Chang 1993.

Study characteristics
Methods	Study design: multicentre, parallel‐group, two‐arm randomised controlled trial Setting: the Rheumatology‐Orthopedic Knee Clinic of the Northwestern Medical Faculty Foundation and the Division of Rheumatology of the Lutheran General Medical Group, Illinois, USA Trial time period: not reported Interventions: arthroscopic surgery versus non‐arthroscopic joint lavage Sample size calculations: authors did not describe how the sample size was estimated Analysis: intention‐to‐treat analysis
Participants	Number of participants Number screened: > 200 (110 ineligible) Number of participants at enrolment: 90 (45 underwent arthroscopic surgery outside the study) Number randomised: 34: 19 in the arthroscopy group and 15 in the lavage group Number included in analyses: 32 participants, 18 in arthroscopy group, 14 in lavage group were included in the 3‐month and 12‐month analysis. Two participants (1 from arthroscopy group, 1 from lavage group) dropped out before treatment was given due to intercurrent medical problems and were excluded from the analysis. Inclusion criteria Persistent knee pain for longer than 3 months, despite conservative medical and rehabilitation management, which restricted work, athletic, or self‐care activities to an extent unacceptable to the patient Weightbearing knee radiographs showing Kellgren‐Lawrence (KL) classification grade 1, 2, or 3 changes Age > 20 years Willingness to attend follow‐up visits at 3 and 12 months Willingness to give written informed consent In participants with bilateral disease, the more symptomatic knee was designated the study knee. Exclusion criteria Knee surgery within 6 months of study entry Total knee replacement Any concurrent illness which would influence functional assessment of the knee or preclude arthroscopic surgery (e.g. severe intermittent claudication or cardiac disease) KL grade 4 changes or radiographs, as determined by one of the authors Baseline characteristics Arthroscopic surgery group (N = 18 included in analyses) Mean (SD) age: 61 (11) years % female: 72 Mean (SD) duration of knee pain (months): 51 (51) Kellgren X‐ray classification (%): Class I ‐ 22, Class II ‐ 28, Class III ‐ 50 Functional class (%): Class I ‐ 5, Class II ‐ 67, Class III ‐ 28 Walk distance (%): > 4 blocks ‐ 44, 2 to 4 blocks ‐ 11, < 2 blocks ‐ 44 Using assistive devise to walk (%): 50 Mean (SD) initial active ROM (degrees): 114 (21) Knee joint tenderness measured on a 4‐point ordinal scale, defined by the American College of Rheumatology (ACR) score of 3 or 4 (%): 29 Knee joint swelling measured on a 4‐point ordinal scale, defined by the ACR score of 3 or 4 (%): 41 Mean (SD) initial Arthritis Impact Measurement Scale (AIMS) scores: pain 6.5 (2.0), physical activity 6.9 (2.0), physical function 2.3 (1.6), social activity 4.4 (2.2), depression 2.8 (2.2), anxiety 4.2 (2.4) Mean (SD) 50‐foot walk time (seconds): 14.9 (4.3) Mean (SD) participant's global assessment (10 cm scale): 4.6 (2.6) Physician's global assessment score of 3 or 4 (%): 12 Joint lavage group (N = 14 included in analyses) Mean (SD) age: 65 (13) years % female: 71 Mean (SD) duration of knee pain (months): 53 (57) Kellgren X‐ray class (%): Class I ‐ 14, Class II ‐ 36, Class III ‐ 50 Functional class (%): Class I ‐ 7, Class II ‐ 79, Class III ‐ 14 Walk distance (%): > 4 blocks ‐ 50, 2 to 4 blocks ‐ 14, < 2 blocks ‐ 36 Using assistive devise to walk (%): 43 Mean (SD) initial active ROM (degrees): 111 (20) Knee joint tenderness measured on a 4‐point ordinal scale, defined by the ACR score of 3 or 4 (%): 31 Knee joint swelling measured on a 4‐point ordinal scale, defined by the ACR score of 3 or 4 (%): 21 Mean (SD) initial AIMS scores: pain 6.1 (2.1), physical activity 5.3 (2.1), physical function 1.7 (1.0), social activity 4.7 (2.6), depression 2.6 (2.0), anxiety 3.9 (2.4) Mean (SD) 50‐foot walk time (seconds): 15.0 (4.7) Mean (SD) participant's global assessment (10 cm scale): 4.6 (2.5) Physician's global assessment score of 3 or 4 (%): 16 Pre‐treatment group differences: the baseline demographic, clinical and functional characteristics were similar between the two groups except for the initial AIMS Physical Activity Score (a statistically significant difference between groups (P < 0.05) was identified).
Interventions	Arthroscopic surgery group Arthroscopic surgery plus physical therapy and analgesia. Arthroscopy was done under general anaesthesia. A diagnostic evaluation was performed, and the anatomic findings were recorded on a standardised form. Following this evaluation, any of the following interventions were performed under arthroscopic guidance: (1) debridement of torn meniscus and removal of meniscal and cruciate ligament fragments; (2) removal of proliferative synovium; and (3) excision of loose articular cartilage fragments. Osteochondral lesions were not drilled. All participants received continuous saline lavage during the procedure and were routinely instructed in partial weightbearing precautions to continue for 10 days following the procedure. If an osteochondral lesion was detected in a weight‐bearing area, this period of protection was increased to 3 weeks. Prior to and following surgery, participants assigned to this group received only non‐narcotic analgesia and physical therapy, consisting of strengthening and flexibility exercises and gait training. Non‐arthroscopic (closed‐needle joint) lavage group Closed‐needle joint lavage plus physical therapy and analgesia. Participants assigned to this group received non‐narcotic analgesia and physical therapy identical to the arthroscopy group. In addition, participants received a tidal knee lavage procedure which was chosen to offset the potentially strong placebo effect of a surgical procedure and to control for the effects on pain and disability of the lavage procedure that occur during the arthroscopic procedure. Tidal knee lavage was performed as described by Ike and colleagues (Ike 1992) under local anaesthesia. A total of 1 litre of saline was injected into and aspirated from the knee in aliquots of 40 cc to 120 cc, depending on the size of the knee capsule. Post‐intervention Participants in both groups received non‐narcotic analgesia and physical therapy.
Outcomes	Outcomes were measured at baseline and at 3 and 12 months of follow‐up Clinical parameters Active and passive range of knee motion in degrees measured using goniometry Knee joint swelling measured on a 4‐point ordinal scale, defined by the American College of Rheumatology (ACR) glossary. Improvement defined as a decrease of at least 1 point on the scale. Knee joint tenderness measured on a 4‐point ordinal scale, as defined by the ACR glossary. Improvement defined as a decrease of at least 1 point on the scale. Pain and functional status measures Arthritis Impact Measurement Scales (AIMS) scales for pain, physical function, physical activity, social activity, depression, and anxiety. The AIMS scales are scored from 0 (best) to 10 (worst) in a self‐administered patient questionnaire. Improvement in the pain score was defined as a decrease of at least 1 point from the baseline score. Observed functional status was assessed with the 50‐foot walk time in seconds. Global measures Patient's global assessment (overall well‐being) measured on a 10 cm visual analogue scale (VAS) scored from 0 (best) to 10 (worst). Improvement from the participant’s perspective was defined as a reduction of > 1 cm from the baseline VAS score. Probability of improvement was defined as the proportion of participants who improved according to this criterion. Physician’s global assessment of disease activity in the knee was made using a 4‐point ordinal scale, ranging from no disease to very severe disease. Improvement from the physician’s perspective was defined as a decrease of at least 1 point on the scale. Probability of improvement was defined as the proportion of participants who improved according to this criterion. Economic measures Costs of the arthroscopic surgery or the tidal knee lavage Costs of medications Costs of other medical services, including physician and therapist visits Indirect costs of OA of the knee, including employment status and the use of paid and unpaid aides to help with activities of daily living Outcomes included in this review at 3 and 12 months Pain measured on AIMS‐P subscale (0 to 10, lower score = less pain) Function measured on AIMS‐PF subscale (0 to 10, lower score = better function) (we multiplied the mean values by –1 so that higher scores = better function, as per Cochrane Handbook guidance, so direction was consistent across function scales) Participant‐reported treatment success (patient's global assessment) measured on VAS
Notes	Funding: supported by grant 9040 from the Robert Wood Johnson Foundation, by MAC grant AR‐30692 from the NIH (NIAMS), and by the Percy Surgical Research Trust of Lutheran General Hospital Trial registration: not reported Adverse events: unclear if measured; not reported Knee surgery (replacement or osteotomy): not reported Progression of knee OA: not reported Withdrawals: 2/34 (6%), 1 (5%) from arthroscopy group and 1 (7%) from lavage group, dropped out after randomisation but before treatment commenced and were excluded from analysis at 3 and 12 months. Missing data at 12 months imputed by trialists using 3‐month outcomes : 7/32 (22%), 5 (28%) from arthroscopy group, 2 (14%) from lavage group. 5 participants (4 from arthroscopy group, 1 from lavage group) dropped out and did not receive the assessment at the 12‐month follow‐up. No reason reported. 2 participants from lavage group were not satisfied with non‐operative management and received arthroscopic surgery after 3 months of follow‐up. Treatment non‐adherence: 3/18 participants from arthroscopy group withdrew before surgery; one improved between enrolment and planned surgery so cancelled surgery, and two developed other medical illnesses that precluded surgery. Data analysis: missing standard deviations for outcomes used in this review: contacted authors but no response received. We imputed SDs.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Used a modification of the pre‐randomisation design described by Zelen (Zelen 1981). Eligible participants were asked if they would accept an arthroscopic procedure if it was offered. Subjects who answered 'yes' were randomly assigned to arthroscopy or lavage and then asked to accept the assigned therapy. The randomisation plan was stratified by study site. No description of sequence generation process provided
Allocation concealment (selection bias)	Unclear risk	Insufficient description of the method of concealment. Use of assignment 'envelopes' described but unclear if safeguards used (e.g. sealed, sequentially numbered, opaque)
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants and investigators was not done
Blinding of outcome assessor Self‐reported outcomes	High risk	Blinding of participants was not done; hence, there was a risk of bias in the measurement of subjective outcomes of pain, physical function, physical activity, social activity, depression, and anxiety (AIMS subscales) and participants' global assessment
Blinding of outcome assessor Assessor‐reported outcome (knee replacement)	Low risk	Blinding of outcome assessors was done. Participants were asked not to disclose their treatment assignment to assessors and to cover actual, or potential, arthroscopy scars
Incomplete outcome data (attrition bias) All outcomes	Low risk	Withdrawals (n = 2) were balanced in number across groups and for similar reasons. Missing data at 12 months were imputed using appropriate methods
Selective reporting (reporting bias)	Unclear risk	Trial registration not done and study protocol not available. Insufficient information to judge high or low risk
Other bias	Low risk	No other biases apparent