To summarise the effects of garlic on cardiovascular-related factors, and to note any potential adverse effects.

Eleven databases were searched from January 1966 to July 1999 for English and non-English citations using the search terms provided in the paper. The databases searched included AMED, CISCOM, the Cochrane Library, EMBASE, MEDLINE and NAPRALERT. The search was updated on PubMed in February 2000. Additional studies were identified by examining the reference lists of pertinent articles, through symposia , and by contacting manufacturers and experts.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) studying the effect of garlic on cardiovascular outcomes, and lasting at least 4 weeks, were considered. For adverse effects, studies of any design that reported adverse clinical symptoms or events associated with garlic exposure were selected.

Specific interventions included in the review

Garlic was compared to placebo, no garlic or another active ingredient. The garlic group of patients received either dehydrated garlic preparations, standardised to the alliin content of the active powder; non-standardised garlic preparations; or garlic in combination with fish oil, hawthorn, soya lecithin, gingko biloba or other ingredients. The comparison group received placebo, non-garlic controls, antilipidaemic agents, antihypertensive agents or another garlic preparation. Concurrent interventions in 10 studies were low-fat, low-cholesterol, high-fibre diets.

Participants included in the review

The participants were healthy volunteers who had one or more of the following: Type 2 diabetes, elevated platelet aggregation, hyperlipidaemia, hypertension, ischaemic heart disease, history of myocardial infarction, or peripheral vascular disease. Participants were recruited from various sources including general practices, academic institutions, the community, specialty clinics and out-patient clinics. The reported age of participants ranged from 14 to 74 years. The participants were all male in five of the studies; the male to female ratio in eight of the studies was unclear; and the rest of the studies included both sexes, with the proportion of males ranging from 30 to 83%.

Outcomes assessed in the review

The review assessed antilipidaemic, antihypertensive, antiglycaemic and antithrombotic effects, effects on cardiovascular morbidity and mortality, and adverse effects.

How were decisions on the relevance of primary studies made?

Decisions were made by two independent reviewers, although it was not indicated whether they were blinded to the results and source of the articles. The authors do not state how differences were resolved.

The criteria used to assess validity were: details of the randomisation procedures, whether equivalence in baseline characteristics between randomised groups were reported, the adequacy of blinding, drop-out rates, and whether intention to treat analysis was used. The authors do not state how the papers were assessed for validity, or how many of the reviewers performed the validity assessment.

Two physicians independently extracted data from the RCTs, whilst one physician extracted data from studies of adverse effects. Any disagreements were resolved by consensus. There was no indication as to whether the physicians were blinded to source, author, journal, affiliations, year of publication or results. The authors indicated that no formal reliability tests of abstractions were made.

Data were extracted for the following categories: authors and year of publication; sample size; age and gender; recruitment setting; characteristics of the participants, e.g. cardiovascular risk factors present; adequacy of randomisation; interventions; duration of trial; end points, e.g. clinical symptoms, lipid levels, blood-pressure, glucose level and thrombotic state; and the percentage drop-out per group.

How were the studies combined?

Placebo-controlled randomised trials with lipid level outcomes were quantitatively pooled because they used similar control groups, and the outcomes were measured using similar parameters at similar follow-up times. The standardised mean differences, adjusted for baseline differences, were used as measures of effect size. Data were pooled using a random-effects model, both with and without outliers. A study evaluating a garlic and fish oil combination was not pooled because of the possible independent effect of fish oil on lipid levels.

How were differences between studies investigated?

Differences between the studies and potential outliers were investigated using Galbraith plots, a standard chi-squared test and funnel plots. Studies were considered outliers if the chi-squared test gave a p-value of less than 0.10 and/or they fell outside the Galbraith or funnel plot.

Subgroup analyses were performed for those trials that used similar dried standardised preparations of garlic, enrolled participants with hypercholesterolaemia, or were of a double-blind design.

Forty-five RCTs (n=2,987) were identified: 37 from scientific journals, 5 from scientific proceedings, 1 as a book chapter, 1 as a thesis and 1 as an abstract. Seventy-three additional studies reporting adverse events were identified.

Antilipidaemic effects.

All garlic preparations combined, compared with placebo: the total cholesterol was reduced by 0.19 mmol/L (95% confidence interval, CI: 0.03 , 0.34) after 4 to 6 weeks of therapy (14 studies), by 0.44 mmol/L (95% CI: 0.32 , 0.57) after 8 to 12 weeks (24 studies), and by 0.03 mmol/L (95% CI: 0.21, +0.28) after 20 to 24 weeks (6 studies).

Standardised dehydrated garlic preparations, compared with placebo: the total cholesterol was reduced by 0.26 mmol/L (95% CI: 0.08, 0.45) after 4 to 6 weeks (8 studies), by 0.50 mmol/L (95% CI: 0.34, 0.66) after 8 to 12 weeks (12 studies), and by 0.07 mmol/L (95% CI: 0.22, +0.37) after 20 to 24 weeks (3 studies).

At 8 to 12 weeks, triglyceride levels were reduced by 0.21 mmol/L (95% CI: 0.09, 0.34) with all types of garlic preparations (17 studies), and by 0.24 mmol/L (95% CI: 0.09, 0.38) with standardised dehydrated garlic tablets (13 studies).

At 8 to 12 weeks, low-density lipoprotein cholesterol levels were reduced by 0.16 mmol/L (95% CI: 0.02, 0.30) with all types of garlic preparations (13 studies), and by 0.17 mmol/L (95% CI: 0.0, 0.34) with standardised dehydrated garlic tablets (10 studies).

At 8 to 12 weeks, high-density lipoprotein cholesterol levels were reduced by 0.02 mmol/l (95% CI: 0.03, +0.07) with all types of garlic preparations (14 studies), and by 0.01 mmol/L (95% CI: 0.05, +0.06) with standardised dehydrated garlic tablets (10 studies).

Subgroup analyses of studies involving only hyperlipidaemic participants or using double-blind methods gave similar results.

An unblinded head-to-head comparison between a standard dehydrated garlic preparation and garlic oil found a statistically-significant reduction in the level of low-density lipoprotein cholesterol, in favour of the dehydrated preparation, after 4 months of treatment. The one trial comparing bezafibrate and a garlic preparation found no difference in lipid level outcomes after 3 months of treatment.

Antihypertensive effects.

Of the 30 trials measuring blood-pressure, only 3 focused on it as a primary outcome. Three trials demonstrated a significant reduction in diastolic blood-pressure of 2 to 7%, whilst one showed a reduction in systolic blood-pressure of approximately 3%. The results of studies comparing antihypertensive effects were not pooled because many of the studies did not present numerical data, different methods of blood-pressure measurement were used, and few studies had a priori hypotheses related to blood pressure.

Antiglycaemic effects.

Of the 12 trials assessing the effect of garlic on serum glucose levels, only one 4-week trial conducted on non-diabetic participants reported a statistically significantly greater reduction in glucose levels with standardised dehydrated garlic tablets, compared with placebo. There were no significant reductions in the glycosylated haemoglobin, serum insulin or C-peptide levels.

Antithrombotic effects.

Four of the 10 trials measuring the effects on spontaneous platelet aggregation reported modest, but significant decreases in platelet aggregation with garlic, compared with placebo. One trial reported significant decreases in epinephrine-induced, but not adenosine diphosphate-induced platelet aggregation. Mixed effects were reported for fibrinolytic activity and plasma viscosity.

Effects on cardiovascular morbidity and mortality.

Two trials assessed the improvement in pain-free walking distance for participants with lower extremity peripheral vascular disease. One trial showed no significant difference between the use of garlic and placebo, whilst the other reported a significant increase in pain-free walking distance when using the garlic oil macerate-soya lecithin-hawthorn oil-wheat germ oil combination.

An unpublished trial assessing reinfarction rates was reanalysed. The results showed there were no significant differences between participants randomised to garlic extract or placebo, in terms of total mortality or myocardial infarction.

Adverse effects.

Eight RCTs reported malodorous breath or body odour with garlic treatment, compared with placebo. Other adverse effects from case reports and case series included abdominal pain or fullness, anorexia, flatulence, dermatitis, rhinitis, Meniere's disease, asthma, myocardial infarction, bleeding, epidural haematoma, increase in international normalised ratio in patients taking warfarin, small intestinal obstruction and oesophageal pain.

The authors concluded that garlic has potentially small short-term benefits on lipid-lowering and antiplatelet factors, whereas it has insignificant effects on blood-pressure and no effect on glucose levels. The conclusions regarding clinical significance were limited by the marginal quality and short duration of many trials, and by the unpredictable release and inadequate definition of the active constituents in the preparations studied.

The objective of the review was stated clearly and the inclusion criteria were appropriate. The search strategy was comprehensive and included efforts to identify unpublished articles; in addition, there were no language restrictions. The assessments of validity and heterogeneity were performed adequately, although the results were not presented. When pooling the studies, the authors took into account the participants' baseline characteristics when summarising the effect sizes. However, it may have been inappropriate to pool the studies, even with adjustment for these baseline variables, if the participants were heterogeneous at baseline in terms of important prognostic factors.

There were no tests for publication bias. It would also have been useful if the results had reported the proportion of non-English publications. Information on the country of publication and participants' ethnic background would have helped in addressing the generalisability of the conclusions.

Allocation concealment has been shown to be an important indicator of trial quality and a description of this factor would have been useful. Also, given the questionable quality of many of the included studies, a sensitivity analysis on the basis of quality would have been informative. The heterogeneity seen in the meta-analyses should have been explored further, rather than simply excluding outliers from the analysis. Despite the limitations of the review, the conclusions drawn were appropriate.

Practice: The authors state that the studies were limited by unclear randomisation procedures, unclear adequacy of blinding and short duration. These, together with unclear definitions of the active ingredients in the garlic preparations studied, limit the clinical significance of the results.

Research: The authors recommend that the major active ingredients in garlic, and their mechanism of action, need to be determined before conducting more trials. The subsequent studies should include well-designed trials of longer duration, with a clear unbiased recruitment and selection process, adequate randomisation, blinding and intention-to-treat analysis.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.