Zhong · Wei sheng yan jiu = Journal of hygiene research 2012 · meta-analysis of observational studies · n=28 studies

[Soy food intake and breast cancer risk: a meta-analysis].

Cited 17 times in the scientific literature.

Level 3 - non-randomized controlled study

Systematic review and meta-analysis of observational studies (case-control and cohort designs)

PubMed 23057338 · record verified 2026-08-28

What was done

Authors searched MEDLINE, PubMed, EMBASE, Sci, and CNKI databases for epidemiological studies published between 1991 and April 2011 evaluating soy food intake and breast cancer risk. A meta-analysis was performed using 28 selected studies (23 case-control and 5 cohort studies) to calculate pooled odds ratios (OR) and 95% confidence intervals (CI), with subgroup analyses stratified by study design, geography (Asian vs. Western countries), and menopausal status.

What was found

Overall soy intake was inversely associated with breast cancer risk across all 28 studies (pooled OR 0.86, 95% CI 0.78–0.94). By design, the protective association was present in case-control studies (pooled OR 0.86, 95% CI 0.77–0.95), but not in overall cohort studies (numerical estimate omitted in abstract). By geography, soy intake was protective in Asian populations (OR 0.76, 95% CI 0.67–0.87), but not in Western populations (OR 1.01, 95% CI 0.93–1.09). In Asian-specific stratified analyses, protection was observed in both case-control studies (OR 0.76, 95% CI 0.65–0.89; 14 studies) and cohort studies (OR 0.81, 95% CI 0.70–0.92; 3 studies). A protective effect was found among postmenopausal women but not premenopausal women, though specific numbers were not reported in the abstract for menopausal subgroups.

Why it matters

This review shows that the inverse association between soy intake and breast cancer risk is largely confined to Asian populations and postmenopausal women, helping explain inconsistencies across previous global studies.

Limits

The meta-analysis is heavily dominated by retrospective case-control studies (23 of 28), which carry inherent risks of recall and selection bias. Total participant numbers, soy intake measurement definitions, and numerical effect sizes for menopausal subgroups and overall cohort studies were not reported in the abstract.

Cited by