Background In a meta-analysis published in October 2001, we reported that new and old classes of antihypertensive drugs had similar long-term efficacy and safety. Furthermore, we observed that in clinical trials in hypertensive or high-risk patients gradients in systolic pressure accounted for most differences in outcome. Objective To test whether our previous conclusions would hold, we updated our quantitative overview with new information from 14 clinical trials presented before 1 March 2003. Methods To compare new and old antihypertensive drugs, we computed pooled odds ratios from stratified 2 x 2 contingency tables. If Zelen's test of heterogeneity was significant, we used a random effects model. In a meta-regression analysis, we correlated odds ratios with corresponding between-group differences in systolic pressure. We then contrasted observed odds ratios with those predicted from gradients in systolic pressure. Main outcomes Differences in achieved systolic blood pressure and incidence of total and cardiovascular mortality, cardiovascular events, stroke, myocardial infarction and heart failure. New versus old drugs In 15 trials, 120574 hypertensive patients were randomized to old drugs (diuretics or beta-blockers) or new agents [calcium-channel blockers, alpha-blockers, angiotensin-converting enzyme (ACE) inhibitors or angiotensin type-1 receptor (AR1) blockers]. Old and new drugs provided similar protection against total and cardiovascular mortality and fatal plus non-fatal myocardial infarction. Calcium-channel blockers, including (-8%, P = 0.07) or excluding verapamil (-10%, P = 0.02), as well as AR1 blockers (-24%, P = 0.0002) resulted in better stroke prevention than did the old drugs, whereas the opposite trend was observed for ACE inhibitors (+10%, P = 0.03). The risk of heart failure was higher (P < 0.0001) on calcium-channel blockers (+33%) and alpha-blockers (+102%) than on conventional therapy involving diuretics. Meta-regression Between-group differences in achieved systolic pressure ranged from 0.1 to 3.2 mmHg in seven actively controlled trials (73 237 patients), and from 2.1 to 22.1 mmHg in seven studies comparing varying intensities of blood pressure lowering (11 128 patients). For these 14 new trials, we predicted outcome from achieved systolic blood pressure using our previously published meta-regression models based on 30 trials with 149 407 patients. In general, predicted and observed odds ratios were similar. Larger reductions in systolic pressure (weighted mean 1.8 mmHg) in two trials accounted for the advantage of AR1 blockers; over conventional therapy in the prevention of stroke. Only for cardiovascular mortality in very old patients (P = 0.02) and for cardiovascular events and myocardial infarction in old Australians (P < 0.05), the observed odds ratios deviated from our predictions based on the gradients in systolic blood pressure. Interpretation The hypothesis that new antihypertensive drugs, such as calcium-channel blockers, alpha-blockers, ACE inhibitors or AR1 blockers might influence cardiovascular prognosis over and beyond their antihypertensive effects remains unproven. The finding that blood pressure differences largely accounted for cardiovascular outcome emphasizes the desirability of tight blood pressure control. However, the level to which blood pressure must be lowered to achieve maximal benefit remains currently unknown.
