Table 3. Subgroup analysis results in meta-analysis.
Heterogeneity | Quality of the evidence(GRADE) | |||||
Subgroup | No. study | WMD(95%CI) | P value | I2 (%) | P | |
Overall | 20 | -1.23 [-1.65, -0.81] | <0.001 | 93 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
BMI | ||||||
<25 | 6 | -0.79 [-1.28, -0.30] | <0.001 | 91 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
≥25 | 14 | -1.48 [-2.16, -0.80] | <0.001 | 94 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
Duration(months) | ||||||
<6 | 10 | -0.94 [-1.41, -0.47] | <0.001 | 87 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
≥6 | 10 | -1.47 [-2.16, -0.78] | <0.001 | 95 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
Dose(mg/day) | ||||||
<2000 | 14 | -1.21 [-1.32, -1.09] | <0.001 | 94 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
≥2000 | 6 | -0.55 [-0.71, -0.40] | <0.001 | 86 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
HOMA-IR ratio | ||||||
<0.735 | 4 | -1.07 [-1.76, -0.38] | <0.001 | 93 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
0.735-0.865 | 4 | -0.49 [-0.72, -0.25] | <0.001 | 7 | 0.360 | ⨁◯◯◯2 VERY LOW |
0.865-1.015 | 3 | -1.06 [-1.86, -0.26] | <0.001 | 23 | 0.270 | ⨁◯◯◯2 VERY LOW |
<1.015 | 4 | -1.32 [-1.56, -1.08] | <0.001 | 0 | 0.600 | ⨁◯◯◯2 VERY LOW |
NR | 5 | -1.98 [-3.59, -0.36] | 0.020 | 98 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
T ratio | ||||||
<0.714 | 4 | -1.19 [-1.67, -0.70] | <0.001 | 40 | 0.170 | ⨁◯◯◯1,2 VERY LOW |
0.714-0.825 | 2 | -1.85 [-2.09, -1.60] | <0.001 | 99 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
0.825-0.886 | 3 | -1.35 [-1.58, -1.11] | <0.001 | 84 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
<0.886 | 4 | -0.74 [-0.88, -0.60] | <0.001 | 91 | <0.001 | ⨁◯◯◯,2 VERY LOW |
NR | 7 | -0.52 [-0.73, -0.31] | <0.001 | 87 | <0.001 | ⨁◯◯◯1,2 VERY LOW |
Abbreviations: WMD, Weighted Mean Difference; HOMA-IR, homeostasis model assessment of insulin resistance; T ratio, Total testosterone ratio; NR, no reported.
1Downgraded by 1 level for substantial heterogeneity (I2 statistic is greater than 40%)
2Downgraded by two levels for the limitations of the included studies and lower total sample size