Effect of Intensive Blood Pressure Lowering on Incident Atrial Fibrillation: A Systematic Review and Meta-Analysis
Matthew J. Singleton1, Lin Y. Chen2, S. Patrick Whalen1, Prashant D. Bhave1, Elijah H. Beaty1, Joseph Yeboah1, Elsayed Z. Soliman13
1Section of Cardiology, Department of Internal Medicine, Wake Forest School of Medicine, Winston-Salem, North Carolina.2Division of Cardiology, Department of Internal Medicine, University of Minnesota Medical School, Minneapolis, Minnesota.3Epidemiological Cardiology Research Center, Wake Forest School of Medicine, Winston-Salem, North Carolina.
The effect of intensive versus standard blood pressure (BP) lowering on the risk of atrial fibrillation (AF) is uncertain. Intensive BP lowering is associated with a lower risk of AF among patients with hypertension. We searched PubMed, EMBASE, and CENTRAL (inception to June 5, 2020) for randomized controlled trials evaluating the effect of intensive versus standard (target systolic BP < 140 mmHg) BP lowering on incident AF. We assessed heterogeneity using the I2 statistic then used fixed-effects meta-analysis models to report pooled treatment effects and 95% confidence intervals. We also tested for publication bias by three funnel plot-based methods. The quality of each study was assessed with the Cochrane Risk of Bias tool. We assessed 16 candidate studies for eligibility from 2,312 published articles, but only three randomized clinical trials were eligible for inclusion and included a combined 12,219 participants with hypertension: Cardio-Sis (Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica), ACCORD-BP (Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial), and SPRINT (Systolic Blood Pressure Intervention Trial). The target systolic BP in the intensive BP arm was <120 mmHg for participants in SPRINT and ACCORD-BP, but <130 mmHg for participants in Cardio-Sis. Participants randomized to intensive BP lowering had significantly lower risk of incident AF compared with those randomized to standard BP lowering (AF incidence 2.2% vs. 3.0%, respectively; pooled hazard ratio (95% confidence interval): 0.74 (0.59 – 0.93)). Intensive BP lowering is associated with a significantly lower risk of incident AF in patients with hypertension. These findings add to the current evidence supporting the benefits of intensive BP control.
Key Words : Atrial fibrillation, Blood pressure, Hypertension, Intensive, Meta-analysis.
Correspondence to: Matthew J. Singleton MD, MBE, MHS, MSc
1 Medical Center Blvd.
Winston-Salem, NC 27157
Atrial fibrillation (AF) is the most common arrhythmia and is associated with a five-fold increased risk of ischemic stroke and a two-fold increased risk of death.1-4 The prevalence of atrial fibrillation continues to increase and is projected to exceed 12 million in the US alone by 2030,5, 6 which has led to calls for additional research exploring primary prevention of AF.7 In observational studies, there is a strong relationship between blood pressure and the risk of AF, with hypertension accounting for the largest population attributable fraction of risk of incident AF.8 There is mounting evidence that intensive blood pressure lowering can decrease the risk of death, atherosclerotic cardiovascular disease, heart failure, and stroke.9, 10 However, the extent to which the risk of AF can be decreased by intensive blood pressure lowering remains uncertain.11, 12 Therefore, a systematic review and meta-analysis was performed to determine whether intensive blood pressure lowering is associated with a reduced risk of incident AF.
Table 1. Participant Baseline Characteristics of Studies Included in Systematic Review and Meta-Analysis of the Association of Blood Pressure Lowering with Incident Atrial Fibrillation.
||Open-label randomized clinical trial
||Age ≥ 55,
SBP ≥ 150 mmHg,
no diabetes, no AF
< 130 mmHg
< 140 mmHg
||Open-label randomized clinical trial
SBP 130 – 180 mmHg,
high CVD risk
< 140 mmHg
||Open-label randomized clinical trial
||No diabetes, no stroke, LVEF ≥ 35%,
SBP 130 – 180, increased CVD risk
135 – 139
Cardio-Sis – Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica ACCORD-BP – Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial SPRINT – Systolic Blood Pressure Intervention Trial SD – standard deviation; SBP – systolic blood pressure (mmHg); AF – atrial fibrillation; CVD – cardiovascular disease; LVEF – left ventricular ejection fraction
We performed a systematic review and meta-analysis according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.13 We searched PubMed, EMBASE, and CENTRAL for articles published from inception to June 1st, 2020, without language or other restrictions. Search terms included: atrial fibrillation AND (blood pressure OR hypertens* OR antihypertens* OR anti-hypertens*) AND (placebo OR control*) AND random. One author (MJS) screened titles and abstracts. For relevant articles, full texts were retrieved and reviewed. The reference lists from all included trials were also reviewed for relevant articles. Trials were eligible for inclusion in the meta-analysis if they were randomized clinical trials comparing two intensities of blood pressure management and included incident AF as a reported outcome in any analysis of the trial. Secondary prevention trials of blood pressure lowering for reducing the risk of recurrent AF after catheter ablation were excluded. Control was defined as the standard blood pressure lowering according to the blood pressure guidelines at the time of conducting the study.
Data were extracted by author MJS using a standardized data extraction form. All included studies and their extracted data were independently reviewed by JY and ES. The primary outcome of this meta-analysis was incident AF. We utilized the Cochrane Risk of Bias tool version 2 to assess the quality of the included studies.14 Domains of assessment included selection bias, performance bias, detection bias, attrition bias, and reporting bias. Hazard ratios (HRs) and 95% confidence intervals (95% CI) were taken from each trial. Chi-squared analysis was used to quantify the degree of heterogeneity between included studies, with a pre-specified p-value of 0.10 used to define significant heterogeneity. Variability and heterogeneity across studies was further assessed using a forest plot and I2 statistics.
Weighted pooled treatment effects were calculated using a fixed-effects meta-analytic model given the low between-study heterogeneity.15 Publication bias was assessed using Egger’s funnel plot.16 Pooled absolute and relative risk reductions were calculated. Statistical analysis was performed using RevMan 5.3.17
The systematic search strategy of published articles yielded 3,142 articles for review [Figure 1]. After excluding duplicates, 2,312 articles were considered for inclusion by reviewing their titles and abstracts. Full texts were retrieved for 16 candidate studies. Of these, one was excluded for being a secondary prevention trial, nine were excluded for not randomizing participants to antihypertensive therapies, and three were excluded for randomizing to a specific class of antihypertensive agent, rather than intensity of therapy, leaving three trials eligible for inclusion in the meta-analysis: Cardio-Sis (Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica),18 ACCORD-BP (Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial),19 and SPRINT (Systolic Blood Pressure Intervention Trial).20
Study flow for systematic review and meta-analysis.
Caption – Of 2,312 unique articles found during the literature search, 16 full-texts were retrieved and 3 studies met criteria for inclusion in the meta-analysis.
Between the three included trials, there were 12,219 participants. The mean age of trial participants was 66.1 ± 8.3 years and 41.0% were women. The mean baseline systolic and diastolic blood pressures were 141.4 ± 15.1 and 79.0 ± 10.9 mmHg, respectively [Table 2]. The median duration of follow-up was 3.8 years. The publication year ranged from 2009 to 2020. Risk of bias was assessed in all included studies, with the overall risk being low in all three studies in all domains assessed [Figure 2]. The funnel plot was not suggestive of publication bias, with relative symmetry about the cumulative effect size [Figure 3].
Risk of bias of included studies.
Caption – All included studies had a low risk of bias in all domains assessed.
Funnel plot of included studies.
Caption – Funnel plot depicting the relationship between treatment effect and study precision is not suggestive of publication bias.
Table 2. Outcomes of Studies Included in Systematic Review and Meta-Analysis of the Association
||Intervention Group BP (mean, SD)
||Control Group BP (mean, SD)
||Systolic: 163.3 (11.3)
Diastolic: 89.6 (8.8)
||Systolic: 27.3 (11.0)
Diastolic: 10.4 (7.5)
||Systolic: 163.3 (11.1)
Diastolic: 89.7 (8.8)
||Systolic: 139.0 (15.0)
Diastolic: 77.5 (9.5)
||Systolic: 139.4 (15.5)
Diastolic: 76.0 (10.2)
||Systolic: 139.5 (15.7)
Diastolic: 78.2 (11.8)
||Systolic: 139.6 (15.3)
Diastolic: 78.1 (11.8)
Cardio-Sis – Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica ACCORD-BP – Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial SPRINT – Systolic Blood Pressure Intervention Trial BP – blood pressure (mmHg); SD – standard deviation; HR – hazard ratio; 95% CI – 95% confidence interval
This systematic review and meta-analysis, which included three trials with 12,219 participants, found intensive blood pressure lowering compared with standard blood pressure control results in a 26% decreased risk of incident AF. The low degree of heterogeneity suggests that this effect is conserved across multiple patient populations. This meta-analysis adds to the known epidemiological association of blood pressure and AF risk in the literature by providing further evidence that intensive blood pressure lowering as a therapeutic strategy that may lower the risk of AF.
Association of Intensive Blood Pressure Lowering with Incident Atrial Fibrillation.
Caption – The blue squares and bars represent the mean values and 95% confidence intervals of the effect sizes, with the area of the blue squares reflecting the weight of the studies. The black diamond represents the combined effect, with the vertical line representing no association.
In aggregate, the evidence available from the epidemiological literature strongly supports the association between higher blood pressure and increased risk of incident AF. However, the question of whether intensive blood pressure lowering decreases the risk of incident AF, or if the propensity for AF might be irreversible such that intensive lowering of blood pressure might not have salutary benefits in prevention of AF, has been unanswered.11 Our findings of a consistent benefit of intensive blood pressure lowering in attenuating the risk of AF lends support to the idea that aggressive control of hypertension may decrease the societal burden of AF.
Though intensive blood pressure lowering was found to decrease the risk of AF in our analysis, it remains unclear whether there is a lower limit of blood pressure below which there is no further reduction in AF risk, or even an increased risk. This is particularly important if intensive blood pressure lowering is to be considered as a tool for lowering the risk of AF, as the target blood pressure depends on the shape of the curve relating blood pressure to AF risk. One study found evidence of a J-shaped curve relating blood pressure to AF risk, such that patients receiving treatment for hypertension with achieved systolic blood pressure < 120 had double the risk of AF compared to those with achieved systolic blood pressure 120 – 129, though the case-control design may allow for residual confounding.21 In contrast, in the Women’s Health Study, a 10 mmHg increase in systolic blood pressure was associated with a 12% increase in risk of AF.22 Importantly, this dose-risk relationship was conserved in all strata of blood pressures tested, with continuously-decreasing risk of AF with progressively decreasing blood pressures, including an systolic blood pressure < 120 mmHg stratum. Findings were similar in a study of Norwegian men, with those in the lowest quartile of systolic blood pressure (88 – 116 mmHg) having the lowest risk of AF.23 An analysis from the ONTARGET/TRANSCEND trials found the group with systolic blood pressure < 120 had the lowest risk of AF.24 Overall, the epidemiological literature suggests that there is no lower limit of blood pressure target below which the benefit in AF prevention is lost, and these findings are in agreement with the result of our meta-analysis.
LESS BENEFIT IN DIABETES?
Of the three trials included in this meta-analysis, only ACCORD-BP did not individually demonstrate a statistically-significant reduction in the risk of incident AF with intensive blood pressure control. This may be explained by ACCORD-BP being underpowered for incident AF, though it should be noted that there were more cases of incident AF in ACCORD-BP than Cardio-Sis. As an alternative explanation, since ACCORD-BP only included participants with diabetes and the other trials specifically excluded participants with diabetes, it is possible that the benefits of intensive blood pressure lowering in decreasing the risk of AF in those with diabetes are less than that seen in those without diabetes. The presence or absence of diabetes may influence the effects of intensive blood pressure lowering on the risk of AF, particularly in light of the fact that intensive blood pressure control has been shown to be beneficial in those without diabetes, but not in those with diabetes, when comparing the SPRINT and ACCORD-BP trials in their primary outcomes.10, 25 It should be noted, however, that although the ACCORD-BP study did not reach statistical significance for reduction in risk of AF, the direction of association still suggested a benefit of intensive blood pressure lowering and the magnitude of the association was only slightly less than that seen in the meta-analysis as a whole.
BENEFIT IN DIVERSE POPULATIONS
The fact that intensive blood pressure lowering appears to confer a lower risk of AF in each of the included trials, despite the differences in their inclusion criteria and participant characteristics, is noteworthy. While ACCORD-BP included 39.2% non-white and SPRINT included 31.6% black participants, Cardio-Sis was a study of nearly all white participants. Since the incidence of AF is approximately 50% higher in whites than blacks,26 the larger magnitude of beneficial effects from intensive blood pressure lowering found in the Cardio-Sis trial may be explained by the predominantly white study population. The mean BMI was 27.8 in Cardio-Sis, 29.9 in SPRINT, and 32.1 in ACCORD. While we cannot exclude an interaction between BMI and intensive blood pressure lowering with regard to AF risk, prior studies have found that the benefits of intensive blood pressure lowering appear to be conserved across all tested BMI strata.27 The proportion of participants with prevalent cardiovascular disease also differed between trials. Only 12% of Cardio-Sis participants had baseline cardiovascular disease, but 19.5% of SPRINT participants and 30.9% of ACCORD-BP participants had prevalent cardiovascular disease. Taken together, the overall consistency of the relationship between intensive blood pressure lowering and AF risk, despite the differences in study populations, is reassuring and increases the generalizability of this finding.
The mechanisms of the observed benefits of intensive blood pressure lowering on reducing AF risk likely involve several pathophysiological pathways.11 Animal studies of the effects of chronic hypertension and its treatment on atrial myocardium have demonstrated that spontaneously hypertensive rats have higher incidence and duration of pacing-induced atrial tachycarrhythmias, with associated deranged calcium handling and increased interstitial fibrosis.28 In an experimental rat model of hypertension in which the atria from rats with surgically-induced partial aortic stenosis were compared with controls, increased afterload led to atrial fibrosis, reduced vectorial conduction velocity, reduced calcium content in the cardiomyocyte sarcoplasmic reticulum, and heterogeneity of conduction velocity, as well as heightened susceptibility to pacing-induced AF and more persistence in AF.29, 30 It is likely that intensive blood pressure lowering either slows progression of these processes or even leads to partial reversal. While it remains unproven that treatment of hypertension can lead to reversal of atrial cardiopathy,31 this is a plausible explanation as there are several studies describing how treatment of hypertension can lead to regression of electrical and structural remodeling in the ventricle, including regression of left ventricular hypertrophy, decreases in left ventricular mass, and improvement in diastolic function.32, 33
LIMITATIONS AND STRENGTHS
Our findings should be interpreted in the context of their limitations. We cannot exclude the possibility that there may be additional studies of intensive blood pressure lowering that reported incident AF as an outcome that were not captured by our search strategy. In addition, despite querying PubMed, EMBASE, and CENTRAL, we found only three studies eligible for inclusion in our meta-analysis, providing us with limited ability to perform more detailed analysis and meta-regression. Surveillance for AF was limited to study electrocardiography in the included studies, so subclinical AF may have escaped detection, though we would not expect a differential effect by intensity of blood pressure lowering. Our data did not include further characterization of AF as paroxysmal or persistent, so any benefits of intensive blood pressure control on slowing progression from paroxysmal AF to persistent AF would not be appreciated in our study. The lack of sex-specific and race/ethnicity-specific event data precludes us from assessing for interaction between demographic characteristics and intensive blood pressure lowering. Since prevalent diabetes was an inclusion criteria for ACCORD and an exclusion criteria for Cardio-Sis and SPRINT, it is possible that the difference in effect size of intensive blood pressure control may be explained by prevalent diabetes serving as a residual confounder, and our meta-analytic methodology with the three included studies does not allow us to teasing out the differential effects in those with and without diabetes. Our study also does not address the effect of intensive blood pressure control on the risk of recurrent AF in patients who have had a prior episode of AF, as Cardio-Sis did not include any participants with prevalent AF and the analyses from ACCORD-BP and SPRINT specifically excluded participants with prevalent AF. Study strengths include the methodological rigor of the underlying randomized controlled trials, rigorous search strategy, inclusion of over 12,000 eligible participants, and the consistent direction of effect observed with low heterogeneity.
In conclusion, this systematic review and meta-analysis of randomized clinical trials published to date demonstrates that intensive blood pressure lowering compared with standard blood pressure control results in a 26% decreased risk of incident AF, and that this effect appears to be consistent in multiple patient populations. Further studies exploring the utility of intensive blood pressure lowering for primary prevention of AF are needed.