Authors: Vidhushei Yogeswaran (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Medicine, University of Washington, Seattle, Washington, USA), Kerri L. Wiggins (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Medicine, University of Washington, Seattle, Washington, USA), Colleen M. Sitlani (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Medicine, University of Washington, Seattle, Washington, USA), Susan R. Heckbert (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Epidemiology, University of Washington, Seattle, Washington, USA), Hooman Kamel (Feil Family Brain and Mind Research Institute, Weill Cornell Medicine, Clinical and Translational Neuroscience Unit, Department of Neurology, New York, NY, USA), Emelia J. Benjamin (Cardiovascular Medicine Section, Department of Medicine, Boston Medical Center, Boston University Chobanian and Avedisian School of Medicine, Boston, MA, USA; Department of Epidemiology, Boston University School of Public Health, Boston, MA, USA), Wondwosen Kassahun‐Yimer (Department of Data Science, University of Mississippi Medical Center, Jackson, MS, USA), Rizwan Kalani (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Medicine, University of Washington, Seattle, Washington, USA; Department of Neurology, University of Washington, Seattle, Washington, USA), Elsayed Z. Soliman (Epidemiological Cardiology Research Center, Department of Internal Medicine, Wake Forest School of Medicine, Winston‐Salem, NC, USA), April P. Carson (Department of Medicine, University of Mississippi Medical Center, Jackson, MS, USA), James S. Floyd (Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA; Department of Medicine, University of Washington, Seattle, Washington, USA; Department of Epidemiology, University of Washington, Seattle, Washington, USA)
Categories: Original Research, atrial cardiomyopathy, atrial fibrillation, Black adults, ischemic stroke, P‐wave indices, Atrial Fibrillation, Cardiovascular Disease, Risk Factors
Source: Journal of the American Heart Association: Cardiovascular and Cerebrovascular Disease
P‐Wave Indices and the Risks of Incident Atrial Fibrillation and Ischemic Stroke in Black Adults: The Jackson Heart Study
Authors: Vidhushei Yogeswaran, Kerri L. Wiggins, Colleen M. Sitlani, Susan R. Heckbert, Hooman Kamel, Emelia J. Benjamin, Wondwosen Kassahun‐Yimer, Rizwan Kalani, Elsayed Z. Soliman, April P. Carson, James S. Floyd
Abstract
Background
Atrial cardiomyopathy, defined as adverse changes to the cardiac atria, is an emerging risk factor for atrial fibrillation (AF). Despite the disproportionate burden of AF‐related complications in Black adults, the relationship between electrocardiographic (ECG) measures of atrial cardiomyopathy, referred to as P‐wave indices (PWIs), and outcomes remains largely unknown in this population.
Methods
In 4900 participants without AF at baseline from the JHS (Jackson Heart Study), a prospective cohort study of Black adults, we assessed atrial cardiomyopathy using PWIs from resting 12‐lead ECGs: PR interval, P‐wave duration, P‐wave axis, and P‐wave terminal force in V1 (PTFV1). Cox proportional hazards models evaluated associations of PWIs with incident AF and ischemic stroke, adjusting for established risk factors.
Results
Over 13.7 years of follow‐up, 396 (9.3%) participants developed incident AF, and 135 (3.8%) experienced ischemic stroke. Each SD‐unit increase in PR interval, P‐wave duration, and PTFV1 was associated with increased AF risk, with the strongest association for PTFV1 (hazard ratio [HR], 1.27 [95% CI, 1.16–1.38]). These PWIs were also associated with AF risk using clinically accepted thresholds, including HR of 1.76 for PTFV1 ≥4000 ms×μV (95%CI, 1.38–2.25). Each SD increase in PTFV1 was associated with a 22% increased risk of ischemic stroke (HR, 1.22 [95% CI, 1.07–1.40]).
Conclusions
In this large cohort study of Black adults, PWIs, particularly PTFV1, were independently associated with an increased risk of incident AF and ischemic stroke. These findings underscore the significance of PTFV1 as an underlying marker of atrial cardiomyopathy in Black adults.
Clinical PerspectiveWhat Is New? In a large community‐based cohort of Black adults, we observed that abnormal P‐wave indices are independently associated with an increased risk of developing AF and ischemic stroke over longitudinal follow‐up.P‐wave terminal force in lead V1 emerged as a crucial marker of atrial cardiomyopathy in Black adults, with higher values of P‐wave terminal force in lead V1 associated with increased risk of both incident AF and ischemic stroke. What Are the Clinical Implications? Accessible 12‐lead ECG‐derived measures of atrial cardiomyopathy, particularly P‐wave terminal force in lead V1, may serve as a potential screening tool to identify those at high risk for AF and ischemic stroke.
Atrial fibrillation (AF) affects more than 10 million adults in the United States ^1^ , ^2^ and is associated with increased risks of ischemic stroke, heart failure, dementia, and death. Although population‐based studies indicate a lower prevalence of clinically detected AF in Black adults compared with White adults, ^3^ , ^4^ which may be due to underdiagnosis, ^5^ Black adults with AF experience higher rates of stroke and AF‐related complications. ^6^ These inequities highlight an urgent need for new strategies to improve the diagnosis of AF in Black adults.
Recent evidence suggests that atrial cardiomyopathy, defined as structural, contractile, or electrophysiologic abnormalities of the cardiac atria, plays a crucial role in the development of AF and its associated complications. ^7^ Atrial cardiomyopathy can be detected through electrocardiographic measures, including P‐wave indices (PWIs). ^8^ , ^9^ PWIs are noninvasive measures of atrial electrical activity hypothesized to reflect subclinical atrial remodeling. ^10^ , ^11^ These indices measure different aspects of atrial depolarization and can be easily obtained from a standard 12‐lead ECG, an affordable and readily available screening tool. ^9^ , ^11^ The PR interval reflects the time from atrial to ventricular depolarization, the P‐wave duration reflects interatrial conduction time, the P‐wave axis measures the net direction of atrial depolarization, and the PTFV1 evaluates the left atrial activation of the P wave. ^9^ , ^12^ Previous studies have observed that these ECG measures of subclinical atrial remodeling can detect which patients may be at high risk for AF ^10^ , ^12^ , ^13^ , ^14^ and ischemic stroke. ^15^ , ^16^ , ^17^
The association between PWIs and clinical outcomes has been inconsistent across multiethnic cohorts. ^10^ , ^18^ , ^19^ For instance, PTFV1 has been independently associated with AF in both the Atherosclerosis Risk in Communities Study ^10^ and the Multi‐Ethnic Study of Atherosclerosis ^19^ but not in the Framingham Heart Study. ^10^ These inconsistencies underscore the need to further investigate these associations, particularly in Black adults, who have been reported to have a high prevalence of abnormal PWIs ^20^ and elevated risk for AF‐related complications. ^6^ , ^21^ In this study, we investigated the association between PWIs, AF, and ischemic stroke within the Jackson Heart Study (JHS), the largest prospective cohort exclusively focused on cardiovascular disease and risk factors in Black adults.
METHODS
JHS data can be requested for the purpose of reproducing study results. Requests to access this data set may be directed to the JHS Coordinating Center.
Data Sources
The JHS is a prospective community‐based cohort study that recruited 5306 Black adults (self‐identified) from the tri‐county area of Jackson, Mississippi. Participants were aged 20 to 95 years at baseline, except in the family study, which also recruited participants aged ≥21. ^22^ Data collection began on September 26, 2000, ^22^ and 3 study visits have been completed to Baseline examination 1 (2000–2004), examination 2 (2005–2008), and examination 3 (2009–2013). Data collected during examinations include demographics, anthropometry, electrocardiography, echocardiography, laboratory values, and medication history. Also, since baseline, participants have completed annual follow‐up calls to assess changes in health status, and medical records have been reviewed to adjudicate cardiovascular events. All participants provided written informed consent, and local institutional review boards approved the study protocols. ^22^ This study followed the Strengthening the Reporting of Observational Results in Epidemiology reporting guideline.
Study Population
We excluded participants with prevalent AF (n=84), paced rhythm (n=3), missing ECG data at examination 1 (n=165), and individuals who refused consent for the use of medical record information (n=180). This led to a total of 4900 participants for analyses of incident AF. For analyses of incident ischemic stroke, we also excluded individuals with prevalent stroke (n=239) or who refused consent for the use of medical record information (n=180) at examination 1, leading to a total of 4693 participants.
P‐Wave Indices
PWIs were measured from a supine 10‐second resting 12‐lead ECG performed at examination 1. ^23^ ECGs were recorded using the Marquette MAC/PC digital ECG recorder (Marquette Electronics, Milwaukee, WI) with simultaneous 12‐lead acquisition by trained JHS technical staff. All ECGs were visually inspected for errors or inadequate quality and were then transmitted via phone modem to the JHS ECG Reading Center. P‐wave and other interval measurements were generated automatically using the well‐validated Minnesota Code Modular ECG Analysis System computer program, which has been used across several clinical trials and population‐based studies. ^23^ , ^24^ PWIs evaluated were PR interval, P‐wave duration, P‐wave axis, and PTFV1. ^9^ , ^10^ The PR interval was measured from the onset of the P‐wave to the start of the QRS segment in ms. Individual PR intervals were coded for all leads, and the median value was calculated. P‐wave duration (ms) was measured from the P‐wave onset to its offset from any individual lead. The net direction of electrical activity on the electrocardiographic reference system determines the P‐wave axis (degrees). PTFV1 was defined as the absolute value of the depth (μV) of the downward deflection (terminal portion) of the P‐wave in ECG lead V1 multiplied by its duration (ms). PR interval, P‐wave duration, and PTFV1 were evaluated as continuous variables. Additionally, PR interval, P‐wave duration, P‐wave axis, and PTFV1 were categorized as abnormal on the basis of clinically accepted PR interval of <120 milliseconds (short) or >200 milliseconds (long); P‐wave duration ≥120 milliseconds; P‐wave axis of any value outside of 0 to 75 degrees; PTFV1 ≥4000 ms*μV. ^9^ , ^12^ , ^25^ , ^26^
Outcomes
AF, defined as either atrial fibrillation or atrial flutter, was identified from study ECGs at in‐person examinations, diagnosis codes from hospitalization surveillance, diagnosis codes from Medicare claims data, and death records, as described previously. ^27^ AF that was first identified on or before the date of baseline examination was considered prevalent AF, and AF identified after the baseline examination was considered incident AF. Prevalent stroke at examination 1 was identified by a self‐reported history of stroke at baseline. Incident ischemic stroke events identified during follow‐up through December 31, 2016, were adjudicated from medical record review and death certificates. ^28^ , ^29^ Data on ischemic stroke subtypes (eg, cardioembolic) were not available for this analysis. Longitudinal follow up for AF and stroke events was conducted through December 31, 2016.
Covariates
Covariates assessed at examination 1 include age, sex, height, weight, systolic blood pressure (SBP), diastolic blood pressure (DBP), self‐reported medication use, current smoking, and medical history using standardized protocols. Body mass index was calculated from weight in kilograms divided by height in meters squared (kg/m^2^). Prevalent heart failure was assessed by self‐report and medication. Prevalent diabetes was defined by hemoglobin A1c ≥6.5%, fasting blood glucose >126 mg/dL, or use of diabetes medications within 2 weeks before the clinic visit. Secondary adjustment variables include additional clinical, echocardiographic, and laboratory risk factors associated with individual clinical outcomes. Coronary artery bypass history was assessed from cardiac procedure history at examination 1. Left ventricular mass, left ventricular ejection fraction, and left atrial maximal internal diameter were assessed from examination 1 echocardiography. Fasting triglycerides, C‐reactive protein levels, and B‐type natriuretic peptide (BNP) were measured from blood specimens collected at examination 1. Fasting serum low‐density lipoprotein levels were estimated using the Friedewald formula.
^23^
The estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equations.
^30^
Statistical Analysis
PWIs were evaluated as continuous variables per SD‐unit increase and as categorical variables based on clinically accepted thresholds. Cox proportional hazards models were used to assess the associations of PWIs with the risk of incident AF, adjusting for age, sex, height, weight, systolic BP, diastolic BP, use of antihypertensive medications, current smoking status, prevalent diabetes, history of heart failure, and history of myocardial infarction. ^31^ Additional models adjusted for estimated glomerular filtration rate, BNP, and echocardiographic variables (left atrial maximal internal diameter, left ventricular mass, left ventricular ejection fraction). Effect modification by sex and age (<60 or ≥60 years) was evaluated in fully adjusted models. Additionally, the functional relationships between continuous PWIs and outcomes were assessed using restricted cubic spline models. Complete case analyses were performed due to minimal missing data in the primary covariates (<5%) for the main analysis.
Cox proportional hazards models were used to evaluate the associations of PWIs with the risk of incident ischemic stroke as a secondary outcome, adjusted for age, sex, BMI, systolic BP, diastolic BP, low‐density lipoprotein, triglycerides, current smoking status, antihypertensive medication use, aspirin use, antiplatelet use, statin use, prevalent diabetes, and prevalent heart failure. ^32^ Additional models adjusted for history of bypass surgery (coronary artery bypass graft), BNP, C‐reactive protein, and echocardiographic variables (left atrial maximal internal diameter, left ventricular mass, left ventricular ejection fraction) and adjusted for intercurrent AF that occurred during follow‐up as a time‐varying covariate. ^32^ , ^33^ Effect modification by sex and age group (<60 or ≥60 years) was evaluated in fully adjusted models. Sensitivity analyses for PR interval associations were performed to adjust for resting heart rate and exclude participants with ECG evidence of Wolff–Parkinson–White pattern, second‐degree heart block, or third‐degree heart block, as well as evaluate age as a spline term.
Separate Cox models were fit for each sex and age (<60 or ≥60 years) group for both outcomes. The proportional hazards assumption was tested using Schoenfeld residuals after all primary Cox proportional models. To further assess the model assumptions, dfbeta residuals were examined and demonstrated no evidence of influential outliers. Martingale residual plots were also inspected and were approximately linear for all continuous covariates. A mild deviation was noted for age, which was subsequently modeled using spline terms in additional sensitivity analyses. All statistical analysis was performed using R Statistical Software version 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
Among the 4900 eligible participants, the mean age was 55±13 years, and 37% were men (Table 1). More than half used antihypertensive medications, 3% had prevalent heart failure, and 5% had prevalent myocardial infarction. Individuals in the highest quartiles of PWIs were older and had a higher prevalence of antihypertensive medication use (Tables S1 through S3). When PWIs were categorized by clinically acceptable thresholds, 1.5% had a short PR interval, 11% had a long PR interval, 46% had abnormal P‐wave duration, 6.6% had abnormal P‐wave axis, and 10% had abnormal PTFV1. Similar baseline demographics were observed in patients with ischemic stroke data (Table S4). Abnormal PWIs were more common in men and older individuals (Table S5).
Atrial Fibrillation
During a median follow‐up period of 13.7 (interquartile range, 11.8–14.6) years, 396 (9.3%) of the 4252 participants with complete covariate data eligible for analysis developed incident AF. In primary analyses adjusted for AF risk factors, each continuous PWI±SD was significantly associated with the risk of incident AF (Table 2). Associations were not substantially altered by additional adjustment for estimated glomerular filtration rate, BNP, or echocardiographic measures of left atrial and left ventricular function. In a model that included all continuous PWIs, only PTFV1 was significantly associated with AF risk (hazard ratio [HR], 1.25, [95% CI, 1.16–1.37]), and the risk estimate was nearly the same as in the primary analysis (HR, 1.27 [95% CI, 1.16–1.38]; Table 2). There was evidence of significant effect modification of the PTFV1–AF association by age (P=0.01): younger individuals (aged <60 years) had a 41% increased risk of AF per each SD higher PTFV1 [95% CI, 18–68%], compared with a 27% increased risk of AF [95% CI, 15%–39%] in older individuals (aged ≥60 years) (Table S6).
Visual inspection of restricted cubic spline plots suggested that the risk of incident AF was higher in the upper and lower distributions of PR interval, P‐wave duration, and P‐wave axis, while PTFV1 demonstrated a gradual increase throughout the range of values observed (Figure). For ischemic stroke, visual inspection also demonstrated a gradual increase in PTFV1 across the distribution of values (Figure S1). Similar patterns were observed in age‐adjusted plots (Figures S2 and S3). When modeled by clinically accepted abnormal thresholds (Table 3) and compared with participants with normal PWI intervals, short PR interval was associated with a >2‐fold increased risk of AF (HR, 2.66 [95% CI, 1.25–5.67]), long PR interval with a 48% increased risk of AF (HR, 1.48 [95% CI, 15%–91%]), abnormal P‐wave duration with a 29% increased risk of AF [95% CI, 5%–60%], and abnormal PTFV1 with a 76% increased risk of AF [95% CI, 38%–125%]. Although short PR interval had the largest point estimate, only 1.5% of participants had this abnormal PWI, and the CIs overlap with the estimate for abnormal PTFV1.
Ischemic Stroke
During a median follow‐up period of 13.8 (interquartile range, 12.8–14.6) years, 135 (3.8%) of the 3573 participants with complete covariate data eligible for analysis developed incident ischemic stroke. In primary analyses, only PTFV1 was significantly associated with ischemic stroke risk (HR, 1.22 [95% CI, 1.07–1.40]; Table 4). Results were similar after adjusting for additional potential confounders and intercurrent AF events that occurred during follow‐up (HR, 1.21 [95% CI, 1.05–1.38]). There was evidence of significant effect modification of the associations of PFTV1 with age and PR interval with sex (Table S7), where women had an increased risk of ischemic stroke (HR, 1.23 [95% CI, 1.03–1.47]) per each SD higher PR interval compared with men (HR, 0.88 [95% CI, 0.65–1.17]). The findings for PTFV1 are similar to those in AF; younger individuals (aged <60 years) had an increased risk of ischemic stroke (HR, 1.44 [95% CI, 1.11–1.86]) compared with older individuals (aged ≥60 years) (HR, 1.17 [95% CI, 0.98–1.39]) per each SD higher PTFV1 (Table S7). When modeled by clinically accepted thresholds, there were no significant associations between PWIs and incident ischemic stroke risk (Table S8). We also repeated PR interval analyses excluding patients with heart block and Wolff–Parkinson–White pattern, and associations remained relatively unchanged (Table S9).
DISCUSSION
In this large community‐based sample of Black adults with longitudinal follow‐up for incident AF and incident ischemic stroke events, PWIs were associated with new‐onset AF, independent of established AF risk factors. These associations were robust to adjustment for additional measures of left atrial and left ventricular function. In particular, PTFV1 was significantly associated with increased AF risk after adjustment for other PWIs, and associations were strongest in younger individuals, for whom there is little information on AF risk factors. ^34^ PTFV1 was also significantly associated with the risk of incident ischemic stroke, even after adjustment for intercurrent AF events, and the magnitude of the association with ischemic stroke risk (HR, 1.44) was nearly identical to the association with AF (HR, 1.41) in participants aged <60 years. These analyses support the relevance of clinically accepted thresholds for PWIs in Black adults in relation to AF risk and highlight the importance of PTFV1 as a generalizable marker of atrial cardiomyopathy with broad clinical relevance in Black adults.
Our results are consistent with findings from previous studies that included predominantly White participants and identified abnormal PWIs as strong predictors of AF. ^9^ , ^12^ , ^14^ PTFV1 is a marker of left atrial enlargement ^35^ and elevated left atrial pressures. ^36^ , ^37^ An analysis from the Atherosclerosis Risk in Communities study of Black and White middle‐aged adults found that PTFV1 was the PWI most strongly associated with AF, ^10^ and PTFV1 was also associated with incident stroke in the Multi‐Ethnic Study of Atherosclerosis cohort. ^16^ Findings from our study, which included more than twice the number of ischemic stroke events in Black adults as in prior studies, ^16^ suggest that PTFV1 may be a potential marker to identify Black adults at higher risk for AF‐related complications. Our stratified analyses suggest that PTFV1 may be especially potent in younger individuals, who face an increased risk of stroke‐related morbidity and death. Recent advances in machine learning and artificial intelligence may enable the automated, widespread screening of PTFV1 across underrepresented cohorts using digital or analog ECGs. ^38^ However, further studies across multiple diverse cohorts are needed to evaluate the clinical utility of these artificial intelligence–driven approaches.
Our study adds to the growing body of evidence that subclinical atrial dysfunction, or atrial cardiomyopathy, may play a significant role in the pathogenesis of AF and stroke. ^31^ Atrial remodeling can create a substrate conducive to the development of AF. ^39^ These structural and functional changes can also impair endocardial function, promote a procoagulant state, ^39^ , ^40^ increase the risk of thrombus formation, and ultimately lead to ischemic stroke. ^39^ , ^41^ , ^42^ The “AF paradox” in Black adults, ^43^ characterized by a lower incidence of clinically detected AF but paradoxically higher rate of AF‐related complications compared with White adults, ^44^ may be explained in part by a higher prevalence of subclinical atrial dysfunction. This hypothesis is further supported by the higher prevalence of abnormal PWIs in Black adults compared with White adults in several population‐based studies. ^20^ , ^45^ However, it is important to note that, although PTFV1 was associated with ischemic stroke in our analysis, we were unable to distinguish cardioembolic stroke from other ischemic stroke subtypes. Therefore, the observed association may reflect both direct cardioembolic risk from atrial cardiomyopathy and indirect associations through shared pathways.
Strengths of this study include the large number of participants, high‐quality data from standardized resting 12‐lead ECG, and longitudinal follow‐up data. However, there are also limitations. First, given the modest number of AF and stroke events, we had limited power to detect effect modification. Second, AF events were assessed from study ECGs and diagnoses made during health care encounters, which may underestimate the true burden of this arrhythmia, especially among Black adults. ^5^ Third, PWIs were obtained from a single baseline ECG and may not reflect temporal variations, ^9^ , ^46^ which may lead to new clinical insights. Fourth, multiple hypotheses were tested in this study, and as a result, some findings may be due to chance. ^47^ Fifth, data on P‐wave morphology are not available in JHS, limiting our ability to evaluate interatrial block, which has been identified as a potent predictor of adverse events, as a PWI risk measure in this cohort. ^48^ , ^49^ Finally, despite the high‐quality assessment of AF risk factors, this is an observational study, and residual confounding is a possible explanation for our findings.
CONCLUSIONS
In this large analysis from a prospective community‐based cohort study of Black adults, we observed that PWIs, widely available noninvasive measures of atrial cardiomyopathy, were associated with increased risk of AF and ischemic stroke. These associations were independent of established clinical risk factors, with PTFV1 emerging as the most potent PWI risk measure. Our findings underscore the significance of PWIs as generalizable risk measures in Black adults and the potential of PTFV1 as a screening tool for AF and ischemic stroke. Additional research is needed to evaluate whether PWIs can improve the identification of Black adults who are at the highest risk for AF and its complications, potentially improving long‐standing health inequities.
Sources of Funding
The JHS is supported and conducted in collaboration with Jackson State University (HHSN268201800013I), Tougaloo College (HHSN268201800014I), the Mississippi State Department of Health (HHSN268201800015I), and the University of Mississippi Medical Center (HHSN268201800010I, HHSN268201800011I, and HHSN268201800012I) contracts from the National Heart, Lung, and Blood Institute and the National Institute on Minority Health and Health Disparities. Dr Benjamin was supported in part by R01HL092577. This work was supported by National Heart, Lung, and Blood Institute award R01HL142599.
Disclosures
None.