Abstract 4422: Initial Antihypertensive Drugs for Heart Failure Prevention: Network and Bayesian Meta-analyses of Clinical Trial Data
Heart failure continues to be the most common admitting diagnosis for Medicare beneficiaries, and is the only cardiovascular endpoint that has continuously increased in incidence in the USA since 1990. Early meta-analyses suggested that active antihypertensive drug therapy decreased heart failure incidence in hypertensives by about 50%. We therefore gathered data from existing clinical trials (minimum follow-up: 1 year) to compare the ability of the various antihypertensive drug classes regarding prevention of heart failure. Heart failure was reported in 53 clinical trials involving 281,626 patients. These data were subjected to both network (Stat Med. 2002;21:2313–2324) and Bayesian (Stat Comp. 2000;10:325–337) meta-analyses. The numbers of subjects with heart failure/number at risk were: Placebo (1264/43,142), Diuretic (1093/35,271), β-blocker (993/52,291), Calcium Channel Blocker (CCB: 2145/76,594), ACE-inhibitor (ACE-I: 1734/48,247), and Angiotensin Receptor Blocker (ARB: 1098/26,141). The incoherence for the network meta-analysis was very small (ω < 0.000008), and the results (using diuretic as reference, 22 arms) were: These results were robust to many sensitivity analyses. After subdividing the trial results according to which diuretic was used initially (chlorthalidone, hydrochlorothiazide, or other), the incoherence was still very small (ω < 0.000005), but HCTZ was significantly better than placebo (only), whereas chlorthalidone was also significantly better than all other non-diuretic drugs. These data indicate that an initial diuretic is superior to placebo and all other active antihypertensive drug classes in preventing heart failure, with chlorthalidone exhibiting somewhat better prevention than HCTZ.