Association of hypertension severity and control with risk of incident atrial fibrillation: The REasons for Geographic And Racial Differences in Stroke (REGARDS) study

Abstract Background The association of hypertension (HTN) severity and control with the risk of incident atrial fibrillation (AF) is unclear. Hypothesis Increased HTN severity and poorer blood pressure control would be associated with an increased risk of incident AF. Methods This analysis included 9485 participants (mean age 63 ± 8 years; 56% women; 35% Black). Participants were stratified into six mutually exclusive groups at baseline—normotension (n = 1629), prehypertension (n = 704), controlled HTN (n = 2224), uncontrolled HTN (n = 4123), controlled apparent treatment‐resistant hypertension (aTRH) (n = 88), and uncontrolled aTRH (n = 717). Incident AF was ascertained at the follow‐up visit, defined by either electrocardiogram or self‐reported medical history of a physician diagnosis. Multivariable logistic regression analyses adjusted for demographic and clinical variables. Results Over an average of 9.3 years later, 868 incident AF cases were detected. Compared to those with normotension, incident AF risk was highest for those with aTRH (controlled aTRH: odds ratio (OR) 2.95; 95% confidence interval (CI) 1.60, 5.43, & uncontrolled aTRH: OR 2.47; 95% CI 1.76, 3.48). The increase in AF risk was smaller for those on no more than three antihypertensive agents regardless of their blood pressure control (controlled OR 1.72; 95% CI 1.30, 2.29 and uncontrolled OR 1.56; 95% CI 1.14, 2.13). Conclusions The risk of developing AF is increased in all individuals with HTN. Risk is highest in those aTRH regardless of blood pressure control. A more aggressive approach that focuses on lifestyle and pharmacologic measures to either prevent HTN or better control HTN during earlier stages may be particularly beneficial in reducing related AF risk.


| INTRODUCTION
The prevalence of atrial fibrillation (AF) has tripled over the past 50 years. 1It is estimated that nearly 50 million individuals are estimated to have AF worldwide, making this the most prevalent clinically significant cardiac arrhythmia. 2 Although individuals with AF do not die directly from the arrhythmia itself, they are at a much higher risk for stroke, heart failure, and cardiovascular mortality. 3,4[11][12][13] Prospective studies looking at AF risk across the spectrum of clinically defined stages (prehypertension, HTN, controlled HTN, uncontrolled HTN, and apparent treatment-resistant hypertension [aTRH]) that concomitantly take blood pressure and medication use into account are lacking.A previous study in the REasons for Geographic And Racial Differences in Stroke (REGARDS) found no differences in AF prevalence across worsening HTN stages; however, this study was cross-sectional. 14We determined whether increasing HTN severity and poorer BP control stratified according to these easily defined clinical categories are associated with a higher risk of incident AF in the REGARDS study, a large biracial prospective cohort study of men and women.Findings from the study will provide important clinical insight into how risk of developing AF may differ according to severity and control of HTN.

| METHODS
Details of the methods of the REGARDS study have been published. 15iefly, REGARDS is a prospective cohort study designed to identify contributors to regional and Black-White disparities in stroke mortality.The study over-sampled Black persons and residents of the stroke belt (North Carolina, South Carolina, Georgia, Alabama, Mississippi, Tennessee, Arkansas, and Louisiana).Between January 2003 and October 2007, using postal mailings and telephone interviews, a total of 30 239 participants were recruited from a commercially available list of residents.Sociodemographic information and medical histories were obtained by a computer-assisted telephone interview (CATI).An in-home examination was performed 3-4 weeks after the telephone interview.Trained staff collected medication information, blood and urine samples, blood pressure readings, and a resting electrocardiogram (ECG).Approximately 10 years after the baseline assessment, 2013-2016, participants who were still alive and active completed a follow-up examination similar to the baseline visit.
The institutional review boards at the collaborating centers approved the REGARDS study protocol, and all participants provided written informed consent.Eligible participants will be those from the REGARDS cohort with baseline data on blood pressure and antihypertensive use as well as follow-up data from the second inhome visit for the development of AF.

| Blood pressure measurements
BP was taken by trained examiners using android sphygmomanometer.BP was measured twice following a standard protocol.
All participants were asked to sit for 5 minutes with feet on floor before BP measurement, and there was a 30-second interval between measurements.The average of two readings was calculated.
BP quality was monitored by central examination of digit preference and retraining of personnel as needed.Antihypertensive medication use was self-reported.Medication adherence was assessed using a four-item validated scale. 16HTN duration was determined based on the self-reported history of short, medium, and long-term (i.e., <10, 10-20, and >20 years, respectively) at the baseline visit.

| Definition of groups based on BP and antihypertensive treatment
We will stratify the cohort into six mutually exclusive groups based on BP control and number of antihypertensive medications used. 17

| AF
AF was identified at baseline and a subsequent follow-up visit approximately 10 years later by (1) scheduled ECG and (2) self-reported history of a physician diagnosis during the CATI survey.The ECGs were read and coded at a central reading center (EPICARE, Wake Forest School of Medicine) by analysts who were blinded to other REGARDS data.Selfreported AF was defined as an affirmative response to the following question: "Has a physician or a health professional ever told you that you had atrial fibrillation?"This question has been shown to be a reliable predictor of incident stroke events as AF detected by ECG. 18

| Covariates
Participant characteristics at baseline were used as covariates.Age, sex, race, household income, education, and smoking status were self-reported.Body mass index (BMI) was measured at the baseline examination.Physically active was defined as ≥4 days of exercise (enough to work up a sweat) per week.Diabetes mellitus was defined as fasting glucose ≥126 mg/dL, nonfasting glucose ≥200 mg/dL, or self-reported current use of antidiabetic medications.C-reactive protein (CRP) measurement used a high-sensitivity particle-enhanced immunonephelometric assay on the BNIII nephelometer (N High-Sensitivity CRP, Dade Behring Inc.) with an interassay coefficient of variation of 2%-6%.Using isotope-dilution mass spectrometry traceable serum creatinine, estimated glomerular filtration rate (eGFR) was calculated using the abbreviated Modification of Diet in Renal Disease study equation. 19Cardiovascular disease included the presence of coronary heart disease (CHD), defined as a self-reported history of myocardial infarction, coronary artery bypass grafting, coronary angioplasty or stenting, or evidence of prior myocardial infarction on the baseline ECG, or prior stroke which was ascertained by participant's self-report at the time of study enrollment.

| Statistical analysis
Participants with missing baseline exposure (blood pressure or antihypertensive use) or covariate measures, poor quality ECG recordings, baseline AF, or missing.AF follow-up data were excluded.
Descriptive statistics for demographic, socioeconomic, lifestyle, anthropometric, and medical history variables at the baseline assessment according to blood pressure groups were calculated using the χ 2 test (for proportions) and analysis of variance (for continuous variables).
Odds ratios (ORs) for incident AF were calculated among blood pressure groups (referent = normotensive).We also performed subgroup analyses to evaluate the effect modification by age (<mean vs. ≥mean), race (white vs. black), and gender (female vs. male).
Multivariable logistic regression was used to compute OR and 95% confidence intervals (CI) for associations involving blood pressure categories and incident AF, overall and stratified by age, race, gender, and geographic region.Multivariable models were adjusted for as follows: Model 1 adjusted for age, sex, height, race, education, income, and geographic region.Model 2 included covariates in Model 1 + BMI, diabetes, total cholesterol, high-density lipoprotein (HDL), eGFR, smoking, exercise, alcohol use, log-transformed CRP, CHD, and stroke.We additionally adjusted duration of HTN and medication adherence to determine whether these factors attenuated associations.We excluded an additional 107 participants for analyses adjusting for HTN duration due to missing data.Statistical significance for all comparisons including interactions was defined as p < .05.SAS version 9.4 was used for all analyses.

| RESULTS
Figure 1 shows the flowchart of participants.There were 20 805 REGARDS participants with complete covariate data and without AF at baseline and 9485 of these were included in the prospective analyses.Baseline characteristics stratified by blood pressure groups are shown in Table 1.Mean (SD) age was 63.3 (8.3) years, 35% were Black, and 56% female.A total of 868 (9.1%) participants developed AF over a median (interquartile range [IQR]) follow-up of 9.3 (1.4) years.Compared to normotensive participants those with worsening HTN severity were older, more likely to be male, be black, smoke, to have diabetes, CHD, a prior stroke, and less likely to have a high school education, have an annual income >$35 000, exercise, or drink alcohol (Table 1).Kidney function and HDL-c were lower while high-sensitivity CRP was higher across worsening HTN severity groups.
Risk of incident AF according to blood pressure groups and is reported in Table 2, respectively.In fully adjusted analyses (Model 2), participants with HTN had a higher risk of incident AF compared to those with normal blood pressure.The risk was nominally highest for those on more than three antihypertensive drugs with either controlled (OR 2.95; 95% CI 1.60, 5.43) or uncontrolled (OR 2.46; 95% CI 1.76, 3.48) blood pressure.
The increase in AF risk was less for those on no more than three antihypertensive agents regardless of their blood pressure control (controlled OR 1.72; 95% CI 1. 30 | 1421 changed after additional adjustment for either duration of HTN (Model 3) or medication adherence (Model 4).
In subgroup analyses, associations were stronger in older participants (Table 3), female participants (Table 4), and white participants (Table 5).Adjusted incident AF risk was not significantly increased across HTN groups in younger participants (mean age < 63.2 years) or black participants.

| DISCUSSION
Compared to those with normotension, the risk of developing AF was highest for those with aTRH.The increase in AF risk was of lower magnitude but still significant for those on no more than three antihypertensive agents.ORs were similar regardless of achieved blood pressure control.No significant associations were observed for Results reported our build upon prior work by discerning whether differences in AF risk exist according to these characteristics.
It is unclear whether the relation between BP and the risk of AF is linear or whether there is a threshold BP value above which the risk is definitively increased.A prior cross-sectional study in REGARDS found that although the prevalence of AF was higher in persons with HTN compared to those without HTN, the severity of HTN was not associated with AF prevalence. 14It is important to note, however, that a higher prevalence ratio for AF was seen in the aTRH groups in models that did not additionally adjust for use of antihypertensive medications.In this regard, cross-sectional findings are like prospective associations reported here.
Prospective studies and randomized trials have had mixed results.Among over 4000 Framingham Heart Study participants, the presence of HTN regardless of achieved control was most associated with incident AF risk. 20Similarly, in an Italian-based cohort of nearly 2500 initially untreated subjects with essential HTN and free of clinical cardiovascular disease no association was observed between blood pressure level the incidence of AF. 21In a nationwide South Korean sample of over 3 million adults, however, AF risk increased with both a higher persistence as well as severity of HTN. 22 has also been suggested that a U-shaped association between HTN and incident AF exists, whereby risk for an SBP > 150 mmHg was similarly increased to an SBP < 120 mmHg compared to those with treated SBP levels between 120 and 129 mmHg. 23This study, however, included only participants with treated HTN and did not distinguish participants according to number of antihypertensive agents used.Finally, while results from individual randomized controlled trials for treatment of HTN have suggested lower ontreatment BP levels are associated with a reduced risk of AF, [24][25][26][27] a large meta-analysis of randomized trials comparing various antihypertensive drug regimens found that differences in blood pressure variability or mean SBP were not related to risk of new AF. 28Our findings suggest that accounting for the number of medications as a measure of severity of HTN may better identify individuals at higher AF risk than single BP measurements alone.
Age is the strongest AF risk factor.In the MESA cohort participants 70-79 and 80-89 years of age had an over seven-and ninefold respective increase in AF risk compared to those 50-59 years old. 5ile the age-adjusted incidence of AF is higher in men compared to women, the lifetime risk is similar.Underlying AF risk factors and taller stature are thought to largely explain the higher AF incidence in men. 1,29spite the generally higher burden of most primary AF risk factors, incident AF rate ratios are lower in blacks compared to whites. 5The lifetime AF risk is one in five for blacks compared to one in three for whites. 1 This disproportionately lower risk in blacks is most notable in the absence of cardiovascular risk factors. 30The impact of HTN severity and control on these associations is unclear. 31We found that associations of worse HTN severity were stronger in older participants but weaker in both male and white participants.Whites with controlled aTRH also had a lower adjusted baseline prevalence of AF in this cohort compared with blacks but gender and age had no impact on associations with AF prevalence based on HTN severity. 14r study has multiple strengths that include the ability to In conclusion, the risk of developing AF was increased all HTN groups and became stronger with worsening HTN severity.BP control did not influence risk.Associations were stronger in older participants, female participants, and white participants.A more aggressive approach that focuses on lifestyle and pharmacologic measures to either prevent HTN or better control HTN during earlier stages may be particularly beneficial in reducing related AF risk.
evaluate the relationship between HTN and incident AF risk with adjustment for multiple HTN characteristics (severity, medication adherence, and duration) in a biracial cohort.Limitations are present as well.Covariates were assessed at baseline only.Duration of HTN relied solely on participant recall.The sensitivity to detect AF was limited to ECG detection at the follow-up visit.The sample size for the aTRH groups were significantly smaller, leading to wider point estimates.Results for these subgroups should be interpreted within this context.Finally, as is true with any observational study, we cannot rule out residual confounding by unmeasured covariates.
Flowchart of study participants.Baseline characteristics according to blood pressure groups.
T A B L E 2 Risk of incident AF across blood pressure groups.Risk of incident AF across blood pressure groups stratified by age.Model 1 = adjusted for age, sex, height, race, education, income, and geographic region.Model 2 = Model 1 + BMI, diabetes, TC, HDL, eGFR, smoking, exercise, alcohol use, log-transformed CRP, CHD, and stroke.Model 3 = Model 2 + duration of HTN.Model 4 = Model 2 + medication adherence.The p value displayed in this table is the result of the Type 3 Analysis of Effects hypothesis test for the blood pressure groups.Risk of incident AF across blood pressure groups stratified by gender.Model 1 = adjusted for age, sex, height, race, education, income, and geographic region.Model 2 = Model 1 + BMI, diabetes, TC, HDL, eGFR, smoking, exercise, alcohol use, log-transformed CRP, CHD, and stroke.Model 3 = Model 2 + duration of HTN.Model 4 = Model 2 + medication adherence.Risk of incident AF across blood pressure groups stratified by race.Model 1 = adjusted for age, sex, height, race, education, income, and geographic region.Model 2 = Model 1 + BMI, diabetes, TC, HDL, eGFR, smoking, exercise, alcohol use, log-transformed CRP, CHD, and stroke.Model 3 = Model 2 + duration of HTN.Model 4 = Model 2 + medication adherence.
T A B L E 3 aT A B L E 4Abbreviations: AF, atrial fibrillation; aTRH, apparent treatment-resistant hypertension; BMI, body mass index; CHD, coronary heart disease; CRP, Creactive protein; eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; HTN, hypertension; TC, total cholesterol.TA B L E 5