Radhakrishnan 2009.

Methods	Design: parallel‐group Number randomly assigned: 100 Number dropped out: 15 (six in soy group and nine in control group) Number analysed: 85 Intention‐to‐treat analysis: no Power calculation: yes, 7% decrease in cholesterol with 80% power and 20% withdrawal rate Duration: six months Timing: not stated Location: Department of Obstetrics and Gynecology, University College of Medical Sciences and Guru Teg Bahadur Hospital, Delhi, India Funding: Dupont Protein Technology International–manufactured soy and protein supplements
Participants	Inclusion: last menstruation at least 12 months previous or six weeks since bilateral oophorectomy; FSH level 40 mlU/mL; unwillingness to take or intolerance to HT; not currently taking lipid‐lowering drugs, antidiabetic medications or herbal supplements; discontinued HT more than three months previously Exclusion: unexplained vaginal bleeding, hypertension, diabetes, liver dysfunction, renal or cardiac disease, active thromboembolic disease, deep vein thrombosis, coronary artery disease, cerebrovascular accident or past history of thromboembolic disease associated with oestrogen use; present or past oestrogen‐dependent malignancy such as breast or endometrial carcinomas, known peanut/legume allergy Mean age, years: 48.07 ± 5.44 in the soy group; 49.71 ± 7.27 in the placebo group Recruitment method: from outpatient clinic
Interventions	Soy group (n = 50): sachets containing 25 g of isoflavone‐rich soy protein isolate with 75 mg of isoflavones in powder form, sweetened with aspartase with mild vanilla flavour Placebo group (n = 50): sachets containing 25 mg of milk protein, which looked and tasted identical to the soy supplement Both sachets contained equal amounts of elemental calcium (900 mg) and other trace elements and vitamins Followed up at baseline and at three and six months
Outcomes	Kupperman Menopausal Index Karyopyknotic Index Maturation value Endometrial thickness Laboratory investigations Bone mineral density Acceptability of treatment
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomly assigned using computer‐generated randomised numbers in blocks of 10
Allocation concealment (selection bias)	Low risk	Precoded sachets were provided by DUPONT Protein Technology International
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	15 stopped the trial prematurely during the initial two months: six in the study group and nine in the control group. Main reasons were gastrointestinal side effects and food intolerance. Three cases in each group were lost to follow‐up
Selective reporting (reporting bias)	Low risk	Original protocol not viewed. Outcomes listed in the methods sections were reported in the results