D'Anna 2007.

Methods	Design: parallel‐group Number randomly assigned: 265 (substudy of larger trial) Number dropped out: by year one, 10/125 in genistein group (after six months of follow‐up) and 8/122 in placebo group (four after three months and four after six months). Reasons given for dropping out were mainly gastrointestinal side effects but also presence of other diseases, loss to follow‐up and inadequate interaction with the woman's family doctor. None withdrew because of treatment inefficacy. By year two, 16/125 in genistein group and 13/122 in placebo group Number analysed: year one: 247; year two: 236 Intention‐to‐treat analysis: no Power calculation for sample size: yes: difference of at least 20% between groups required 97 participants in each group Duration: one and two years (two publications) Timing: not stated Location: two separate university centres in Italy Funding: Italian Ministry of Scientific Research and Technology and University of Messina, Italy
Participants	Inclusion criteria: 50 to 70 years old; postmenopausal; FSH > 50 IU/L; serum oestradiol < 100 pmol/L. In the substudy of 247 participants, only women with vasomotor symptoms were evaluated Exclusion criteria: clinical or laboratory abnormalities that suggested various disorders; coagulopathy; use of oral or transdermal oestrogen, progestin, androgen or other steroids; use of bisphosphonates, cholesterol‐lowering therapy or cardiovascular medications in previous six months; smoking > 10 cigarettes/d; BMD of the femoral neck > 0.795 g/cm3 Age, years: mean 53 in both groups Recruitment: women referred from university departments
Interventions	Phytoestrogen genistein twice per day (each tablet contained 27 mg of total isoflavone) Placebo twice per day
Outcomes	Number and severity of hot flushes Endometrial thickness
Notes	The study is a substudy of a larger study evaluating bone loss and CVD prevention. The substudy consisted of women with vasomotor symptoms, and characteristics of participants did not differ from those in the parent study
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Computerised database"
Allocation concealment (selection bias)	Unclear risk	Not reported
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	Unclear risk	Although dropouts were minimal at both one and two years, outcomes were measured in a subgroup of the total study population
Selective reporting (reporting bias)	Unclear risk	Side effects were presumably collected, as the authors stated that treatments were "well tolerated" but the results were not reported