to the editor: The authors state in the article1 about alternative therapies for congestive heart failure and hypercholesterolemia that despite a large body of data showing Allium sativum (garlic) extracts to be effective for lowering serum total cholesterol levels, "more recent and rigorous studies have failed to substantiate these benefits."¹ They do not address the clinical fact that prior studies have used garlic extracts that contain allicin—which at least one of the recent negative trials was not studying (instead, it was evaluating garlic oil, known to contain minimal allicin).² This was thoroughly addressed in a letter replying to that study.³ This trial was also quite modest, involving only 25 patients, which is hardly more "rigorous" than prior research involving hundreds of patients.⁴ Another relatively recent study involved only children⁵ and it is difficult to compare such research with prior trials, all of which were conducted in adults. How can the authors state that garlic is “not efficacious” when negative trials are published in light of the much larger body of prior positive research?

The net total of the evidence that garlic lowers cholesterol is undoubtedly better than any other natural product known to me with the possible exception of soy protein and Trigonella foenum-graecum (fenugreek) seed, which the authors identify as effective based on published literature. This makes it difficult to understand how the authors can state that Monascus purpureus (red yeast rice) extracts and Commiphora mukul (guggul) extracts are more promising or effective than garlic extracts. The number of trials on these agents is much smaller than trials on garlic and are, generally speaking, much less rigorous (particularly in the case of guggul).

The authors do not comment on the possible anticoagulant side effects of garlic. There is at least one case report of multiple hemorrhages occurring in a patient taking garlic extracts.⁶ Although one case of hemorrhage among millions, if not billions, of doses taken is hardly reason for major concern, the possibility should at least be mentioned. Furthermore, the theoretical potential for hemorrhage when garlic extracts are combined with warfarin, heparin, aspirin, or other anticoagulant drugs could also have been mentioned.

in reply: Dr Yarnell's letter once again reminds us how difficult it is to critically interpret the medial literature and how often we face the task of analyzing studies that reach contradictory conclusions. Such contradictory results can be attributed to poor design (lack of power, inclusion in meta-analysis of poorly designed studies, etc.), improper data analysis, as well as publication bias and other types of bias.

Such is the case with garlic and its purported ability to lower lipid levels in humans. Indeed, the 1993 meta-analysis quoted by Dr. Yarnell did find garlic to be beneficial as a lipid-lowering agent; however, a subsequent 1994 letter published in the Annals of Internal Medicine clearly describes the shortcomings of this research.¹ It points out flaws in methodology, failure to include well-designed studies that showed no change in serum lipids with garlic supplements, inadequate sensitivity analysis and, finally, that two (of only five) "original" studies included in this meta-analysis were sponsored by garlic supplement manufacturers.

A 1996 letter in the Lancet also addresses this 1993 meta-analysis, as well as two more recent 1994 and 1996 meta-analysis.² This letter concludes, "at this stage the evidence does not support the use of garlic powder for cardioprotection." The letter also states that publication bias is a strong possibility.

The most recent meta-analysis concludes that garlic has a modest effect on lowering total cholesterol, but it neither lowers low-density lipoprotein (LDL) nor increases high-density lipoprotein (HDL) levels, although these results lack power.3 The 4 to 6 percent reduction in cholesterol in this meta-analysis is the same as could be expected from dietary measures alone. It concludes that the use of garlic "is not an efficient way to decrease total serum cholesterol and its use may not be clinically meaningful."³

Results from the two most recent, randomized clinical trials (which included 101 moderately hypercholesterolemic adults) further support our stance that garlic is not an effective cholesterol-lowering agent.^4,5 The data did not demonstrate any change in total cholesterol, LDL or HDL levels with garlic tablets or powder supplementation after three months.

Dr. Yarnell states that we should have mentioned the possible side effects of garlic supplements. Because we were not recommending the use of these supplements, we believed this was unnecessary.

We realize how problematic it is to critically steer through the sea of medical literature, especially when conflicting studies and personal biases are concerned. We must resist the urge to "cite scripture for our purpose" and instead be as discriminating as possible in making recommendations that affect our patients’ health and pocketbooks.

Should future evidence demonstrate garlic's benefit in lowering cholesterol, we would be happy to revise our position. We would certainly like to endorse an effective, "natural," and less expensive means of lowering cholesterol. Unfortunately, the current literature does not find that garlic fits this bill. We thank Dr. Yarnell for promoting this discussion and highlighting the dangers of a partial interpretation of the medical literature.