Network Meta-analysis of Heart Failure Prevention by Antihypertensive Drugs

We agree with the conclusions of the “network meta-analysis” of heart failure prevention by antihypertensive drugs, which used a Bayesian technique and 26 clinical trials.¹ We have reported similar results supporting the superiority of a diuretic using both the network meta-analytic technique of Lumley,² and a Bayesian technique, which included data from all 47 or 53 of the then published hypertension clinical trials.^3- 4 We noted the great preponderance of evidence favoring chlorthalidone, which accounts for more than 90% of their heart failure cases. Although a network meta-analysis in 2004 suggested no significant difference between chlorthalidone vs another diuretic as initial therapy to prevent heart failure,⁵ data from all 61 hypertension trials involving 355 225 subjects (from the first Veterans Administration trial⁶ to the very recent Valsartan Amlodipine Randomized Trial⁷), using the therapeutic categories of Sciarretta et al¹ suggests a major difference in precision among diuretics (Figure). “Other diuretics” has the smallest number of heart failure cases of any treatment, nearly 4-times smaller than that for chlorthalidone. The P value for chlorthalidone is nearly 5 orders of magnitude smaller than that of “other diuretics.” Our results were robust to many sensitivity analyses, such as when combining “conventional therapy” and “β-blocker,” including only studies that randomized subjects to initial therapy, omitting data from studies done in nonhypertensive patients, and omitting data from the Antihypertensive and Lipid-Lowering to prevent Heart Attack Trial (ALLHAT). We add to the conclusions of Sciaretta et al¹ that the clinical trial evidence is strongest, by far, for chlorthalidone as the most effective antihypertensive agent to prevent heart failure.