Table 3.
Subgroup analysis of pooled data from the 21 studies included
| Subgroup parameter | Pooled intervention group(n) | Pooled control group(n) | Mean difference | P-value | 95% CI | Heterogeneity (I2) | p-value for heterogeneity |
|---|---|---|---|---|---|---|---|
| Fasting glucose levels | 335 | 345 | 3.29 mg/dl | 0.09 | -0.53, 7.11 | 92 | <0.00001 |
| Hb1Ac | 225 | 241 | 0.04% | 0.45 | -0.06, 0.13 | 37 | 0.12 |
| BMI change | 233 | 251 | -0.69 kg/m2 | 0.02 | -1.27, -0.12 | 0 | 0.68 |
| Insulin levels | 304 | 319 | -0.01mIU/l | 0.97 | -0.40, 0.38 | 0 | 0.47 |
| SBP | 471 | 465 | -0.30 mm Hg | 0.68 | -1.76, 1.15 | 13 | 0.3 |
| DBP | 471 | 465 | -1.05 mm Hg | 0.06 | -2.15, 0.04 | 35 | 0.08 |
| Oxidized LDL | 98 | 96 | -0.29 U/l | 0.93 | -6.99, 6.41 | 54 | 0.07 |
Legend: A negative mean difference suggests an improvement in the intervention group over the control group. Several other subgroup analyses were carried out, but on measurements other than main lipid measurements. The analysis results can be found in Appendix A4.