sidered in this study and matching our inclusion criteria were included in our work, further indicating that our metaanalysis does not carry any selection bias.With regard to the specific trials suggested by Nakao and colleaguesintheirletter,webelievethattheydonotfitoursearch and appropriateness criteria, and thus they were not considered. The purpose of our analysis, in fact, was to specifically analyze the impact of different antihypertensive therapies on the prevention of HF. In the HIJ-CREATE study, 4 only subjects with angiographic evidence of coronary artery disease were enrolled,resultinginaprevalenceofHFattheenrollmentaround the 20%, with no provided information about the hemodynamic status and the specific therapy of subjects with HF. In the CASE-J 5 and KYOTO-HEART study, 6 subjects with pre-existing HF werealsoenrolledandnoinformationabouttheirhemodynamic status and their specific treatment for HF were provided. In the CASE-J study, the prevalence of subjects with pre-existing HF in the 2 arms of treatment is not known.On the other hand, in the Jikey Heart study, 7 which was considered in our analysis, and which also included a proportion of subjects with baseline HF, these patients were enrolled only if they had been stabilized receiving a standard therapy for HF for at least 1 month. In addition, the percentage of baseline HF was exactly the same in the 2 arms of treatment.On the basis of these considerations, we still believe that the inclusion of these trials in our analysis, rather than their exclusion would have represented a potential source of bias. In any event, even including the results of these trials, this would lend negligible impact on the results of our meta-analysis.With regard to the letter of Elliott and colleagues, we appreciated the fact that the results and conclusions of our article are fully supported by their analysis. In their letter, these authors confirm that diuretics and angiotension-converting enzyme inhibitors are the most effective classes of antihypertensive drugs in the prevention of HF. The hypothesis that chlorthalidone is superior to the other diuretics is intriguing, although many factors should be taken into consideration before drawing final conclusions. First of all, the dosage of other diuretics, mostly thiazides, is hardly comparable to the dose of chlorthalidone in most of the studies. In addition, chlorthalidone was used in much larger and long-term studies. Indeed, the trend of a superior effect of chlorthalidone compared with "other diuretics" is only linked to indirect comparisons and influenced by the inclusion of the results of trials conducted more than 3 decades ago, which may not reflect current clinical hypertension management. Obviously, as meta-analyses generate hypotheses, it would be strongly desirable to perform a head-to-head comparison between chlorthalidone and other diuretics to clarify this issue. Our current study 1 cannot address this problem because we did not perform intraclass comparisons in view of the high risk to miss a sufficient...