FOSTAR 2016

Methods	Fish Oil in knee OSTeoARthritis (FOSTAR) RCT, parallel, (n‐3 EPA + DHA vs low n‐3), 24 months Summary risk of bias: low
Participants	Adults aged 40+ years with knee osteoarthritis N: 101 intervention, 101 control Level of risk for CVD: low Men: 41% intervention, 60% control Mean age in years (SD): 60.8 (10) intervention, 61.1 (10) control Age range: > 40 Smokers: not reported Hypertension: not reported Medications taken by at least 50% of those in the control group: none reported Medications taken by 20%‐49% of those in the control group: not reported at baseline, but 'during' includes Vit. D ˜ 32% Medications taken by some, but less than 20% of the control group: not reported at baseline, but 'during' includes Glucocorticoid, HRT/anti‐resorptive, both ˜ 10% Location: Australia Ethnicity: not reported
Interventions	Type: supplementary food (enriched orange juice) Comparison: high EPA + DHA vs low EPA + DHA plus ALA (replacement unclear, but low omega 3) Intervention: 1‐3 × a day drink of fruit juice mixed with day total = 15 mL of fish oil supplement (18% EPA, 12% DHA, 4.5 g/day total omega 3). Dose: 4.5 g/d EPA + DHA Control: liquid oral oil 15 mL sunola oil/day (which contains fish oil 2 mL plus 13 mL canola oil) (total omega‐3 fat: ≥ 0.45 g EPA + DHA from 15 mL) Compliance: assessed by measuring the oil volume in returned bottles, compliance was > 80% in both groups. Both groups had increases from baseline in plasma EPA and DHA with the high‐dose group having substantially larger increases, consistent with compliance with study oil Length of intervention: 24 months
Outcomes	Main study outcome: change in pain scale of WOMAC index Dropouts: 18 intervention, 16 control Available outcomes: mortality, CVD events, adverse events, analgesic use, bone marrow density, weight gain and serum fatty acids Response to contact: yes
Notes	Data on quality of life and pain score are presented in a figure and not in a usable format Study funding: government funding
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random allocation sequence
Allocation concealment (selection bias)	Low risk	A security‐protected central automated allocation procedure was used to allocate participants to one of the 2 treatment arms. This was performed centrally at one pharmacy and then used to allocate and administer the oil at each site
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Citrus flavouring was added to both oils to achieve comparable taste and optimise masking. Both were provided in identical dark 500‐mL bottles with similar labelling. At the end of the study, 52% of participants were unsure which group to which they had been allocated (50% high dose, 50% low dose). Of the remaining who thought they knew which group they were allocated, only 57% answered correctly, suggesting that blinding had been well maintained
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Participants and staff involved in patient care and assessment of BMD remained blinded throughout the study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Oil intolerance in 1st year differed, others appear similar, but numbers confused
Selective reporting (reporting bias)	High risk	Prospectively registered August 2007, recruitment started July 2007, outcomes published 2016. Variety of outcomes such as quality of life stated in trials registry but not published.
Attention	Low risk	Same contact and instruction schedule for all participants.
Compliance	Low risk	Assessed by measuring the oil volume in returned bottles, compliance was > 80% in both groups. Both groups had increases from baseline in plasma EPA and DHA with the high‐dose group having substantially larger increases, consistent with compliance with study oil
Other bias	Low risk	None noted