Forty-seven RCTs were included (n=1,813 analysed): 32 parallel studies and 15 cross-over studies. The studies included 11 studies of premenopausal women (n=579 analysed), 35 studies of postmenopausal women (n=1,165 analysed) and one study of perimenopausal women (n=69 analysed). Sample size ranged from 10 to 304 women analysed. Intervention duration ranged from four to 104 weeks; 29 studies lasted four to 12 weeks and only two studies lasted more than one year.
Study quality varied. All except one were randomised. Thirty-five studies blinded patients, 31 blinded outcome assessors, nine assessed compliance, 20 reported commercial funding, 42 reported an adequate duration of intervention and 25 fully reported drop-outs. Ten studies were classified as being at low risk of bias.
Primary outcomes for premenopausal women: There was no significant difference between soy/isoflavone and control in circulating total oestradiol, oestrone or sex hormone binding globulin (based on six to 11 comparisons per analysis). No significant heterogeneity was found. Findings were similar after excluding crossover studies and using only data from the luteal phase.
Primary outcomes for postmenopausal women: There was a non-statistically significant increase in circulating total oestradiol associated with soy/isoflavone groups (SMD 0.13, 95% CI -0.01 to 0.27; 21 studies); there was no significant difference when mean differences were used. The funnel plot suggested some evidence of publication bias. There was no significant difference between soy/isoflavone and control in circulating total oestrones (seven studies) or sex hormone binding globulin (17 studies). No significant heterogeneity was found.
Secondary outcomes for premenopausal women: Soy/isoflavones were associated with a statistically significant reduction in follicle stimulating hormone (SMD -0.45, SE 0.1735) and luteinizing hormone (SMD -0.34, SE 0.1709) and a significant increase in menstrual cycle length (MD 1.05, SE 0.4694).
Secondary outcomes for postmenopausal women: There was no significant difference between soy/isoflavone and control in follicle stimulating hormone or luteinizing hormone.
Results for other secondary outcomes and subgroup analyses were reported.
One author had links with a manufacturer of soy/isoflavone products.