Brandt 2005.

Methods	Randomised controlled trial with 2 parallel groups Randomisation stratified by centre Trial duration: 30 months Multicentre trial including 6 centres No power calculation reported
Participants	431 participants with radiologically confirmed knee osteoarthritis were randomised Number of females: 431 (100%) Average age: 54.9 years Average BMI: 36.7 kg/m2 Severity of knee osteoarthritis: 59% with Kellgren/Lawrence grade 2 and 41% with Kellgren/Lawrence grade 3 Duration of knee complaints: not reported
Interventions	Experimental intervention: doxycycline, 100 mg twice daily Control intervention: placebo, twice daily Treatment duration: 30 months Analgesics other than study drugs allowed and intake was similar between groups
Outcomes	Extracted pain outcome: 15.2 m (50‐feet) walking pain after 30 months Extracted function outcome: WOMAC disability subscore after 30 months Primary outcome: joint space narrowing in the tibiofemoral compartment of the contralateral knee
Notes	ClinicalTrials.gov Identifier: NCT00000403 The trial was supported by the a non‐profit organisation (NIH grants R01‐AR‐43348, P60‐AR‐20582 and R01‐AR‐44370). It is unclear whether this trial received funding from a commercial body
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Subjects (...) were randomly assigned" Comment: no mention of the mechanism used for sequence generation
Allocation concealment (selection bias)	Unclear risk	Quote: "Patients were allocated randomly to treatment groups in blocks of 6" Comment: no mention of concealment of allocation
Blinding of patients described?	Low risk	Quote: "matched placebo" Comment: indistinguishable interventions and the description of a double‐blind phase implies blinding of participants
Blinding of physicians?	Low risk	Comment: clearly distinguished between single blind run‐in period and double‐blind phase. Blinding of physicians probable
Blinding of outcome assessors?	Low risk	Comment: depending on the outcome, participants or physicians were the assessors, both of which were blinded
Intention‐to‐treat analysis? All outcomes	High risk	Pain outcome: 69 of 218 participants (32%) excluded in experimental group and 55 of 213 participants (26%) excluded in control group Function outcome: 69 of 218 participants (32%) excluded in experimental group and 55 of 213 participants (26%) excluded in control group
Free of selective reporting?	High risk	Comment: there were 2 instruments (SF‐36 and Pain at rest) assessed during the trial but that results were not available in any of the trial reports. These instruments were identified in the trial registration and a multiple trial report (Mazzuca et al. 2004). Trial registration occurred after trial main report (Brandt et al. 2005) was published
Funding by commercial body avoided?	Unclear risk	No information provided regarding funding from commercial body