Results
Exclusion and inclusion of study participants are shown in Figure 1. Baseline characteristics of individuals free of atrial fibrillation, heart failure, ischaemic heart disease, diabetes mellitus, and hyperthyroidism in the nested 1:1 matched studies within the entire Danish population 1995 through 2010 are shown in Table 1 and Supplementary Material Online, Table S2 . Baseline characteristics of all individuals free of atrial fibrillation, heart failure, ischaemic heart disease, diabetes mellitus, and/or hyperthyroidism on antihypertensive monotherapy and of the entire Danish population are shown in Supplementary Material Online, Table S3 .
(Enlarge Image)
Figure 1.
Exclusion and inclusion of study participants from the entire Danish population from 1995 through 2010. ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker.
Cumulative Incidence of Atrial Fibrillation
ACEi and ARB monotherapy were both associated with lower cumulative incidences of atrial fibrillation compared with β-blocker and diuretic monotherapy, but not compared with calcium-antagonists (Figure 2). Median follow-up times for ACEi and ARB monotherapy were 6.8 years (interquartile range 3.0–9.7) and 6.6 years (5.9–9.2) when compared with β-blocker, 5.9 years (2.0–8.8) and 6.7 years (3.4–9.4) when compared with diuretic, and 5.9 years (2.4–7.9) and 6.6 years (3.3–9.5) when compared with calcium-antagonist monotherapy. Median follow-up time for ACEi compared with ARB monotherapy was 6.5 years (3.3–8.9). Incidence rates for the individual classes of antihypertensive medication in the individual nested 1:1 matched studies are shown in Table 2 .
(Enlarge Image)
Figure 2.
Cumulative incidences of atrial fibrillation as a function of follow-up time in nested 1:1 matched studies within the entire Danish population from 1995 through 2010. Calculations of cumulative incidence allowed for competing risk of death. ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker.
Incidence Rates of Atrial Fibrillation as a Function of Calendar-year
Incidence rates of atrial fibrillation as a function of calendar-year increased more in the groups treated with β-blocker or diuretic monotherapy compared with incidence rates of atrial fibrillation in the other three groups of antihypertensive monotherapy (Figure 3).
(Enlarge Image)
Figure 3.
Incidence per year of atrial fibrillation in nested 1:1 matched studies within the entire Danish population as a function of calendar-year (1995 through 2010). ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker.
Risk of Atrial Fibrillation and Stroke
The hazard ratio of atrial fibrillation for ACEi monotherapy was 0.12 (95% CI: 0.10–0.15) compared with β-blocker, 0.51 (0.44–0.59) compared with diuretic, 0.97 (0.81–1.16) compared with calcium-antagonist, and 1.46 (1.04–2.05) compared with ARB monotherapy (Figure 4). In contrast, for risk of stroke as an endpoint influenced by lowering blood pressure rather than renin-angiotensin system blockade per se, hazard ratios for ACEi monotherapy vs. other antihypertensive monotherapies did not differ from 1.0 (Figure 4).
(Enlarge Image)
Figure 4.
Risk of atrial fibrillation and stroke in nested 1:1 matched studies within the entire Danish population from 1995 through 2010. Individuals with atrial fibrillation were excluded from the analyses of stroke risk. Hazard ratios were adjusted multivariably for defined daily dose, ethnicity, highest obtained level of education, geographical residency, and paired matching was included in the model as a stratification variable, in the studies matched for antihypertensive medication, gender, age at first prescription, and propensity score. ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; HR, hazard ratio; CI, confidence interval.
The hazard ratio of atrial fibrillation for ARB monotherapy was 0.10 (95% CI: 0.07–0.14) compared with β-blocker, 0.43 (0.32–0.58) compared with diuretic, 0.78 (0.56–1.08) compared with calcium-antagonist, and 0.68 (0.49–0.96) compared with ACEi monotherapy (Figure 4). In contrast, for risk of stroke, hazard ratios for ARB monotherapy vs. other antihypertensive monotherapies did not differ from 1.0 (Figure 4). Corresponding comparisons among β-blocker, calcium-antagonist, and diuretic monotherapy are presented in Supplementary Material Online, Figure S5.
(Enlarge Image)
Supplementary Figure 5.
Sensitivity Analyses
First, when β-blocker monotherapy included sotalol (see Supplementary Material Online, Figure S1), results were similar to those shown in Figure 4; however, when calcium-antagonists included verapamil hazard ratios of atrial fibrillation associated with ACEi and ARB monotherapy were 0.52 (0.44–0.60) and 0.41 (0.30–0.54) (see Supplementary Material Online, Figure S1), whereas corresponding hazard ratios were 0.97 (0.81–1.16) and 0.78 (0.56–1.08) when verapamil was excluded (Figure 4). Second, when sotalol, alprenolol, and oxprenolol were excluded from the β-blocker group (see Supplementary Material Online, Figure S2), results were similar to Figure 4. Third, when sotalol, atenolol, and bisoprolol were excluded from the β-blocker group (see Supplementary Material Online, Figure S3), results were similar to Figure 4. Fourth, when individuals with diseases predisposing to atrial fibrillation were included (see Supplementary Material Online, Figure S4), results were similar to those shown in Figure 4. The only exception was a slight increase in risk of stroke associated with ACEi monotherapy compared with β-blocker monotherapy (see Supplementary Material Online, Figure S4). Fifth, when we calculated hazard ratios of atrial fibrillation and stroke among β-blocker, calcium-antagonist, and diuretic monotherapy β-blocker monotherapy was associated with the increased risk of atrial fibrillation compared with calcium-antagonists and diuretics (see Supplementary Material Online, Figure S5). Sixth, when adjustment additionally included new onset risk factors for atrial fibrillation during follow-up and hospitalizations due to any cause during follow-up (see Supplementary Material Online, Table S4 and Table S5 ) and for chronic kidney disease and renal hypertension at baseline ( Table 1 and Supplementary Material Online, Table S2), results were similar (compare Supplementary Material Online, Figure S6 with Figure 4). Seventh, when limiting the analysis to only validated atrial fibrillation diagnoses (see Supplementary Material Online, Figure S7), results were similar to those shown in Figure 4. Finally, hazard ratios of atrial fibrillation were similar in strata of age and gender compared with overall analyses (see Supplementary Material Online, Figure S8).
(Enlarge Image)
Supplementary Figure 1.
(Enlarge Image)
Supplementary Figure 2.
(Enlarge Image)
Supplementary Figure 3.
(Enlarge Image)
Supplementary Figure 4.
(Enlarge Image)
Supplementary Figure 6.
(Enlarge Image)
Supplementary Figure 7.
(Enlarge Image)
Supplementary Figure 8.