Twenty-five studies were included in the review, but only 11 RCTs (n=590) were included in the meta analysis (10 RCTs for systolic blood pressure and 11 RCTs for diastolic blood pressure). All studies reported adequate randomisation and double-blinding. Eight RCTs reported losses to follow-up that ranged from 9% to 20%; three RCTs reported no data on follow-up.
Garlic was more effective than placebo at reducing systolic blood pressure (WMD -4.56mmHg, 95% CI -7.36 to -1.77, p<0.001; 10 RCTs). There was evidence of statistical heterogeneity (p=0.01). Subgroup analysis of hypertensive participants also showed a greater reduction in the garlic group compared to placebo (WMD -8.38mmHg, 95% CI -11.13 to -5.62, p<0.001; four RCTs). There were no statistically significant differences between garlic and placebo groups for studies of normotensive participants (six RCTs). There was no evidence of heterogeneity for the subgroup analyses.
There were no statistically significant differences between garlic and placebo groups for mean diastolic blood pressure (11 RCTs). There was evidence of statistical heterogeneity (p<0.00001). Subgroup analysis of normotensive participants also showed no statistically significant differences between groups (eight RCTs). However, garlic was more effective than placebo at reducing diastolic blood pressure in hypertensive participants (WMD -7.27mmHg, 95% CI -8.77 to -5.76, p<0.001; three RCTs). There was no evidence of heterogeneity for the subgroup analyses.
Meta-regression analysis reported that systolic blood pressure or diastolic blood pressure at start of intervention was a significant predictor for heterogeneity (p=0.03 systolic and p=0.02 diastolic). No other variables showed significant associations with outcomes (data not reported). There was no evidence of publication bias. Other results were reported.