Garlic preparations and pressure: the largest estimate was not the strongest evidence
A review of nine trials found lower office pressure, but study quality changed the systolic estimate.

Illustrative food image, not a photograph of the study intervention.
What the review included
This review examined garlic preparations rather than a garlic-containing dinner. The authors searched for randomized trials comparing a garlic preparation with placebo in people with hypertension, requiring baseline and follow-up blood-pressure values and at least four weeks of follow-up. Nine double-blind trials with 482 participants qualified. Actual follow-up ranged from eight to 26 weeks, and all trials reported office measurements. That description sets the boundary of the evidence: a collection of relatively small preparation-versus-placebo trials. It does not directly test ordinary amounts of garlic in cooking, and it does not establish that different garlic products are interchangeable.
Rohner A, Ried K, Sobenin IA, Bucher HC et al.. American journal of hypertension, 2015. PMID 25239480 · DOI
The pooled numbers
Across the included trials, the weighted mean difference favored garlic preparations by 9.1 mmHg for systolic pressure, with a 95% confidence interval from 5.4 to 12.7 mmHg lower than placebo. The corresponding diastolic estimate was 3.8 mmHg lower, with an interval from 1.0 to 6.7 mmHg lower. These are striking pooled estimates, but the abstract immediately pairs them with substantial inconsistency. Heterogeneity was 71% for systolic and 80% for diastolic pressure, and statistically significant for both. Reporting the pooled reductions without that qualification would make a variable body of evidence look more uniform than the authors found it to be.
Rohner A, Ried K, Sobenin IA, Bucher HC et al.. American journal of hypertension, 2015. PMID 25239480 · DOI
Better methods changed the reading
The review did more than pool all nine trials together. Analyses were also restricted to higher-quality studies using intention-to-treat analysis, or to studies with concealed allocation and standardized, blinded pressure measurement. Under those restrictions, systolic effect sizes became smaller and heterogeneity disappeared. Diastolic effect sizes did not show the same reduction. This is a reason to resist selecting the largest systolic number as the definitive answer. The abstract describes most included trials as moderate quality and the studies as rather small. Its quality-sensitive analyses support a cautious interpretation instead of a confident claim that the overall estimate will reproduce in a better trial.
Rohner A, Ried K, Sobenin IA, Bucher HC et al.. American journal of hypertension, 2015. PMID 25239480 · DOI
Preparations are not a kitchen dose
The authors concluded that garlic preparations may lower blood pressure in people with hypertension, but that the evidence was not strong. They called for a well-conducted, adequately powered trial of longer duration. Their assessment is more restrained than a headline built only around a 9.1 mmHg systolic difference. The abstract does not establish an equivalent amount of fresh garlic for a recipe, nor does it compare a meal with medication. All included trials measured office pressure, with follow-up no longer than 26 weeks. The appropriate takeaway is a suggestive preparation-specific signal whose magnitude and reliability remain limited by trial quality and inconsistency.
Rohner A, Ried K, Sobenin IA, Bucher HC et al.. American journal of hypertension, 2015. PMID 25239480 · DOI