The review included 190 studies, comprising 409 treatment groups and 28,922 patients.
The frequency of DAEs was highest with calcium-channel blockers (6.7%) then alpha-adrenergic blockers (6.0%), ACE inhibitors (4.7%), beta-blockers (4.5%), placebo (4.3%) and, finally, angiotensin-receptor blockers and diuretics (both 3.1%). The DAE risk differences relative to placebo were calculated for diuretics (0.027, 95% CI: 0.001, 0.053), beta-blockers (0.018, 95% CI: -0.008, 0.044), ACE inhibitors (0.014, 95% CI: -0.002, 0.029), calcium-channel blockers (-0.005, 95% CI: -0.022, 0.012) and angiotensin-receptor blockers (0.02, 95% CI: 0.001, 0.038). The advantage of both diuretics and angiotensin-receptor blockers relative to within-study placebo was statistically significant, (P=0.038 in both cases). The DAE event rate was not significantly different from placebo for any of the other categories of antihypertensive agent.
The range of total AE frequency among different classes of antihypertensive agents was narrow, extending from 39.3% (diuretics) to 32.3% (beta-blockers). The AE frequency with placebo was 37.3%. The frequency ranking of antihypertensive agents varied by organ system. Rankings of agents for AEs by organ system did not parallel frequency rankings of DAEs for the same organ system.
A meta-analysis of risk difference of DAE frequency in placebo-controlled monotherapy trials supported the frequency rankings of the DAE frequency, as computed by class of antihypertensive agent in the treatment groups. Tests for heterogeneity were not significant in any instance for these meta-analyses. The risk of DAEs was greater in placebo groups than for same-study beta-blocker, ACE inhibitor, diuretic or angiotensin-receptor blocker groups, of which the latter two reached statistical significance. However, for studies of calcium-channel blockers, the risk of DAE was higher for the calcium-channel blocker group than for the placebo group, but it was not statistically significant.
A meta-analysis of the risk difference of DAE frequency in active comparison trials showed statistical significance for four comparisons: diuretics versus calcium-channel blockers, beta-blockers versus calcium-channel blockers, angiotensin-receptor blockers versus ACE inhibitors, and ACE inhibitors versus alpha-adrenergic blockers. In each case, the latter drug group had a significantly higher frequency of DAEs than the former.
When stratifying groups by study duration of 1 month or less versus more than 1 month, the results differed by drug class. In ACE inhibitor, calcium-channel blocker and placebo groups the frequency of DAEs more than doubled from short- to long-term studies, suggesting that not all DAEs with these drugs occur in the first month. Little difference was seen in DAE frequency between short- and long-term studies for beta-blocker, angiotensin-receptor blocker and alpha-adrenergic blocker groups. The overall DAE frequency was 4.0% in fixed-dosage studies and 5.4% in dose-titration studies.
The analysis of variance procedure, undertaken to assess the impact of covariates on DAE frequency, showed that covariates of interest relating to the study were geographic location, year of study and quality score. This was explained by the differences in reporting practices. For patient characteristics, only concurrent left ventricular hypertrophy had a significant effect on the frequency of DAEs in the model containing all covariates. However, when studied in a univariate model, significance was not present.
The average Jadad quality score was 3.2 (range: 1 to 5), with 80% of the studies scoring 3 or higher.