Decreases in systolic blood pressure during exercise may predispose to arrhythmias such as atrial fibrillation (AF) because of underlying abnormal autonomic tone. We examined the association between systolic blood pressure response and incident AF in 57,442 (mean age 54 ± 13 years, 47% women, and 29% black) patients free of baseline AF who underwent exercise treadmill stress testing from the Henry Ford ExercIse Testing project. Exercise systolic blood pressure response was examined as a categorical variable across clinically relevant categories (>20 mm Hg: referent; 1 to 20 mm Hg, and ≤0 mm Hg) and per 1-SD decrease. Cox regression, adjusting for demographics, cardiovascular risk factors, medications, history of coronary heart disease, history of heart failure, and metabolic equivalent of task achieved, was used to compute hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between systolic blood pressure response and incident AF. Over a median follow-up of 5.0 years, a total of 3,381 cases (5.9%) of AF were identified. An increased risk of AF was observed with decreasing systolic blood pressure response (>20 mm Hg: HR 1.0, referent; 1 to 20 mm Hg: HR 1.09, 95% CI 0.99, 1.20; ≤0 mm Hg: HR 1.22, 95% CI 1.06 to 1.40). Similar results were obtained per 1-SD decrease in systolic blood pressure response (HR 1.08, 95% CI 1.04 to 1.12). The results were consistent when stratified by age, sex, race, hypertension, and coronary heart disease. In conclusion, our results suggest that a decreased systolic blood pressure response during exercise may identify subjects who are at risk for developing AF.
Several studies have shown that a reduction in systolic blood pressure below the resting value (e.g., exercise-induced hypotension) during incremental exercise stress testing is associated with an increased risk of cardiovascular events. Explanations for this abnormal blood pressure response include the presence of severe obstructive coronary artery disease that results in left ventricular dysfunction, and neurocardiogenically mediated vasodilation. Alterations in autonomic tone are associated with an increased risk for atrial fibrillation (AF) development, and it is plausible that subjects who experience decreases in systolic blood pressure during exercise have abnormal autonomic regulation that predisposes to this arrhythmia. However, to our knowledge, this hypothesis has not been explored. Therefore, the purpose of this analysis was to examine the risk of AF associated with systolic blood pressure response during exercise treadmill testing using data from the Henry Ford ExercIse Testing (FIT) project, a racially diverse registry of men and women aimed to elucidate the association between cardiorespiratory fitness and cardiovascular outcomes.
Details of the design, procedures, and methods used in the FIT project have been previously described. Briefly, the project population consists of 69,885 consecutive patients who underwent physician-referred exercise treadmill stress testing in the Henry Ford Health System, including affiliated hospitals and ambulatory care centers throughout the metropolitan area of Detroit, Michigan from 1991 to 2009. Data regarding treadmill testing, medical history, and medications were collected by laboratory staff at the time of testing. Follow-up data were collected from electronic medical records and administrative claims databases. The FIT project was approved by the Henry Ford Health System Institutional Review Board.
In this analysis, we examined the association between exercise systolic blood pressure response (peak systolic blood pressure−systolic blood pressure at rest) and the risk of new-onset AF. Patients with a history of AF (n = 1,975) or severe valve disease (n = 579) were excluded. We also excluded patients with missing baseline characteristics, medication data, and/or follow-up data (n = 1,577). Additionally, AF follow-up data were available beginning in 1995, and participants with stress tests before this time were excluded (n = 8,312). The final sample included 57,442 (mean age 54 ± 13 years, 47% women, and 29% black) patients.
Demographics and clinical characteristics were obtained at the time of treadmill testing. Smoking status was self-reported. Diabetes mellitus was defined as a previous diagnosis of diabetes, the use of hypoglycemic medications including insulin, or a database-verified diagnosis of diabetes. Obesity was self-reported. Hypertension was defined as a previous diagnosis of hypertension or a database-verified diagnosis. The blood pressure at the time of stress testing was not used to diagnose hypertension. Hyperlipidemia was defined as a previous diagnosis of any major lipid abnormality or a database-verified diagnosis of hypercholesterolemia or dyslipidemia. Coronary heart disease was defined as a history of previous myocardial infarction, coronary angioplasty, coronary artery bypass grafting surgery, or coronary angiography with evidence of obstructive coronary artery disease. Heart failure was defined as a previous clinical diagnosis of systolic or diastolic heart failure.
Exercise treadmill stress testing was conducted using the Bruce protocol. Patients aged <18 years at the time of testing or those who underwent pharmacologic stress testing, modified Bruce, or other non-Bruce protocol tests were excluded from the database. Heart rate at rest was measured from the electrocardiogram at rest, and blood pressure was manually measured before each stress test with each participant in the upright position. Heart rate was measured continuously during testing, and blood pressure values were measured every 3 minutes. Peak heart rate and blood pressure were the values recorded closest to the end of the exercise test for each participant. Initial treadmill speed was set at 2.7 km/h and increased to 4.0, 5.4, 6.7, 8.0, 8.8 km/h on 3, 6, 9, 12, and 15 minutes, respectively. Exercise workload was expressed in metabolic equivalent of task (MET). We examined the association between exercise systolic blood pressure response as a categorical variable across clinically relevant categories (>20 mm Hg: referent; 1 to 20 mm Hg, and ≤0 mm Hg) and as a continuous variable per 1-SD decrease in systolic blood pressure response.
AF events were ascertained through linkage with administrative claim files from services delivered by the system-affiliated group practice and/or reimbursed by the system’s health plan. These files included the appropriate International Classification of Disease Code, Ninth Revision for AF (427.31). A new diagnosis was considered present when the appropriate International Classification of Disease Code, Ninth Revision code, either primary or secondary, was identified in at least 3 separate follow-up encounters. Complete follow-up for AF events was available through March 2010.
Categorical variables were reported as frequency and percentage, whereas continuous variables were reported as mean ± SD. Statistical significance for categorical variables was tested using the chi-square method and the Wilcoxon rank-sum procedure for continuous variables. Follow-up time was defined as the date of exercise stress testing until the date of censoring or AF development. Patients were censored when they lost contact with the health system (death, loss to follow-up, or end of follow-up, which was March 2010). Kaplan-Meier estimates were used to compute cumulative incidence curves for AF by systolic blood pressure response, and the differences in estimates were compared using the log-rank procedure. Cox regression was used to compute hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between systolic blood pressure response and AF. We also constructed a restricted cubic spline model to examine the graphical dose-response relationship between systolic blood pressure response and AF at the 5th, 50th, and 95th percentiles. Multivariate models were constructed as follows: Model 1 adjusted for age, sex, and race; Model 2 adjusted for Model 1 covariates plus smoking, hypertension, diabetes, obesity, hyperlipidemia, coronary heart disease, heart failure, antihypertensive medication use, lipid-lowering medication use, aspirin, and METs achieved. We tested for interactions between our main effect variable and age (dichotomized at the median age for study participants), sex, race (non-white vs white), hypertension, and coronary heart disease.
Several sensitivity analyses were performed. The normal increase in exercise systolic blood pressure response is 10 mm Hg per 1 MET achieved. Therefore, subjects who did not complete the first stage of stress testing possibly did not achieve a high enough workload for their systolic blood pressure to increase. Accordingly, a sensitivity analysis was performed in which patients with <5 METs achieved were excluded. Owing to the potential for exercise systolic blood pressure to be attenuated in patients with elevated systolic blood pressure at rest, we performed an additional analysis excluding those with systolic blood pressure values >150 mm Hg at rest.
The proportional hazards assumption was not violated in our analysis. Statistical significance was defined as p <0.05 for the main effect model and tests for interaction. SAS version 9.3 (Cary, North Carolina) was used for all analyses.
Baseline characteristics stratified by systolic blood pressure response are listed in Table 1 . Differences were noted for all characteristics between blood pressure groups except for smoking. Over a median follow-up of 5.0 years, a total of 3,381 cases (5.9%) of AF were identified. The incidence rate (per 1,000 person-years) of AF increased with decreases in systolic blood pressure response (>20 mm Hg: 9.3, 95% CI 8.9 to 9.6; 1 to 20 mm Hg: 18.9, 95% CI 17.4 to 20.5; ≤0 mm Hg: 30.1, 95% CI 26.4 to 34.3). The cumulative incidence of AF by systolic blood pressure response is shown in Figure 1 (log-rank p <0.0001).
|Characteristics||Systolic Blood Pressure Response (mm Hg)||P-value ∗|
(n = 1,623)
|1 to 20 |
(n = 6,243)
(n = 49,576)
|Age, mean (SD) (years)||63 (13)||58 (14)||53 (12)||<0.0001|
|White||1,101 (68%)||4,090 (66%)||31,684 (64%)||0.0004|
|Male||807 (50%)||2,748 (44%)||26,702 (54%)||<0.0001|
|Smoker||713 (44%)||2,637 (42%)||21,262 (43%)||0.42|
|Obesity †||345 (21%)||1,428 (23%)||12,099 (24%)||0.0006|
|Diabetes mellitus||489 (30%)||1,506 (24%)||9,459 (19%)||<0.0001|
|Hypertension ‡||1,390 (86%)||4,707 (75%)||30,949 (62%)||<0.0001|
|Hyperlipidemia §||1,396 (86%)||5,194 (83%)||40,384 (81%)||<0.0001|
|Coronary heart disease ¶||420 (26%)||1,278 (20%)||4,910 (10%)||<0.0001|
|Heart failure||80 (4.9%)||225 (3.6%)||670 (1.4%)||<0.0001|
|Aspirin||500 (31%)||1,688 (27%)||9,543 (19%)||<0.0001|
|Antihypertensive medications||1,183 (73%)||3,782 (61%)||21,734 (44%)||<0.0001|
|Lipid-lowering therapies||590 (36%)||1,989 (32%)||11,640 (23%)||<0.0001|
|METs achieved, mean (SD)||6.2 (3.0)||7.5 (3.0)||9.3 (2.9)||<0.0001|
In a multivariate Cox regression analysis, an increased risk of AF was observed with decreasing systolic blood pressure response ( Table 2 ). The results remained similar when stratified by age, sex, race, hypertension, and coronary heart disease ( Table 2 ). The graphical representation between systolic blood pressure response and AF risk is shown in Figure 2 . The risk of AF increased with lower systolic blood pressure response, with the highest risk observed in those with peak systolic blood pressure values lower than resting values.
|Events/No at risk||Model 1 ∗ |
|P-value||Model 2 † |
|P-value||Interaction ‡ P-value|
|Systolic Blood Pressure Response|
|>20 mm Hg||2,592/49,576||1.0||–||1.0||–|
|1 to 20 mm Hg||566/5,677||1.41 (1.29, 1.55)||<0.0001||1.09 (0.99, 1.20)||0.074|
|≤0 mm Hg||223/1,400||1.69 (1.47, 1.94)||<0.0001||1.22 (1.06, 1.40)||0.0069|
|Per 1-SD Decrease||3,381/57,442||1.22 (1.18, 1.26)||<0.0001||1.08 (1.04, 1.12)||<0.0001|
|Age § ¶|
|<53 years||686/28,303||1.33 (1.23, 1.45)||<0.0001||1.12 (1.04, 1.21)||0.0026||0.34|
|≥53 years||2,695/29,139||1.37 (1.32, 1.42)||<0.0001||1.13 (1.09, 1.17)||<0.0001|
|Female||1,369/27,185||1.17 (1.11, 1.23)||<0.0001||1.07 (1.02, 1.13)||0.0090||0.71|
|Male||2,012/30,257||1.25 (1.19, 1.30)||<0.0001||1.09 (1.04, 1.14)||0.0002|
|Non-White||920/20,567||1.23 (1.16, 1.31)||<0.0001||1.07 (1.00, 1.14)||0.041||0.98|
|White||2,461/36,875||1.20 (1.15, 1.25)||<0.0001||1.09 (1.04, 1.13)||<0.0001|
|No||580/20,396||1.11 (1.02, 1.21)||0.019||1.06 (0.97, 1.15)||0.20||0.99|
|Yes||2,801/37,046||1.21 (1.17, 1.26)||<0.0001||1.09 (1.05, 1.13)||<0.0001|
|Coronary heart disease ¶|
|No||2,365/50,834||1.17 (1.13, 1.22)||<0.0001||1.09 (1.04, 1.13)||<0.0001||0.11|
|Yes||1,016/6,608||1.16 (1.09, 1.24)||<0.0001||1.06 (0.99, 1.13)||0.069|