TABLE 2.
Subgroup analysis for phytoestrogen intake and risk of T2D1
| Subgroups by study characteristics | Number of studies | Participants, n | T2D cases, n | RR (95% CI) | P-heterogeneity2 |
|---|---|---|---|---|---|
| Study population | |||||
| Only women | 7 | 183,919 | 8117 | 0.91 (0.83-0.99) | 0.65 |
| Men and women | 3 | 28,927 | 1604 | 0.88 (0.79, 0.98) | |
| Study design | |||||
| Nested case-control | 2 | 7950 | 425 | 0.97 (0.78, 1.20) | 0.45 |
| Cohort | 7 | 204,896 | 9296 | 0.89 (0.83, 0.96) | |
| Type of phytoestrogen | |||||
| Isoflavones | 5 | 106,006 | 3428 | 0.86 (0.77, 0.98) | 0.30 |
| Flavonoids | 4 | 106,840 | 6293 | 0.93 (0.86, 1.01) | |
| Genistein | 3 | 39,964 | 2232 | 0.87 (0.73, 1.05) | NA |
| Daidzein | 3 | 39,964 | 2232 | 0.89 (0.84, 0.95) | NA |
| Soy products | 3 | 72,997 | 1304 | 0.81 (0.68, 0.97) | NA |
| Location | |||||
| Asia | 4 | 73,997 | 1621 | 0.83 (0.87, 0.98) | 0.05 |
| Other | 5 | 138,849 | 8100 | 0.93 (0.88, 0.98) | |
| Difference between phytoestrogen intake in | |||||
| highest vs. lowest quantile | |||||
| Median or less (5.33-fold) | 5 | 123,123 | 3461 | 0.88 (0.78, 1.00) | 0.49 |
| Higher than median (5.33-fold) | 4 | 89,723 | 6260 | 0.92 (0.87, 0.999) | |
| BMI, kg/m2 | |||||
| Median or less (≤25.89) | 5 | 80,551 | 4146 | 0.91 (0.86, 0.97) | 0.70 |
| Higher than median (>25.89) | 4 | 132,295 | 5575 | 0.88 (0.78, 1.00) | |
| Age | |||||
| Median or less (≤53.87) | 5 | 133,736 | 4131 | 0.88 (0.79, 0.98) | 0.48 |
| Higher than median (>53.87) | 4 | 79,110 | 5590 | 0.92 (0.87, 0.98) | |
1“Study population” indicates studies conducted only in women: investigation performed only among female population, after excluding studies [Muller et al. (14), Zamora-Ros et al. (15) and Knekt et al. (52)] that reported overall results for male and female subjects but stated that they tested the interaction term with sex. “Study design” indicates that only prospective cohort and nested case-control studies were included. “Type of phytoestrogen”: “Soy products” estimates were pooled together for soy beans, soy milk, soy flour, and other soy products. “Location”: “Asia” (South Korea, Japan, and 2 studies from China; “Other”: Europe and 2 studies from the United States). NA, not applicable; T2D, type 2 diabetes.
2 P values for heterogeneity were evaluated by using random-effects meta-regression. P values were calculated between 2 or 3 groups that were considered to be a source of heterogeneity; the groups are indicated in the table (if >5 studies were included).