Campagnoli 2005.

Methods	Design: cross‐over Number of women screened: not stated Number randomly assigned: 36 Number analysed: 29 Number dropped out: seven (not clear which group: three medical reasons, four family reasons) Intention‐to‐treat: no Power calculation: yes: 95% power to detect at least a 20% greater reduction in hot flushes in active arm compared with placebo Duration: 12 + 12 weeks Timing: November 1999 to December 2000 Location: hospital in Torino, Italy Funding: Medestea International (manufacturer of active treatment)
Participants	Inclusion criteria: minimum of five moderate to severe hot flushes/d, good general health, 45 to 58 years, BMI 18 to 28, surgical menopause (bilateral oophorectomy for at least three months or in spontaneous menopause with no menses for over six months), menopausal hormone profile (oestradiol < 30 pg/mL, FSH > 40 UI/L) Exclusion criteria: use of drugs that influence vasomotor symptoms, hormone therapy or tibolone in previous six months, consumption of soy‐based food more than once per week, use of drugs that might reduce absorption of isoflavones Mean age of completers, years: 51 Recruitment method: menopause clinic
Interventions	Standardised soy extract 200 mg (Soy select) capsules (60 mg/d isoflavones) Placebo capsules Dose, duration and timing of administration: two capsules per day in two doses for 12 weeks, then switched to alternate treatment without a washout period
Outcomes	Number of hot flushes per week after treatment (at end of first phase of study)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Schedule prepared by the manufacturer of the product using computer‐generated randomisation list and distributed sequentially
Allocation concealment (selection bias)	Low risk	Adequate: investigators blind to treatment allocation
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	High risk	High dropout
Selective reporting (reporting bias)	Low risk	All potential outcomes reported