Moseley 1996.

Study characteristics
Methods	Study design: single centre, three‐arm, double‐blind, randomised, placebo‐controlled trial Setting: Houston Veterans Administration (VA) Medical Center, Texas, USA Trial time period: June 1992 Interventions: arthroscopic debridement versus arthroscopic lavage versus placebo surgery Sample size calculations: a priori sample size calculation was not done Analysis: statistical analysis not done
Participants	Number of participants Number of participants screened for eligibility: not reported Number of participants enrolled: 10 Number randomised: 10 participants: 5 participants were randomised to the placebo group, 3 were randomised to the arthroscopic lavage group and 2 to the arthroscopic debridement group Number included in analysis: a statistical analysis was not performed because of the small number of subjects and responses. However, data were available from all 10 participants at 2 and 6 weeks and from 9 participants at 3 and 6 months and reported as means. Inclusion criteria Symptomatic osteoarthritis of the knee in spite of a minimum of 6 months of non‐operative treatment including non‐steroidal anti‐inflammatory medication At least moderate knee pain (~ 4 on a 0 to 10 scale) on average over a week’s time Age under 70 No medical problems that placed the participant at significant risk for complications from a general anaesthetic (as determined by the general medicine consult service at the Houston VA Medical Center) Exclusion criteria Not reported Baseline characteristics Arthroscopic debridement (N = 2) Mean intensity of worst knee pain on a scale where 1 = no pain, 10 = severe pain: 9.0 Mean average intensity of knee pain on a scale where 1 = no pain, 10 = severe pain: 7.0 Mean intensity of today's pain on a scale where 1 = no pain, 10 = severe pain: 6.0 Number of days this week with knee pain: 6 Average knee extension (negative values refer to flexion contracture): ‐12.0 Average knee flexion: 111 Average knee crepitus on a scale: 0 = none, 3 = severe: 1.5 Average knee effusion on a scale: 0 = none, 3 = severe: 0 Average global knee tenderness on a scale: 0 = none, 9 = severe: 5.0 50‐foot walk in seconds: 12.0 Arthroscopic Lavage (N = 3) Mean intensity of worst knee pain on a scale where 1 = no pain, 10 = severe pain: 8.5 Mean average intensity of knee pain on a scale where 1 = no pain, 10 = severe pain: 7.5 Mean intensity of today's pain on a scale where 1 = no pain, 10 = severe pain: 5.5 Number of days this week with knee pain: 6 Average knee extension (negative values refer to flexion contracture): ‐5.0 Average knee flexion: 105 Average knee crepitus on a scale: 0 = none, 3 = severe: 1.5 Average knee effusion on a scale: 0 = none, 3 = severe: 1.0 Average global knee tenderness on a scale: 0 = none, 9 = severe: 4.0 50‐foot walk in seconds: 11.0 Placebo surgery (N = 5) Mean intensity of worst knee pain on a scale where 1 = no pain, 10 = severe pain: 8.4 Mean average intensity of knee pain on a scale where 1 = no pain, 10 = severe pain: 6.8 Mean intensity of today's pain on a scale where 1 = no pain, 10 = severe pain: 6.8 Number of days this week with knee pain: 5.6 Average knee extension (negative values refer to flexion contracture): ‐5.2 Average knee flexion: 125 Average knee crepitus on a scale: 0 = none, 3 = severe: 1.0 Average knee effusion on a scale: 0 = none, 3 = severe: 0 Average global knee tenderness on a scale: 0 = none, 9 = severe: 2.8 50‐foot walk in seconds: 11.5 Pre‐treatment group differences: there were no obvious differences in baseline characteristics between the three groups except for greater flexion contracture in the debridement group compared to the other two groups.
Interventions	The arthroscopic debridement and arthroscopic lavage groups both received a standard general endotracheal anaesthetic that was routinely used at the medical centre. The placebo arthroscopy group received a lesser anaesthetic, an intravenous tranquilliser (benzodiazepine, droperidol, or both) in conjunction with an opioid (fentanyl or sufentanil), without having placement of an endotracheal tube. This combination of drugs, along with the local anaesthetic injected at the stab wound sites, sedated the participant and rendered them insensitive to pain. The placebo participants breathed spontaneously throughout the procedure, and they inhaled oxygen supplemented by a nasal cannula. End‐tidal carbon dioxide was continuously monitored throughout the procedure. Using this type of anaesthesia for the placebo group minimised the potential complications from induction of general anaesthesia and placement of an endotracheal tube. Once anaesthetised, all participants had their knees examined and then prepared and draped in the usual manner. All participants had a gram of cephalosporin antibiotic administered intravenously as prophylaxis against infection. Bupivacaine (0.25%) with epinephrine was then injected into the skin where the arthroscopy portals were to be made. Arthroscopic debridement The arthroscopic debridement participants had three stab wounds made and an arthroscope inserted in the inferolateral portal; an inflow cannula was inserted in the superomedial portal, and the various operating instruments were inserted from the inferomedial portal. The knee was distended with sterile saline from the inflow cannula, and a constant flow of fluid was lavaged through the knee as is typical for arthroscopic surgery. A minimum of 10 litres of fluid was lavaged through the knee. After a diagnostic arthroscopic examination, the arthroscopic instruments were used to shave the rough articular cartilage (chondroplasty), remove loose debris, trim torn or degenerated meniscal fragments, and correct any other soft tissue abnormalities that could interfere with the mechanical function of the knee. However, no abrasion arthroplasty or removal of bone spurs was performed. At the end of the procedure, the instruments were removed, the portals were closed with absorbable suture, and a sterile compression dressing was applied. The average time for the surgery was 45 minutes. Arthroscopic lavage The arthroscopic lavage participants had a procedure identical to that of the debridement participants, except that no operating instruments were used to remove or trim the various parts of the knee. A diagnostic arthroscopic examination was performed, and a minimum of 10 litres of fluid was lavaged through the knee. A minimum of 30 minutes was spent performing the lavage procedure, and the total time spent in the operating room was approximately 1 hour. Placebo surgery Participants undergoing placebo arthroscopy were prepared, draped, examined and injected with local anaesthetic in the same manner as the other two groups. Three stab wounds were made in the skin with a scalpel, but no instruments of any kind were placed into the knee. The knee was manipulated, instruments were requested and passed, saline was splashed, and a standard arthroscopic debridement was simulated as closely as possible in the event the participant was not totally unaware during the event. A minimum of 30 minutes was spent performing the placebo surgery, and the typical time spent in the operating room was 1 hour. Participants in all three groups spent approximately an hour in the operating theatre. All participants were taken to the recovery room and treated the same as any participant having any arthroscopic procedure. Once the participant returned to the orthopaedic ward from the recovery room, he/she was observed until stable enough to be sent home. All participants were discharged from the hospital the afternoon of surgery or the next morning with an oral narcotic analgesic (typically acetaminophen with codeine) for use as required. Before discharge, participants were fitted with crutches and instructed to discontinue using them as soon as they could walk comfortably without a limp. At the first post‐operative visit, participants were instructed to resume their pre‐operative anti‐inflammatory medications and to resume walking and other activities of daily living as soon as their symptoms would allow.
Outcomes	Outcomes were measured at 2 and 6 weeks, and 3 and 6 months. Study outcomes Pain (5 domains: intensity of worst knee pain; average intensity of knee pain; intensity of today's pain; pain relief experienced after surgery; number of days this week with knee pain). Domains 1 through 3 were each measured on a 10‐point scale where 1 = no pain, 10 = severe pain. Domain 4 was measured on a 10‐point scale where 1 = no relief, 10 = complete relief. Mobility (activity) was measured but the details of the measurement scale and the findings were not reported in this paper General well‐being was measured but the details of the measurement scale and the findings were not reported in this paper Satisfaction with surgery, measured using two questions (would you recommend the surgery to your family or friends?; do you feel the operation was worthwhile?) each using a 5‐point Likert scale (strongly agree to strongly disagree) 50‐foot walk, time in seconds Knee extension and flexion measured via physical examination Knee crepitus and effusion measured on a 4‐point scale: 0 = none, 3 = severe Knee tenderness measured on a 10‐point scale: 0 = none, 9 = severe The subjective data (pain, mobility, general well‐being, satisfaction with surgery) were collected from a questionnaire designed specifically for this study (not included in trial report). The questionnaire was reported to be based on the AIMS‐2, the SF‐36, the Wisconsin Brief Pain Questionnaire and the Knee Society's Knee Rating Scale. Outcomes used in this review at 3 and 6 months Pain ‐ average intensity of knee pain, measured on a 10‐point scale (lower score = less pain) Participant reported success ‐ satisfaction with surgery at 6 months, measured as number of participants reporting 'strongly agree' or 'slightly agree' for item 'do you feel the operation was worthwhile?'
Notes	Funding: not reported 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: 1/10 in total. 1/2 (50%) from the arthroscopic debridement group Analysis: we included arthroscopic debridement versus placebo and we excluded the lavage treatment arm from this review. Data imputations: SDs for pain were not reported at 3 and 6 months (and no baseline values were reported). We used SD values from Sihvonen 2013 for pain at 3 and 6 months for this study in Analysis 2.1 as Sihvonen and colleagues also used a pain numerical rating scale. For satisfaction with surgery, we assumed dropouts (one in arthroscopic debridement group) were unsatisfied and used the number randomised as the denominator.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information about sequence generation process given, other than stating it was randomised. "An unbalanced randomization scheme was used to determine if physician and patient blinding could truly be maintained in the placebo surgery group."
Allocation concealment (selection bias)	Unclear risk	Sealed 'randomisation envelopes' were used, which were opened in the operating theatre, to reveal which procedure the participant was to receive. Insufficient information provided about whether appropriate safeguards were used (e.g. opaque, sequentially‐numbered envelopes)
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel (other than the operating surgeon) were blinded to the treatment assignment group. Identical preparation, stab wounds and post‐operative care were performed in all groups so it is unlikely that participants could guess their group assignment. "All postoperative care was performed by orthopaedic residents, nurses and other personnel who were blinded to the type of treatment that the patient received."
Blinding of outcome assessor Self‐reported outcomes	Low risk	Blinding of participants was done; low risk of bias in the measurement of pain and satisfaction with surgery
Blinding of outcome assessor Assessor‐reported outcome (knee replacement)	Low risk	Knee replacement not measured
Incomplete outcome data (attrition bias) All outcomes	High risk	1/2 (50%) participants withdrew from the arthroscopic debridement group and 0/5 withdrew from the placebo surgery group. The reason for withdrawal was not reported.
Selective reporting (reporting bias)	High risk	Trial registration not done and protocol not available. Data on mobility and general well‐being were reported as being collected in the methods but results not reported. Pain measured but outcome data not adequately reported (no measure of variance)
Other bias	Low risk	No other bias apparent