.5% (0.75 fewer points on a 0-10 scale)	14.4% (I)	6	Statistically significant	Platinum
		*(Bennell 2005)	(4.8%, 10%)	(9.2%, 19.3%)

---

 

Table 3. Clinical relevance: WOMAC Physical function

---

Outcome (scale)	# patients (#trials)	Control baseline m*	Wt absolute change	Relative % change	NNT(B)	Statist significance	Quality of evidence
Improvement in physical function (WOMAC 0-68)	2275 (16)	23.0*	3.9% (2.6 fewer points on a 0-68 scale)	11.4% (I)	8	Statistically significant	Platinum
		(*Thomas 2002)	(2.6%, 5.3%)	(7.8%, 15.5%)

---