Drs. Lerner and Hulley mention only the null studies excluded, but most of the excluded studies favored garlic. Our objective was to study the effect of garlic on hypercholesterolemic patients, and we included trials in which most patients had cholesterol levels greater than 200 mg/dL. Most studies did not enroll patients based on cholesterol levels alone but rather on the presence of coronary artery disease, diabetes, hypertension, or hyperlipidemia of varying definitions. Thus, an arbitrary percentage cutoff was necessary. One study of 33 patients [1] showing no effect of garlic was excluded, but so was another randomized, placebo-controlled study (68 patients) that showed a 23% reduction in cholesterol levels that was associated with garlic [2].

We agree that including missing data would give more precise estimates. The study with null results had a short-duration, crossover design and no washout period [3]. As Dr. Hulley has stated, the crossover design is disadvantageous because of "an added complexity of analysis and interpretation created by the elusive problem of carryover effects" [4]. Therefore, one cannot assume that garlic had no effect unless data from the first period were reported. Again, a study showing a positive effect of garlic was also excluded for failure to meet this criterion [5].

We did a sensitivity analysis on noncontrolled trials to be sure that we did not overlook studies showing no effect and to examine the magnitude and consistency of garlic's effect. The cholesterol-lowering effect of garlic in these eight studies was homogeneous and seen in the garlic-treated arms of the primary analysis. By contrast, meta-analysis showed that placebo groups had a 4% decrease in serum cholesterol levels, a better estimate of "regression to the mean" or "placebo effect".

Because the quality of the included trials and their appearance in unfamilar journals also concerned us, we did a comprehensive, critical quality assessment and reported the results in both the body of the paper and in the appendix.

Millions of people already take garlic supplements for their alleged health benefits. On the basis of our rigorous, unbiased, quantitative review of the literature, we believe that garlic does indeed lower cholesterol levels.