Knight 1999.

Methods	Design: parallel‐group Number screened for inclusion: not stated Number randomly assigned: 37 initially. When one woman dropped out after randomisation and before commencing treatment, another was recruited to the same treatment Number dropped out: two in high‐dose group withdrawn on GP advice Number lost to follow‐up: Number analysed: 35 Intention‐to‐treat analysis: no Power calculation: not reported Duration: 12 weeks Location: hospital outpatient clinic, Sydney, Australia Funding: industry funded (Novogen, Australia)
Participants	Inclusion criteria: postmenopausal women (bilateral oophorectomy or amenorrhoea for least six months with typical symptoms of menopause, FSH > 40 IU/L), having at least three hot flushes daily, 40 to 65 years of age Exclusion criteria: HT use within previous six weeks, allergy to isoflavones, current bowel, liver or gallbladder disease, diabetes requiring medication, malignancy other than skin cancers, contraindications to HT use, vegetarians, regular soy product users, taking liver enzyme–inducing medications Age, years: high dose 51.1 (± 8.8), medium dose 54.5 (± 4.4), placebo 53.1 (± 2.5) Recruitment method: through university department of obstetrics and gynaecology
Interventions	Phytoestrogen: isoflavones: one Promensil tablet plus three placebo tablets Phytoestrogen: isoflavones: four Promensil tablets Formulation: Each Promensil tablet contained 40 mg total isoflavones (genistein 4.0 mg, daidzein 3.5 mg and their methylated precursors biochanin 24.5 mg and formonetin 8.0 mg) Placebo four tablets identical to active tablets Dose, duration and timing of administration: four tablets daily (packed in individual sachets) for 12 weeks Participants advised not to alter their usual diet during the study
Outcomes	Menopausal symptoms: daily flush diary Quality of life: Greene Menopause Scale Compliance: pill counts
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Randomisation performed remotely by external statistician
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Stated as double‐blind
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Stated as double‐blind
Incomplete outcome data (attrition bias) All outcomes	Low risk	Minimal dropout
Selective reporting (reporting bias)	High risk	Adverse events not reported