22/01/2026
Outcomes in Patients With Atrial Fibrillation Stratified by Body Mass Index and Heart Failure Status
Author links open overlay panel
Jean Jacques Noubiap MD, PhD a b
,
Lisa A. Kaltenbach MS c
,
Karen Chiswell PhD c
,
Ulrich Flore Nyaga MD, MPH d e
,
Mina K. Chung MD f
,
Jeroen M. Hendriks RN, PhD a g h
,
Larry R. Jackson II MD, MHSc c
,
Andrea M. Russo MD i
,
Annabelle Santos Volgman MD j
,
Meghan Reading Turchioe PhD, RN k
,
Ohad Ziv MD l
,
Jonathan P. Piccini MD, MHS c
,
Prashanthan Sanders MBBS, PhD a
Show more
Add to Mendeley
Share
Cite
https://doi.org/10.1016/j.jacadv.2025.102531
Get rights and content
Under a Creative Commons license
Open access
Abstract
Background
An obesity-survival benefit, called the “obesity paradox,” has been variably reported in patients with heart failure (HF) and those with atrial fibrillation (AF), but inconsistencies have been observed.
Objectives
The purpose of this study was to assess how the interaction between body mass index (BMI) and HF status impacts AF-related outcomes.
Methods
Patients hospitalized for AF in the Get With The Guidelines-Atrial Fibrillation registry from 2013 to 2021 and linked to Medicare claims were included. Adjusted Cox proportional hazards models were used to assess the association between BMI and outcomes, stratified by HF status (no HF, HF with preserved ejection fraction, HF with mid-range ejection fraction, and HF with reduced ejection fraction). The outcomes were mortality, cardiovascular rehospitalization, thromboembolism, and myocardial infarction within 1 year.
Results
In total, 21,850 patients (mean age 77 years, 42.3% male) were included: 29.5% underweight/normal (BMI 0.05).
Conclusions
Higher BMI was associated with increased survival in patients without obesity, irrespective of HF status, but not in patients with obesity and AF.
Central Illustration
Download: Download high-res image (769KB)
Download: Download full-size image
Previous article in issue
Next article in issue
Key words
atrial fibrillationheart failuremortalitymyocardial infarctionobesitystroke
Abbreviations and Acronyms
AFatrial fibrillationARBangiotensin II receptor blockerBMIbody mass indexCMScenters for medicare & medicaid servicesHFheart failureHFmrEFheart failure with mid-range ejection fractionHFpEFheart failure with preserved ejection fractionHFrEFheart failure with reduced ejection fractionICDinternational classification of diseasesLVEFleft ventricular ejection fractionRCTsrandomized controlled trialsTIAtransient ischemic attack
Although obesity is a well-established risk factor for the development of heart failure (HF),1, 2, 3, 4 several studies have reported counterintuitive findings suggesting that, among people with chronic HF, overweight and mild to moderate obesity are associated with substantially lower mortality compared with normal weight.5, 6, 7, 8 This “obesity paradox” has been mostly described in patients with HF with reduced ejection fraction (HFrEF), whereas it was inconsistently observed in those with HF with preserved ejection fraction (HFpEF).5,6 The majority of studies that described the “obesity paradox” in HF measured obesity by body mass index (BMI), and studies that used other surrogates of adiposity confirmed this obesity paradox, but with some discrepancies.8, 9, 10, 11
Obesity is commonly seen in patients with atrial fibrillation (AF), a prevalent arrhythmia12 associated with increased risk of ischemic stroke,13 ischemic heart disease,14 and premature death.15 Obesity is an important driver of the occurrence and progression of AF,16, 17, 18, 19 but its prognostic implications in patients with AF are complex. Long-term sustained weight loss has been shown to be associated with a significant reduction of AF burden and maintenance of sinus rhythm,17, 18, 19, 20, 21, 22, 23 prompting the inclusion of weight reduction and other risk factor control as a new pillar of AF management.18,19 However, the impact of obesity on hard outcomes, such as death or ischemic stroke, in patients with AF is unclear. Randomized controlled trials (RCTs) of oral anticoagulation showed lower risks of ischemic stroke and mortality associated with obesity,24, 25, 26, 27 supporting an “obesity paradox.” Conversely, observational studies mostly showed either increased mortality or no impact on mortality and ischemic stroke of obesity compared with normal weight.27, 28, 29
To our knowledge, the influence of HF on the relationship between BMI and AF-related outcomes has not been reported. In view of the above-described inconsistencies, the complex interplay between HF and obesity and their frequent coexistence in patients with AF,26,29 we conducted the current study to assess whether the interaction between BMI and HF status (no HF, HFpEF, or HFrEF) has an impact on mortality, thromboembolism (ischemic stroke, systemic embolism, and transient ischemic attack [TIA]), myocardial infarction, and cardiovascular hospitalization, in a large, real-world cohort of patients with AF.
Methods
Study design and data sources
This cohort study was performed using data from the GWTG-AFIB (Get With The Guidelines-Atrial Fibrillation) registry, a national, voluntary quality improvement program initiated by the American Heart Association in June 2013, aiming to improve cardiovascular health and outcomes in patients with AF through adherence to guideline-recommended therapies.30 The GWTG-AFIB registry collects patient-level data from >150 participating hospitals across the United States, including sociodemographic and clinical characteristics, medical history, diagnosis, treatment, hospital characteristics, and in-hospital outcomes. We also used fee-for-service Medicare claims data from the U.S. Centers for Medicare & Medicaid Services (CMS) which include inpatient information on demographics, date of service, diagnoses recorded using the International Classification of Diseases (ICD)-9th Revision codes and the ICD-10th Revision codes (Supplemental Appendix), and date of death. Data from the GWTG-AFIB registry (2013-2023) and CMS (2013-2021) were linked through a procedure previously described.31
Study cohort
We included patients admitted as an inpatient with a primary or secondary diagnosis of AF and discharged between January 1, 2013, and June 30, 2021, in 152 sites participating in the GWTG-AFIB registry. Our cohort was limited to patients aged ≥65 years who were linked to Medicare fee-for-service claims. We excluded patients with 10% indicating an imbalance between the study groups.
Death and adverse cardiovascular cumulative incidence rates were estimated with follow-up time starting on the date of index discharge and ending when the outcome occurred or at the censor date from CMS claims. We calculated the cumulative incidence for 30-day and 1-year postdischarge primary and secondary outcomes by BMI categories and stratified by HF status. Death cumulative incidence event rates were estimated using the Kaplan-Meier method and presented graphically. Nonfatal outcome rates were estimated using cumulative incidence functions that account for competing risk of death and are presented graphically. Differences by BMI categories within each stratum of HF status were assessed using the log-rank test for fatal endpoints and Gray’s test for nonfatal endpoints.
Cox proportional hazards models with robust variance estimation to account for within-hospital clustering were used to determine the association between BMI (as a continuous measure) and fatal outcomes, with adjustment for the following a priori specified potential confounders: demographics (age, s*x, insurance status), medical history (CAD, prior stroke or TIA, diabetes, hemodialysis, hypertension, liver disease, obstructive sleep apnea, peripheral artery disease, chronic obstructive pulmonary disease, prior hemorrhage, prior myocardial infarction, prior percutaneous coronary intervention, smoker, thyroid disease), type of AF, admission estimated glomerular filtration rate, medications at discharge (beta-blocker, calcium-channel blocker, ACEi/ARB, ARNi, statin, antiarrhythmic, antiplatelet, direct oral anticoagulant, warfarin), and hospital characteristics (region, teaching/nonteaching, number of beds, rural/urban location). Cause-specific Cox proportional hazards regression models were used to account for the competing risk of death for nonfatal outcomes. Models were fit stratified by HF status group. First, we graphically assessed the functional form of BMI on each unadjusted time-to-event outcome using restricted cubic splines with 5 knots selected a priori32 and located at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles of BMI distribution, and with BMI on the x-axis and log HR on the y-axis. The functional form was determined to be significantly nonlinear and, for simplicity of interpretation, was approximated with a linear spline (“broken stick model”) with a knot at BMI = 30 kg/m2, allowing us to characterize the HR for increments in BMI for values
