van de Weijer 2002.

Methods	Design: parallel‐group Number screened for inclusion: 42 (six ineligible, 24 did not return to clinic, tqo recorded inadequate data in diary during screening phase) Number randomly assigned: 30 Number dropped out: six (three in each group, mainly because of lack of efficacy) Number lost to follow‐up: nil Number analysed: 26 Intention‐to‐treat analysis: no Power calculation: not stated Duration: 12 weeks Timing: not stated Location: university clinic, The Netherlands Funding: industry funded (Novogen Ltd, Australia)
Participants	Inclusion criteria: postmenopausal women 49 to 65 years of age with at least 12 months' amenorrhoea, average of > five hot flushes daily Exclusion criteria: HT or antibiotics within 12 weeks of study entry, undiagnosed vaginal bleeding, active liver or renal disease, history of allergy to foodstuffs, cardiovascular disease or thromboembolism Age, years: active arm 52.5 (SD 5.2), placebo 54.2 (SD 7.4) Recruitment method: not stated
Interventions	Phytoestrogen: isoflavones Formulation: Promensil 40‐mg tablets (daidzein, genistein, biochanin, formononetin) Placebo tablets Dose, duration and timing of administration: two tablets each morning for 12 weeks Participants given list of foods to avoid, including legumes and isoflavone supplements
Outcomes	Menopausal symptoms: daily diary for number of hot flushes, list of 21 symptoms to score on 4‐point scale Measures of change in quality of life
Notes	Baseline hot flush count = average count of last seven days from four‐week screening phase
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Blank envelopes containing allocation shuffled, numbered consecutively, then given consecutively to participants
Allocation concealment (selection bias)	Low risk	Participants took envelope to pharmacy, where number in envelope was matched with batch number of medication
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Both participants and researchers blinded
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Both participants and researchers blinded
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Attrition > 10% and unbalanced between groups
Selective reporting (reporting bias)	High risk	Adverse events not reported