TABLE 1.
Subgroup analysis for RCTs included in our review1
| Subgroups by study characteristics | Number of studies | Mean difference (95% CI)2 | I 2 for heterogeneity,3 % | P-heterogeneity4 |
|---|---|---|---|---|
| Phytoestrogen use and mean serum glucose change5 | ||||
| Study population | ||||
| Healthy women | 12 | −0.1 (−0.19, −0.01) | 97.3 | |
| Other women | 5 | −0.2 (−0.54, 0.13) | 96.1 | |
| Median number of participants | ||||
| ≤60 | 8 | −0.14 (−0.3, 0.02) | 93.2 | 0.28 |
| >60 | 9 | −0.1 (0.21, −0.003) | 97.9 | |
| Menopausal status | ||||
| Postmenopausal | 14 | −0.07 (−0.15, 0.01) | 97.2 | 0.17 |
| Other | 3 | −0.44 (−0.94, 0.07) | 91.5 | |
| Age, y | ||||
| ≤55.15 | 9 | −0.06 (−0.18, 0.07) | 96.6 | 0.66 |
| >55.15 | 8 | −0.18 (−0.31, −0.05) | 97.6 | |
| BMI, kg/m2 | ||||
| ≤25.77 | 9 | −0.07 (−0.18, 0.05) | 97.5 | 0.44 |
| >25.77 | 7 | −0.23 (−0.43, −0.02) | 97.1 | |
| Route of administration | ||||
| Tablet/capsule | 14 | −0.07 (−0.15, 0.01) | 97.2 | 0.17 |
| Other | 3 | −0.44 (−0.94, 0.07) | 91.5 | |
| Dosage, mg/d | ||||
| ≤86 | 9 | −0.17 (−0.36, 0.02) | 96.8 | 0.29 |
| >86 | 7 | −0.02 (−0.11, 0.07) | 97.2 | |
| Other | 1 | NA | NA | |
| Intervention duration, wk | ||||
| ≤16 | 7 | −0.09 (−0.24, 0.07) | 96.2 | 0.42 |
| >16 | 10 | −0.15 (−0.24, −0.06) | 96.7 | |
| Location | ||||
| Asia | 6 | −0.07 (−0.15, 0.01) | 90.2 | 0.81 |
| Other | 11 | −0.15 (−0.28, −0.02) | 97.9 | |
| Risk of bias | ||||
| High | 8 | −0.33 (−0.5, −0.16) | 97.1 | 0.62 |
| Low to medium | 9 | 0.03 (−0.06, 0.12) | 96.2 | |
| Design | ||||
| RCT crossover | 4 | 0.01 (−0.31, 0.03) | 94.8 | 0.33 |
| RCT | 13 | −0.15 (−0.26, −0.05) | 97.4 | |
| Phytoestrogen use and mean serum insulin change6 | ||||
| Study population | ||||
| Healthy women | 10 | 1.51 (−2.7, 5.73) | 89.1 | 0.64 |
| Other women | 5 | −6.87 (−16.13, 2.39) | 95.7 | |
| Median number of participants | ||||
| ≤50 | 6 | −4.63 (−11.17, 1.92) | 94.8 | 0.68 |
| >50 | 9 | 1.51 (−44.08, 7.11) | 90.2 | |
| Menopausal status | ||||
| Menopausal | 12 | 0.98 (−2.76, 4.71) | 92.6 | 0.20 |
| Other | 3 | −12.77 (−25.8, 0.25) | 74.3 | |
| Age, y | ||||
| ≤56.61 | 8 | −2.31 (−10.14, 5.51) | 92.1 | 0.33 |
| >56.61 | 7 | −0.28 (−4.85, 4.29) | 93.3 | |
| BMI, kg/m2 | ||||
| ≤26.25 | 8 | 1.51 (−1.95, 4.97) | 90 | 0.13 |
| >26.25 | 7 | −5.36 (−15.71, 4.98) | 91 | |
| Route of administration | ||||
| Tablet/capsule | 9 | 2.51 (−4.15, 9.17) | 91.5 | 0.23 |
| Other | 6 | −5.24 (−10.31, −0.17) | 93.7 | |
| Dosage, mg/d | ||||
| ≤78 | 8 | 0.94 (−6.77, 8.66) | 93.2 | 0.44 |
| >78 | 6 | −1.45 (−5.59, 2.69) | 90.3 | |
| Other | 1 | NA | NA | |
| Intervention duration, wk | ||||
| ≤12 | 8 | −2.32 (−8.82, 4.18) | 93.2 | 0.33 |
| >12 | 7 | 0.42 (−4.7, 5.54) | 91.8 | |
| Location | ||||
| Asia | 4 | −1.85 (−8.69, 4.99) | 94.1 | 0.88 |
| Other | 11 | −0.86 (−6.12, 4.41) | 92.1 | |
| Risk of bias | ||||
| High | 6 | −11.95 (−20.54, −3.35) | 88 | 0.19 |
| Low to medium | 9 | 3.91 (1.54, 6.27) | 76.2 | |
| Design | ||||
| RCT crossover | 4 | −1.02 (−8.2, 6.17) | 91.2 | 0.29 |
| RCT | 11 | −1.13 (−6.43, 4.18) | 92.9 | |
| Phytoestrogen use and mean HOMA-IR change | ||||
| Study population | ||||
| Healthy women | 4 | −0.61 (−1.2, −0.02) | 97.3 | 0.27 |
| Other women | 7 | −0.08 (−0.35, 0.20) | 97.8 | |
| Median number of participants | ||||
| <60 | 6 | −0.49 (−1.05, 0.06) | 96.5 | 0.11 |
| ≥60 | 5 | −0.02 (−0.18, 0.13) | 94.7 | |
| Menopausal status | ||||
| Menopausal | 8 | −0.14 (−0.37, 0.09) | 97.9 | 0.15 |
| Other | 3 | −0.62 (−1.41, 0.17) | 91.5 | |
| Age, y | ||||
| ≤60 | 6 | −0.38 (−0.85, 0.08) | 97.9 | 0.17 |
| >60 | 5 | −0.13 (−0.38, 0.11) | 96.4 | |
| BMI, kg/m2 | ||||
| ≤27.3 | 5 | −0.09 (−0.38, 0.2) | 98.4 | 0.09 |
| >27.3 | 6 | −0.44 (−0.9, 0.02) | 96.1 | |
| Route of administration | ||||
| Tablet/capsule | 5 | −0.16 (−0.371, 0.39) | 98.2 | 0.20 |
| Other | 6 | −0.28 (−0.5, −0.05) | 96 | |
| Dosage, mg/d | ||||
| ≤93 | 5 | −0.51 (−1.1, 0.08) | 98 | 0.35 |
| >93 | 5 | 0.02 (−0.17, 0.22) | 95.4 | |
| Other | 1 | NA | NA | |
| Intervention duration, wk | ||||
| <16 | 5 | −0.43 (−0.82, −0.03) | 96.5 | 0.18 |
| ≥16 | 6 | −0.11 (−0.44, 0.21) | 98.2 | |
| Location | ||||
| Asia | 3 | −0.12 (−0.48, 0.24) | 97.8 | 0.34 |
| Other | 8 | −0.32 (−0.65, 0.02) | 97.1 | |
| Risk of bias | ||||
| High | 6 | −0.62 (−1.07, −0.17) | 91.6 | 0.10 |
| Low to medium | 5 | 0.1 (0.02, 0.18) | 75 | |
| Design | ||||
| RCT crossover | 3 | −0.17 (−0.66, 0.33) | 93.2 | 0.09 |
| RCT | 8 | −0.3 (−0.64, 0.04) | 97.9 | |
1“Healthy women” are considered premenopausal or postmenopausal women included in an RCT; “Other women” are women with metabolic syndrome, glucose intolerance, or unrecognized diabetes (without antidiabetic medications); women treated for breast cancer in the previous 6 mo; and osteopenic and obese women. “Median number of participants” was calculated separately for each outcome. “Menopausal status” indicates postmenopausal women vs. adult women (above the age of 18 y). “Median age” was calculated separately for each outcome. “Median BMI” was calculated separately for each outcome [Liu et at. (45) did not report BMI]. “Route of administration” includes the use of tablets/capsules and other routes of administration (shake, powder, flower). “Dosage” indicates the mean dosage based on all included RCTs was 80 mg/d. “Intervention duration” indicates that the median intervention duration was calculated on the basis of all included RCTs; we compared RCTs with durations of intervention of ≤14 and >14 wk. “Location” refers to study location; studies done in Asia vs. studies done in Europe, America, and Australia. “Risk of bias” means that studies are considered to be low risk of bias if allocation concealment, blinding of participants, and outcome assessors were all coded “yes”; if a compliance assessment was done; and the number of dropouts and reasons for dropout were reported. In the case that ≥3 quality criteria were not met, the study was classified as having a high risk of bias; others were classified as having a moderate risk of bias. “Study design” refers to RCT vs. RCT crossover design. NA, not applicable; RCT, randomized controlled trial.
2Mean difference refers to mean difference of changes between treatment groups in serum glucose, insulin, and HOMA-IR (subjects using phytoestrogens as compared with subjects from control/placebo group).
3 l 2 describes the percentage of variation across studies that is due to heterogeneity rather than chance.
4 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 source of heterogeneity; the groups are indicated in the table.
5Glucose in mmol/L.
6Insulin in pmol/L.