Sage Journals: Discover world-class research

Abstract

Background

Standard Modifiable Cardiovascular Risk Factors (SMurfs) influence Atrial fibrillation (AF). However, data on their cumulative prognostic effect in Asian non-valvular AF (NVAF) patients receiving oral anticoagulants (OACs) remain limited. This study investigated the association between the cumulative burden of SMurfs and the prognosis of these NVAF patients on oral anticoagulation (OAC) therapy.

Methods

This retrospective cohort study used Tianjin Health and Medical Data Platform (THMDP) data on 20 782 NVAF patients on OACs between 2015 and 2020. Patients were categorized into groups by the number of baseline SMurfs (0, 1-2, or 3-4: hypertension, diabetes, hyperlipidemia, smoking). Cox proportional hazards models assessed the association between SMurf burden group and the primary outcome (all-cause mortality) and secondary outcomes. Subgroup and sensitivity analyses were performed to evaluate the consistency and robustness of the findings.

Results

After adjusting for confounders, compared to the SMurf-less group, the hazard ratios (HRs) for all-cause mortality were 1.20 (95% confidence interval [CI]: 1.01-1.44) for the SMurf (1-2) group and 1.67 (95% CI: 1.33-2.08) for the SMurf (3-4) group. For ischemic stroke, the hazard ratio (HR) was 1.44 (95% CI: 1.03-1.99) for the SMurf (3-4) group, while the association was not significant for the SMurf (1-2) group. A more pronounced association between SMurfs and all-cause mortality was observed in patients with an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m².

Conclusion

In Asian NVAF patients receiving OACs, a higher SMurf burden is significantly associated with an increased risk of all-cause mortality and ischemic stroke.

Keywords

atrial fibrillation risk factor prognosis

What's New?

In a large-scale cohort (>20 000) of anticoagulated Asian patients with non-valvular atrial fibrillation (NVAF), this is the first study to quantify the cumulative, dose-response impact of standard modifiable cardiovascular risk factors (SMurfs) on prognosis. We demonstrate that a simple SMurf count is a powerful, independent predictor of all-cause mortality and ischemic stroke.

This prognostic impact is significantly amplified in patients with chronic kidney disease, identifying a critical subgroup for targeted intervention. These findings highlight the insufficiency of anticoagulation alone and provide robust, real-world evidence for implementing comprehensive risk management, reinforcing the “Comorbidity” pillar of the Atrial Fibrillation Better Care (ABC) pathway to address the patient's total residual risk.

Introduction

Atrial fibrillation (AF) is the most common type of cardiac arrhythmia globally, often linked to complications like stroke and systemic embolism.^1,2 AF patients commonly have multiple comorbidities, leading to high disability and mortality rates, imposing substantial health and economic burdens.

The approach to AF management has evolved significantly towards a more holistic and integrated strategy. A pivotal development was the introduction of the “Atrial Fibrillation Better Care” (ABC) pathway in 2017, which established a simple yet comprehensive framework for AF management: “A” for anticoagulation/avoid stroke, “B” for better symptom management, and “C” for cardiovascular and comorbidity optimization.³ The effectiveness of this integrated approach is now well-supported by a robust body of evidence. Its benefits have been validated in major randomized clinical trials, such as the mAFA-II and MIRACLE-AF trials in China,^4,5 and have been consistently demonstrated in large-scale observational studies⁶ and further substantiated by meta-analyses showing that structured risk factor management improves AF-related outcomes.⁷ This integrated approach was formally incorporated into major international guidelines, including the 2020 European Society of Cardiology (ESC) guidelines.⁸ More recent guidelines, such as the 2023 ACC/AHA guidelines (with its similar SOS pathway) and the 2024 ESC AF-CARE pathway, have further reinforced and built upon this established focus on integrated care, underscoring that comprehensive risk factor management is a central pillar of optimal AF care.^1,2 Therefore, AF should not be viewed in isolation, a thorough evaluation and management of comorbidities and risk factors are essential for patients with AF.

While oral anticoagulation (OAC) is the cornerstone for preventing cardioembolic stroke in AF, a substantial residual risk for adverse outcomes persists, even in fully anticoagulated patients. This residual risk could be largely driven by the progression of underlying atherosclerotic disease, which is influenced by the burden of standard modifiable cardiovascular risk factors (SMurfs). These factors, comprising hypertension, diabetes mellitus (DM), hyperlipidemia, and smoking, are well-established drivers of atherosclerosis.^9–11 Although the benefits of managing these risk factors individually in patients with AF are recognized, the cumulative prognostic impact of harboring multiple SMurfs on this residual risk is not well-defined.

Therefore, this study aims to investigate the association between the cumulative burden of SMurfs and adverse prognostic outcomes in a large, real-world cohort of non-valvular AF patients receiving oral anticoagulants. By quantifying the dose-response relationship between the number of SMurfs and the risk of adverse outcomes, we seek to underscore the importance of the baseline risk factor profile and provide a stronger evidence base for comprehensive risk management in this population.

Methods

Study Design and Population

The data for this study were primarily sourced from the Tianjin Health Medical Data Platform (THMDP), an electronic medical record-based healthcare data platform in Tianjin, a provincial-level municipality located in Northern China with a large, diverse urban population of 13.9 million. The THMDP includes electronic health information from 97.8% (261/267) of community healthcare centers and 78 public hospitals, including 100% (42/42) of tertiary general hospitals and 36 secondary hospitals. The system employs a unique personal identification mechanism that enables continuous monitoring and coverage of the entire healthcare pathway for patients in Tianjin, based on hospital visits and information from community healthcare centers. The initial cohort included 134 516 patients who were first received OACs between January 2015 and June 2020. The index date was defined as the date of the first OAC prescription. The final study population was identified from this cohort by applying a multi-step exclusion process based on International Classification of Diseases, 10th Revision (ICD-10) codes. To define the target population of NVAF, patients were first excluded if they had no prior hospital discharge diagnosis of atrial fibrillation (ICD-10: I48). Subsequently, to ensure a true NVAF cohort and remove other primary indications for OAC, patients were also excluded if they had a prior discharge diagnosis of significant valvular heart disease or mechanical valve implantation, or if their OAC prescription was documented solely for the prevention of venous thromboembolism (VTE) complications. Additional exclusion criteria included: a prior OAC prescription within 12 months before admission; initiation of OAC therapy more than 90 days after admission; a diagnosis of trauma; and missing low-density lipoprotein cholesterol (LDL-C) data at baseline. This 90-day interval from hospital admission to the initiation of OAC therapy was established to ensure that the baseline data recorded at admission accurately represented the patients’ clinical status at the time of treatment initiation, thereby minimizing potential misclassification bias.¹² The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and received full approval from the Ethics Committee of Tianjin Medical University General Hospital (Approval No: IRB2022-YX-235-01). The committee waived the requirement for individual informed consent due to the retrospective nature of the research and the use of fully anonymized, de-identified data, which posed minimal risk to patient privacy. The study's flowchart outlines the detailed inclusion and exclusion criteria (Figure S1 in the Supplement).

Definition of Exposure and Outcomes

The exposure in this study was defined as the number of baseline SMurfs, which includes hypertension, hyperlipidemia, diabetes, and smoking. Hypertension and diabetes were identified based on ICD-10 codes. The smoking status of patients was obtained by searching the personal registration information at community health service centers. Hyperlipidemia was defined as LDL-C ≥ 3.37 mmol/L and/or total cholesterol (TC) ≥ 5.18 mmol/L.⁹ Participants were grouped based on their baseline SMurf counts into three categories: SMurf-less, SMurf (1-2), and SMurf (3-4). This categorization, which groups patients into low, intermediate, and high-burden categories, is consistent with the methodology used in prior research on the cumulative burden of SMurfs.⁹

The primary outcome of this study was all-cause mortality. Patient mortality data was primarily collected from the hospital electronic medical records system, in conjunction with the THMDP, and death registration records from community health care centers. The study also assessed secondary outcomes such as ischemic stroke (IS), intracranial hemorrhage (ICH), gastrointestinal bleeding (GIB), and acute myocardial infarction (AMI), as well as composite outcomes of ischemic events (IS and AMI) and bleeding events (ICH and GIB). Among them, ICH includes intracerebral hemorrhage, subarachnoid hemorrhage, epidural hemorrhage, or subdural hemorrhage. The identification of ICH, IS, and GIB was based on ICD-10 codes and clinically relevant examinations results such as computed tomography (CT), magnetic resonance imaging (MRI), endoscopy. The identification of AMI was based on ICD-10 codes. Follow-up began on the index date and continued until the first outcome occurred, the maximum follow-up period of 3 years was attained, until December 31, 2021, or discontinuation of OACs (no new prescription within 183 days after the last prescription date), whichever came first.

Study Variables

The study's baseline variables encompassed demographic factors, clinical parameters, scoring scales, medical history, and medication use. Demographic factors included age, sex, body mass index (BMI), smoking status, and alcohol consumption. The main laboratory test information included hemoglobin (Hb), TC, LDL-C, high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), uric acid. Estimate Glomerular filtration rate (eGFR) was calculated using the equation provided by the Chronic Kidney Disease Epidemiology Collaboration.¹³ The scoring scales included CHA₂DS₂-VASc score, HAS-BLED score, DOAC score. Most medical history of patients was determined by searching for diagnoses that appeared in the system before the inclusion date of the population, according to ICD-10 codes, including stroke, venous thromboembolism, ischemic heart disease, peripheral vascular disease, bleeding or coagulation disorders, hemiplegia, dementia, chronic pulmonary diseases, gastrointestinal diseases, liver diseases, anemia, cancer, heart failure, connective tissue diseases, psychiatric disorders, except for hyperuricemia (defined as a serum uric acid concentration ≥420 umol/L for both sexes)¹⁴ and hyperlipidemia (as previously described). Baseline medication use was determined by having at least one relevant prescription record within 6 months prior to baseline exposure, including anti-platelet agents, oral anticoagulants (categorized by OAC type: non-vitamin K antagonist oral anticoagulants [NOACs] vs warfarin), non-steroidal anti-inflammatory drugs (NSAIDs), gastric mucosal protective agents, glucocorticoids, amiodarone, anti-thyroid agents, selective serotonin reuptake inhibitors, statins, anti-hypertensive agents, anti-diabetic agents. A detailed list of the specific drugs included in each of these classes is provided in Table S1 in the Supplement.

Statistical Analyses

The distribution of SMurfs and internal risk factor composition is visually represented using donut charts and bar graphs. Normally distributed continuous variables are expressed as mean ± standard deviation, while non-normally distributed ones are shown as median (25^th-75^th percentiles). Categorical variables were expressed as frequencies (percentages). Group differences are evaluated using analysis of variance (ANOVA) and Kruskal-Wallis test for normally and non-normally distributed continuous variables, and Chi-square tests for categorical variables. The extent of missing baseline data for study variables is detailed in Table S2 in the Supplement. MissForest was predominantly employed for imputing these missing values.

Kaplan-Meier curves were used to visualize survival for the primary outcome of all-cause mortality. For all non-fatal secondary outcomes, cumulative incidence function curves were generated to appropriately visualize event probabilities. The association between the SMurf burden group and primary and secondary outcomes is primarily evaluated using Cox proportional hazards models, utilizing SMurf-less as the reference. The proportional hazards assumption for all outcomes was verified using Schoenfeld residuals, with global tests indicating no substantial violation. Four models are developed in the study: Model 1: Unadjusted. Model 2: Adjusted for key demographic and clinical confounders, including age, sex, eGFR, alcohol consumption, and a comprehensive list of medical histories. Model 3: Further adjusted for most of baseline medication use, importantly including the specific type of OACs. Model 4: Additionally adjusted for anti-hypertensive agents, anti-diabetic agents, and statins. The results are presented as hazard ratio (HR) and 95% confidence intervals (CI).

This study categorized specific causes of death based on baseline SMurf distribution into circulatory system diseases, respiratory system diseases, cancer, and other specific causes.¹⁵ Subgroup analyses were conducted to explore the association between the SMurf burden group and all-cause mortality across different population strata, including age, sex, eGFR, and various health conditions and medication use. A series of sensitivity analyses were performed to confirm the robustness of our findings. In particular, to rigorously control for baseline confounding and approximate causal inference, a propensity score-based inverse probability of treatment weighting (IPTW) analysis was conducted for the primary outcome. Simultaneously, to account for the high mortality rate as a competing event, Fine-Gray proportional sub-distribution hazard models were employed for all non-fatal secondary outcomes. Moreover, to clarify the causal framework regarding baseline medication use (anti-hypertensive agents, anti-diabetic agents, statins), a Directed Acyclic Graph (DAG) was constructed (Figure S2 in the Supplement). Within this framework, these specific medications were primarily conceptualized as markers of unmeasured disease severity (confounding by indication). Accordingly, this assumption was verified by repeating the primary analysis after systematically excluding these treatment variables. Beyond this, additional sensitivity analyses were performed to further assess the impact of controlling specific risk factors on outcomes within the SMurf framework, specifically examining the effects of anti-hypertensive agents in hypertensive patients, anti-diabetic agents in diabetic patients, and statins in hyperlipidemic patients. Finally, to further validate our results, we employed several other approaches, including: utilizing the Fine-Gray proportional sub-distribution hazards model to account for competing risks when assessing cause-specific mortality; excluding extreme populations (those with bleeding or coagulation disorders, hemiplegia, dementia, cancer, peripheral vascular disease, or Hb < 90 g/L) that could disproportionately influence overall mortality; and excluding populations with major nontraditional risk factors such as chronic pulmonary diseases, hyperuricemia, connective tissue diseases, and psychiatric disorders. All statistical analyses were primarily performed using STATA (version 16.0) and R (version 4.1.3). A two-tailed P-value ≤ .05 was considered statistically significant.

Results

Patient Characteristics

A total of 20 782 patients were finally included in this study. The proportions of the SMurf-less, SMurf (1-2), and SMurf (3-4) in the total population were 12.2%, 77.5%, and 10.4% respectively (Figure S3 in the Supplement). The median follow-up times for the three groups were 208, 225, and 225 days, respectively, and the prevalence of SMurf within the population as well as the SMurf (1-2) and SMurf (3-4) was ranked in the order of hypertension, DM, hyperlipidemia, and smoking (Figure S4 in the Supplement). The average age of the population was 68.4 years, with a majority of males (55.8%). Patients in higher SMurf burden groups showed elevated uric acid and lipid levels (TC, TG, LDL-C), along with higher CHA₂DS₂-VASc, HAS-BLED, and DOAC scores. Most patients (82.2%) used NOAC, predominantly rivaroxaban. As the number of SMurfs increased, there was a gradual rise in the proportion of patients with pre-existing comorbidities. Specifically, heart failure and hyperuricemia were more prevalent in the SMurf (1-2), accounting for 37.6% and 18.7%, respectively. There was a gradual rise in the number of patients using additional medications as the number of SMurfs increased, such as anti-platelet agents, NSAIDs, gastric mucosal protective agents, statins, anti-hypertensive agents, and anti-diabetic agents, among others. Baseline characteristics are presented in Table 1.

Table 1.

Patient Characteristics by SMurf Burden Groups at Baseline.

Characteristics	Overall (N = 20 782)	SMurf-Less (N = 2534)	SMurf (1-2) (N = 16 097)	SMurf (3-4) (N = 2151)	P Value
Patient demographics
Age, mean (SD), years	68.4 (10.9)	65.0 (13.6)	69.0 (10.5)	67.6 (9.5)	<.001
Male sex, n (%)	11 590 (55.8)	1453 (57.3)	8895 (55.3)	1242 (57.7)	.022
Current smoking, n (%)	2915 (14.0)	0 (0.0)	1807 (11.2)	1108 (51.5)	<.001
Drinking daily, n (%)	1353 (6.5)	74 (2.9)	983 (6.1)	296 (13.8)	<.001
BMI, median (IQR), kg/m²	25.3 (23.2, 27.8)	23.9 (22.4, 26.3)	25.3 (23.3, 27.9)	26.0 (23.7, 28.4)	<.001
Clinical characteristics
eGFR, median (IQR), mL/min/1.73 m²	85.8 (71.1, 95.2)	89.8 (77.4, 98.7)	85.2 (70.3, 94.6)	85.5 (68.7, 95.7)	<.001
Hb, median (IQR), g/L	134 (122, 147)	135 (121, 148)	134 (121, 147)	135 (122, 149)	.008
LDL-C, median (IQR), mmol/L	2.7 (2.1, 3.3)	2.4 (2.0, 2.8)	2.6 (2.1, 3.2)	3.6 (3.0, 4.1)	<.001
HDL-C, median (IQR), mmol/L	1.1 (0.9, 1.3)	1.1 (0.9, 1.3)	1.1 (0.9, 1.3)	1.1 (0.9, 1.3)	.590
TC, median (IQR), mmol/L	4.3 (3.6, 5.0)	4.0 (3.4, 4.4)	4.2 (3.5, 4.9)	5.4 (4.7, 5.9)	<.001
TG, median (IQR), mmol/L	1.5 (1.1, 2.1)	1.3 (1.0, 1.8)	1.5 (1.1, 2.1)	1.9 (1.4, 2.6)	<.001
Uric acid, median (IQR), umol/L	322.0 (256.0, 391.0)	311.0 (248.0, 378.0)	323.0 (256.0, 393.0)	327.0 (261.0, 394.0)	<.001
Clinical scale, median (IQR)
CHA₂DS₂-VASc score	3 (2, 5)	2 (1, 3)	4 (2, 5)	4 (3, 5)	<.001
HAS-BLED score	1 (0, 2)	0 (0, 1)	1 (0, 2)	1 (1, 2)	<.001
DOAC score	4 (2, 6)	2 (0, 4)	4 (3, 6)	5 (3, 7)	<.001
OAC type, n (%)
Warfarin	3693 (17.8)	496 (19.6)	2831 (17.6)	366 (17.0)	.034
NOAC	17 089 (82.2)	2038 (80.4)	13 266 (82.4)	1785 (83.0)	.034
Specific NOAC type, n (%)
Dabigatran	3203 (15.4)	408 (16.1)	2522 (15.7)	273 (12.7)	<.001
Rivaroxaban	13 849 (66.6)	1629 (64.3)	10 714 (66.6)	1506 (70.0)	<.001
Apixaban	37 (0.2)	1 (0.03)	30 (0.2)	6 (0.3)	.110
Previous history, n (%)
Hypertension	15 902 (76.5)	0 (0.0)	13 784 (85.6)	2118 (98.5)	<.001
Diabetes	5669 (27.3)	0 (0.0)	3959 (24.6)	1710 (79.5)	<.001
Hyperlipidemia	5438 (26.2)	0 (0.0)	3778 (23.5)	1660 (77.2)	<.001
Stroke	3016 (14.5)	134 (5.3)	2443 (15.2)	439 (20.4)	<.001
Venous thromboembolism	713 (3.4)	61 (2.4)	561(3.5)	91 (4.2)	.002
Ischemic heart disease	909 (4.4)	36 (1.4)	740 (4.6)	133 (6.2)	<.001
Peripheral vascular disease	6648 (32.0)	488 (16.1)	5238 (32.5)	1002 (46.6)	<.001
Bleeding or coagulation disease	767 (3.7)	60 (2.4)	595 (3.7)	112 (5.2)	<.001
Hemiplegia	654 (3.1)	27 (1.1)	535 (3.3)	92 (4.3)	<.001
Dementia	498 (2.4)	32 (1.3)	416 (2.6)	50 (2.3)	<.001
Chronic pulmonary disease	3153 (15.2)	196 (7.7)	2543 (15.8)	414 (19.2)	<.001
Gastrointestinal disease	1926 (9.3)	116 (4.6)	1548 (9.6)	262 (12.2)	<.001
Liver disease	5382 (25.9)	398 (15.7)	4258 (26.5)	726 (33.8)	<.001
Anemia	1094 (5.3)	88 (3.5)	875 (5.4)	131 (6.1)	<.001
Cancer	1386 (6.7)	126 (5.0)	1067 (6.6)	193 (9.0)	<.001
Heart failure	7641 (36.8)	780 (30.8)	6054 (37.6)	807 (37.5)	<.001
Hyperuricemia	3763 (18.1)	378 (14.9)	3011 (18.7)	374 (17.4)	<.001
Connective tissue disease	742 (3.6)	79 (3.1)	576 (3.6)	87 (4.0)	.230
Psychiatric disorder	456 (2.2)	37 (1.5)	352 (2.2)	67 (3.1)	<.001
Concomitant medication use, n (%)
Anti-platelet agents	6908 (33.2)	427 (16.9)	5448 (33.8)	1033 (48.0)	<.001
NSAIDs	1212 (5.8)	86 (3.4)	936 (5.8)	190 (8.8)	<.001
PPI/H2RA	3567 (17.2)	233 (9.2)	2820 (17.5)	514 (23.9)	<.001
Glucocorticoid	652 (3.1)	45 (1.8)	498 (3.1)	109 (5.1)	<.001
Amiodarone	731 (3.5)	54 (2.1)	577 (3.6)	100 (4.6)	<.001
Anti-thyroid agents	110 (0.5)	13 (0.5)	81 (0.5)	16 (0.7)	.330
SSRIs	166 (0.8)	12 (0.5)	132 (0.8)	22 (1.0)	.079
Statins	5508 (26.5)	320 (12.6)	4372 (27.2)	816 (37.9)	<.001
Anti-hypertensive agents	8606 (41.4)	517 (20.4)	6854 (42.6)	1235 (57.4)	<.001
Anti-diabetic agents	4433 (21.3)	0 (0.0)	3148 (19.6)	1285 (59.7)	<.001

Abbreviations: BMI, body mass index; eGFR, estimated glomerular filtration rate; Hb, hemoglobin; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; DOAC, direct-acting oral anticoagulants; OAC, oral anticoagulant; NOAC, non-vitamin K antagonist oral anticoagulants; NSAIDs, nonsteroidal anti-inflammatory drugs; PPI/H2RA, proton pump inhibitors/H2 receptor antagonists.

Associations Between SMurf and All-Cause Mortality and the Distribution of Death Reason

The study observed a total of 1831 cases of all-cause death over a maximum follow-up duration of 3 years, representing 8.8% of the study population. Kaplan-Meier curve for all-cause mortality is shown in Figure S5 in the Supplement. Table 2 displays the univariate and multivariate adjusted Cox models for all-cause mortality. Initially, without adjusting other factors, the HR for all-cause mortality were 1.42 (95% CI: 1.20-1.69, P < .001) for SMurf (1-2) and 1.89 (95% CI: 1.54-2.32, P < .001) for SMurf (3-4) compared to SMurf-less. Subsequent adjustments were made for age, sex, eGFR, alcohol consumption, and medical history. After these adjustments, the HR for all-cause mortality were 1.21 (95% CI: 1.01-1.44, P = .035) for SMurf (1-2) and 1.78 (95% CI: 1.44-2.20, P < .001) for SMurf (3-4). Additional adjustments were then performed for baseline medication use, anti-hypertensive agents, anti-diabetic agents, and statins. Following these adjustments, the HR for all-cause mortality were 1.20 (95% CI: 1.01-1.44, P = .041) for SMurf (1-2) and 1.67 (95% CI: 1.33-2.08, P < .001) for SMurf (3-4). The comprehensive results of the final multivariable model were presented in Table S3 in the Supplement, and the results of IPTW analysis also confirmed the robustness of these findings (Table S4 in the Supplement). To further explore this dose-response relationship, we also modeled the SMurf count as a continuous variable. In the fully adjusted model, each additional SMurf was associated with a 22% increase in the risk of all-cause mortality (HR per SMurf: 1.22, 95% CI: 1.14-1.31, P < .001). Furthermore, to provide a more granular view of the risk gradient, we also calculated the hazard ratios for each specific SMurf count (1, 2, 3, and 4) separately, with detailed results presented in Table S5 in the Supplement. This analysis clearly illustrated an incremental increase in risk with each additional SMurf. Additionally, an analysis deconstructing the SMurf burden into its individual components was performed, with the independent prognostic weights for each risk factor detailed in Table S6 in the Supplement.

Table 2.

Associations Between SMurf and All-Cause Mortality.

	SMurf Burden Groups, HR (95%CI)
Model	SMurf-less	SMurf (1-2)	SMurf (3-4)
1. Unadjusted	1[Reference]	1.42 (1.20 to 1.69)	1.89 (1.54 to 2.32)
2. Adjusted for age, sex, eGFR, alcohol consumption, medical history^a	1[Reference]	1.21 (1.01 to 1.44)	1.78 (1.44 to 2.20)
3. Further adjusted for most of baseline medication use^b	1[Reference]	1.22 (1.03 to 1.46)	1.80 (1.46 to 2.23)
4. Further adjusted for anti-hypertensive agents, anti-diabetic agents, and statins.	1[Reference]	1.20 (1.01 to 1.44)	1.67 (1.33 to 2.08)

Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs).

Model 2 adjusted for age, sex, eGFR, drinking daily, stroke, venous thromboembolism, ischemic heart disease, peripheral vascular disease, bleeding or coagulation disease, hemiplegia, dementia, chronic pulmonary disease, gastrointestinal disease, liver disease, anemia, cancer, heart failure, hyperuricemia, connective tissue disease, psychiatric disorder.

Model 3 included all covariates in model 2 plus anti-platelet agents, OAC type, NSAIDs, PPI/H2RA, glucocorticoid, amiodarone, anti-thyroid agents, SSRIs.

Table S7 presents the distribution of specific cause of death, categorized as SMurf-less, SMurf (1-2), and SMurf (3-4). The number of deaths in each category were 140, 1435, and 256 respectively. Among all deaths, the highest proportion (68.7%) was attributed to circulatory system diseases, followed by cancer (14.7%), and the lowest proportion was due to respiratory system diseases (5.0%). The internal cause of death within each category is similar to the overall distribution of cause of death.

Associations Between SMurf and Secondary Outcomes

The study also examined other outcomes such as ICH, IS, GIB, AMI, composite ischemic outcomes, and composite hemorrhagic outcomes (Table 3). Cumulative incidence function curves for all secondary outcomes are shown in Figures S6-S11 in the Supplement. The numbers of these events were 196, 877, 388, 371,1212, and 568 cases respectively. For composite ischemic outcomes, in the unadjusted scenario, the HR for events in SMurf (1-2) and SMurf (3-4) were 1.51 (95% CI: 1.22-1.87, P < .001) and 1.82 (95% CI: 1.40-2.35, P < .001) respectively when using SMurf-less as reference. After full adjustment, the HR for SMurf (1-2) was not statistically significant, while for SMurf (3-4), the HR for composite ischemic events was 1.46 (95% CI: 1.11-1.94, P = .007). In terms of IS, the study found that without adjustment, the HR in SMurf (1-2) and SMurf (3-4) were 1.44 (95% CI: 1.13-1.85, P = .004) and 1.63 (95% CI: 1.21-2.21, P = .002) respectively. Upon further adjustment for demographic variables, clinical parameters, medical history, and medication use, the HR for IS in SMurf (1-2) was 1.22 (95% CI: 0.94-1.57, P = .128), while for SMurf (3-4) it was 1.44 (95% CI: 1.03-1.99, P = .031), showing statistical significance. After adjusting for medical history, the HRs for SMurf (1-2) and SMurf (3-4) in relation to AMI lost their statistical significance.

Table 3.

Associations Between SMurf and Secondary Outcomes.

Event	Event/Total (%)	Model	SMurf (1-2) Versus SMurf-Less	P Value	SMurf (3-4) Versus SMurf-Less	P Value
Composite Ischemic Outcome^a	1212/20 872 (5.8%)
		Model1	1.51 (1.22 to 1.87)	<.001	1.82 (1.40 to 2.35)	<.001
		Model2	1.21 (0.97 to 1.50)	.091	1.44 (1.10 to 1.87)	.008
		Model3	1.20 (0.96 to 1.49)	.109	1.41 (1.08 to 1.84)	.012
		Model4	1.21 (0.97 to 1.51)	.084	1.46 (1.11 to 1.94)	.007
Ischemic Stroke	877/20 782 (4.2%)
		Model1	1.44 (1.13 to 1.85)	.004	1.63 (1.21 to 2.21)	.002
		Model2	1.19 (0.93 to 1.54）	.168	1.33 (0.98 to 1.82）	.072
		Model3	1.19 (0.92 to 1.53）	.178	1.32 (0.96 to 1.81）	.082
		Model4	1.22 (0.94 to 1.57）	.128	1.44 (1.03 to 1.99）	.031
Acute Myocardial Infarction	371/20 782 (1.8%)
		Model1	1.63 (1.09 to 2.43)	.017	2.21 (1.38 to 3.53)	.017
		Model2	1.20 (0.80 to 1.80)	.390	1.60 (0.99 to 2.60)	.055
		Model3	1.18 (0.78 to 1.77)	.436	1.55 (0.95 to 2.51)	.080
		Model4	1.17 (0.77 to 1.76)	.464	1.46 (0.88 to 2.42)	.145
Composite Hemorrhagic Outcome^b	568/20 782 (2.7%)
		Model1	1.41 (1.05 to 1.89)	.024	1.36 (0.93 to 1.98)	.118
		Model2	1.08 (0.80 to 1.47)	.603	1.04 (0.70 to 1.54)	.834
		Model3	1.08 (0.79 to 1.46)	.638	1.02 (0.69 to 1.51)	.924
		Model4	1.05 (0.77 to 1.43)	.757	0.94 (0.62 to 1.42)	.767
Intracranial Hemorrhage	196/20 782 (0.9%)
		Model1	2.20 (1.20 to 4.06)	.011	2.12 (1.02 to 4.39)	.028
		Model2	1.82 (0.98 to 3.37）	.059	1.67 (0.79 to 3.52）	.177
		Model3	1.83 (0.98 to 3.40）	.057	1.67 (0.79 to 3.53）	.179
		Model4	1.85 (0.99 to 3.45）	.053	1.84 (0.85 to 3.97）	.123
Gastrointestinal Bleeding	388/20 782 (1.9%)
		Model1	1.19 (0.85 to 1.67)	.305	1.09 (0.70 to 1.72)	.694
		Model2	0.89 (0.63 to 1.26）	.524	0.85 (0.53 to 1.35）	.483
		Model3	0.89 (0.63 to 1.25）	.494	0.83 (0.52 to 1.32）	.424
		Model4	0.84 (0.59 to 1.20）	.345	0.70 (0.43 to 1.15）	.159

Composite ischemic outcomes were defined as ischemic stroke and acute acute myocardial infarction.

Composite hemorrhagic outcomes were defined as Intracranial hemorrhage and Gastrointestinal bleeding.

Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs). Model 1 was unadjusted. Model 2 adjusted for age, sex, eGFR, drinking daily, stroke, venous thromboembolism, ischemic heart disease, peripheral vascular disease, bleeding or coagulation disease, hemiplegia, dementia, chronic pulmonary disease, gastrointestinal disease, liver disease, anemia, cancer, heart failure, hyperuricemia, connective tissue disease, psychiatric disorder. Model 3 included all covariates in model 2 plus anti-platelet agents, OAC type, NSAIDs, PPI/H2RA, glucocorticoid, amiodarone, anti-thyroid agents, SSRIs. Model 4 included all covariates in model 3 plus anti-hypertensive agents, anti-diabetic agents, and statins.

In terms of composite hemorrhagic outcomes, the HR for SMurf (1-2) and SMurf (3-4) did not show statistical significance compared to SMurf-less in both partially adjusted and fully adjusted scenarios. After adjusting for medical history, the HR for ICH in SMurf (1-2) and SMurf (3-4) lost statistical significance. However, the overall trend remained consistent with the primary outcome. In terms of GIB, the HR for SMurf (1-2) and SMurf (3-4) did not show statistical significance, regardless of adjustment levels. Moreover, after adjustments, the HR values changed direction but remained statistically non-significant. Interestingly, when further adjusting for anti-hypertensive agents, anti-diabetic agents, and statins in both ICH and IS (Model 4), there was a slight increase in the HR for these outcomes in SMurf (1-2) and SMurf (3-4) compared to Model 3. Consistent results were observed for all non-fatal outcomes when accounting for the competing risk of death using Fine-Gray models in Table S8 in the Supplement.

Subgroup Analyses and Sensitivity Analyses

The subgroup analyses indicated that the findings were generally consistent with the main effects across all stratified groups after full adjustment. Furthermore, these analyses revealed a more pronounced association between SMurf and all-cause mortality in patients with eGFR < 60 mL/min/1.73 m² when compared to SMurf-less (P_int < .001), as presented in Table 4.

Table 4.

Subgroup Analysis of the Association Between SMurf and All-Cause Mortality.

		Unadjusted Model		Model 4
Characteristics	Event/Total (%)	SMurf (1-2) versus SMurf-less	SMurf (3-4) versus SMurf-less	SMurf (1-2) versus SMurf-less	SMurf (3-4) versus SMurf-less	P _int
Age
<65	250/7046 (3.6)	1.32 (0.88 to 2.00)	2.16 (1.33 to 3.48)	0.94 (0.61 to 1.45)	1.16 (0.67 to 2.00)	.670
≥65	1581/13 736 (11.5)	1.27 (1.05 to 1.53)	1.73 (1.38 to 2.17)	1.21 (0.99 to 1.47)	1.55 (1.21 to 1.98)	.670
Sex
Male	1034/11 590 (8.9)	1.51 (1.20 to 1.90)	1.95 (1.49 to 2.57)	1.30 (1.03 to 1.65)	1.90 (1.41 to 2.57)	.268
Female	797/9192 (8.7)	1.31 (1.01 to 1.70)	1.80 (1.31 to 2.46)	1.10 (0.84 to 1.44)	1.42 (1.01 to 1.99)	.268
eGFR
≥60	1325/17 910 (7.4)	1.38 (1.14 to 1.68)	1.75 (1.38 to 2.22)	1.27 (1.04 to 1.55)	1.66 (1.29 to 2.15)	<.001
<60	506/2872 (17.6)	1.05 (0.72 to 1.55)	1.47 (0.96 to 2.26)	1.03 (0.70 to 1.53)	1.74 (1.09 to 2.77)	<.001
Daily drinking
No	1676/19 429 (8.6)	1.45 (1.22 to 1.74)	1.96 (1.58 to 2.43)	1.24 (1.03 to 1.49)	1.76 (1.40 to 2.23)	.435
Yes	155/1353 (11.5)	0.71 (0.36 to 1.42)	0.77 (0.37 to 1.61)	0.81 (0.39 to 1.67)	1.07 (0.47 to 2.41)	.435
Stroke
No	1568/17 766 (8.8)	1.46 (1.22 to 1.74)	1.89 (1.52 to 2.35)	1.20 (1.00 to 1.44)	1.61 (1.27 to 2.05)	.514
Yes	263/3016 (8.7)	1.15 (0.59 to 2.25)	1.72 (0.85 to 3.50)	1.19 (0.60 to 2.37)	2.06 (0.97 to 4.38)	.514
Heart failure
No	753/13 141 (5.7)	1.54 (1.17 to 2.03)	2.42 (1.76 to 3.32)	1.36 (1.03 to 1.80)	1.88 (1.33 to 2.65)	.073
Yes	1078/7641 (14.1)	1.23 (0.98 to 1.54)	1.45 (1.10 to 1.90)	1.11 (0.88 to 1.40)	1.48 (1.10 to 1.99)	.073
Anti-platelet agents
No	1183/13 874 (8.5)	1.43 (1.18 to 1.73)	1.82 (1.41 to 2.34)	1.20 (0.99 to 1.46)	1.49 (1.13 to 1.96)	.319
Yes	648/6908 (9.4)	1.54 (1.02 to 2.34)	2.19 (1.41 to 3.42)	1.27 (0.83 to 1.95)	1.98 (1.24 to 3.17)	.319
NSAIDs
No	1671/19 570 (8.5)	1.39 (1.16 to 1.66)	1.79 (1.45 to 2.22)	1.19 (0.99 to 1.42)	1.63 (1.29 to 2.06)	.957
Yes	160/1212 (13.2)	1.73 (0.81 to 3.70)	2.55 (1.12 to 5.77)	1.51 (0.69 to 3.32)	2.42 (0.99 to 5.90)	.957
PPI/H2RA
No	1402/17 215 (8.1)	1.41 (1.17 to 1.71)	1.90 (1.51 to 2.39)	1.23 (1.01 to 1.49)	1.80 (1.40 to 2.31)	.379
Yes	429/3567 (12.0)	1.18 (0.76 to 1.83)	1.42 (0.88 to 2.32)	1.02 (0.64 to 1.61)	1.20 (0.71 to 2.04)	.379

Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs). Model 4 included age, sex, eGFR, drinking daily, stroke, venous thromboembolism, ischemic heart disease, peripheral vascular disease, bleeding or coagulation disease, hemiplegia, dementia, chronic pulmonary disease, gastrointestinal disease, Liver disease, anemia, cancer, heart failure, hyperuricemia, connective tissue disease, psychiatric disorder, anti-platelet agents, OAC type, NSAIDs, PPI/H2RA, glucocorticoid, amiodarone, anti-thyroid agents, SSRIs, anti-hypertensive agents, anti-diabetic agents, and statins.

To test the robustness of our covariate adjustment strategy, we performed sensitivity analyses for the primary outcome (Table S9 in the Supplement). These analyses demonstrated that the association between a higher SMurf burden and all-cause mortality remained strong and statistically significant across all tested combinations of medication exclusion. Notably, the hazard ratios were modestly strengthened when these protective medication classes were excluded from adjustment, further supporting the robustness of our primary finding. Among hypertensive individuals, sensitivity analysis of anti-hypertensive medication use post full adjustment demonstrated a protective effect across most of outcomes compared to non-users besides ICH, albeit without statistical significance (Table S10 in the Supplement). In the diabetic population, sensitivity analysis of anti-diabetic medication use post full adjustment also showed a protective effect for most outcomes, including IS, AMI, composite ischemic outcomes, and ICH, but these effects were not statistically significant. However, there was a significant association between anti-diabetic medication use and higher all-cause mortality (HR: 1.45, 95% CI: 1.16-1.81). For composite hemorrhagic outcomes and GIB, there was a non-statistically significant increase in risk (Table S11 in the Supplement). For individuals with hyperlipidemia, sensitivity analysis revealed that statin use was associated with an increased risk of GIB (HR: 2.39, 95% CI: 1.22-4.67) (Table S12 in the Supplement). The results of competitive risk analysis indicate that SMurf primarily increases the risk of death related to circulatory system diseases (Table S13-S15 in the Supplement). After excluding for extreme populations that could potentially influence overall mortality (Table S16 in the Supplement), as well as baseline populations with nontraditional risk factors (Table S17 in the Supplement), the effect of SMurf on mortality remained consistent with the main findings.

Discussion

The study utilized data from THMDP for a retrospective observational study on non-valvular AF patients receiving oral anticoagulation therapy. Our findings suggest that in Asian populations with non-valvular AF, a higher burden of SMurfs is significantly associated with increased risks of all-cause mortality and ischemic events, particularly ischemic stroke. Subgroup analyses also revealed a more pronounced association between SMurf and all-cause mortality in patients with eGFR < 60 mL/min/1.73 m² when compared to SMurf-less. Furthermore, competitive risk analysis indicated that SMurf primarily increases the risk of death related to circulatory system diseases. The sensitivity analysis for all-cause mortality further confirmed the robustness of the main results.

Patients with AF often present a complex clinical profile, characterized by multiple risk factors and comorbidities, advanced age, frailty¹⁶ and inappropriate polypharmacy. This clinical complexity, also referred to as the clinical complexity of AF patients,¹⁷ necessitates targeted treatment and management of relevant risk factors and comorbidities. Also, the burden of comorbidity and risk factors among patients with AF has risen in the last two decades, as highlighted by a cohort study from the UK.¹⁸ In our study population, the baseline prevalence of hypertension, diabetes, hyperlipidemia, and smoking among AF patients was 76.5%, 27.3%, 26.2%, and 14.0%, respectively. Furthermore, a significant proportion of patients had a history of heart failure, peripheral vascular disease, liver disease, hyperuricemia, chronic pulmonary disease, and stroke. Analyses from the GLORIA-AF registry studies have revealed that AF patients with various patterns of comorbidities and more complex clinical phenotypes are associated with poor prognosis.^19,20 Previous research has also confirmed that in populations with AF, the risk of all-cause death, cardiovascular death, major bleeding, and systemic embolism increases with worsened long-term health conditions, chronic diseases, and comorbidity burden.^21,22 Recent ACC/AHA guidelines have highlighted the significance of risk factors and lifestyle choices in the secondary prevention of AF to avoid negative outcomes. They also stress the importance of multidisciplinary care for individuals with multiple chronic conditions.² Therefore, the management of pre-existing chronic conditions, risk factors, and comorbidities in AF is crucial. Our study uniquely incorporates SMurf into the secondary prevention strategies for AF populations. We investigate how SMurf, typically studied in coronary heart disease (CHD), impact the prognosis of NVAF patients on OAC therapy.

The importance of SMurf in managing CHD populations has been well established. Studies like the Framingham Heart Study have confirmed the impact of these risk factors on vascular health and the subsequent changes in prognosis post-intervention.²³ A recent study in China focusing on acute myocardial infarction further highlights the importance of managing these traditional risk factors in this population,⁹ which emphasizes the critical role of SMurf in determining individual prognosis and stresses the need for timely and effective medical treatment. Furthermore, the “SMurf paradox” may be explained in this study by the possibility that individuals with fewer SMurfs may have worse outcomes due to a wider range of unfavorable and uncontrolled risk factors, as well as suboptimal evidence-based medical management. Previous research has shown limited evidence on the cumulative prognostic impact of SMurf on the prognosis of AF patients. A retrospective observational study from Australia suggested that a combination of traditional and nontraditional risk factors could impact the prognosis of AF patients, including all-cause death and major adverse cardiovascular events (MACE). However, the study did not find a statistically significant association between SMurf and AF prognosis in its primary outcomes. Additionally, the study was limited to a single tertiary center hospital with a small sample size. While nontraditional risk factors were categorized separately, the study only included a few factors (chronic kidney impairment, obstructive sleep apnea, obesity), potentially lacking reliability.²⁴ Prior research has also emphasized the impact of nontraditional risk factors in cardiovascular diseases.²⁵ Our study incorporates several nontraditional cardiovascular risk factors such as chronic pulmonary disease, hyperuricemia, connective tissue diseases, psychiatric disorders, etc, as covariates in the analysis. This is aimed at better controlling for the impact of these risk factors on the study population and elucidating the prognostic significance of SMurf.

In both ICH and IS outcomes, regardless of SMurf (1-2) or SMurf (3-4), the HR values in Model 4 showed a slight increase compared to Model 3. This increase may be attributed to the protective effects of the overall management practices such as BP reduction, glycemic control, and lipid lowering. Therefore, after adjusting for these factors, the HR exhibited a slight elevation. Subgroup analysis revealed a more evident association between SMurf and all-cause mortality in patients with eGFR < 60 mL/min/1.73 m², but the effect of SMurf is diluted in the majority of other subgroups, thus further multidimensional assessments are needed in subsequent studies. As for the interaction between individuals with reduced kidney function and AF is currently intricate. Poor kidney function can lead to increased release of ACE, promoting atrial fibrosis and abnormal electrical conduction.²⁶ Uremic toxins in these patients may affect calcium ion levels in circulatory and myocardial cells, impacting cellular electrical activity.²⁶ Renal anemia and electrolyte imbalances from kidney dysfunction can worsen AF progression.²⁷ Conversely, AF can contribute to further kidney function decline through mechanisms like thrombus formation, microinfarctions, systemic inflammation, cardiac dysfunction, activation of the renin-angiotensin-aldosterone system, and nephrotoxic treatments, ultimately resulting in adverse outcomes.²⁸

Our analysis underscores the overarching importance of the cumulative SMurf burden, while also providing nuanced insights into the individual risk factors within our anticoagulated AF cohort. Hypertension, as the most prevalent SMurf, reaffirmed its critical role. Our sensitivity analysis, though not reaching statistical significance for all outcomes, suggested a beneficial impact of BP-lowering therapy on mortality and ischemic events, aligning with findings from large meta-analyses and the ARREST-AF study which demonstrate that aggressive risk factor management, including BP control, improves AF-related outcomes.^29,30 The associations with diabetes and hyperlipidemia revealed greater complexity. In our analysis, the use of anti-diabetic agents was paradoxically associated with a higher risk of all-cause mortality, and statin use was associated with GIB. Crucially, as visualized in our DAG, these counterintuitive associations should not be interpreted causally, they represent classic confounding by indication. These findings likely reflect underlying clinical complexity in a real-world setting: anti-diabetic use serves as a marker for more advanced diabetes, or relates to treatment legacy effects and the specific risks associated with certain older drug classes, such as sulfonylureas.^31–33 Similarly, the bleeding signal observed with statins was attenuated after excluding patients with pre-existing liver disease and gastrointestinal diseases (Table S18 in the Supplement), highlighting that such risks are likely driven by underlying comorbidities. Furthermore, pharmacological interactions may contribute. Given that the majority of our cohort received rivaroxaban, statins might competitively inhibit the CYP3A4 and P-glycoprotein pathways shared with rivaroxaban, potentially increasing anticoagulant plasma concentrations and contributing to the observed GIB risk. Smoking, another key modifiable factor, is a well-documented independent predictor of thromboembolic events and mortality in AF patients, and studies have shown significant benefits of smoking cessation.^34–36 Our findings reinforce its importance as a critical component of the total risk burden. While these analyses of individual factors provide valuable clinical insights, they should not overshadow the central thesis of our study. The clear, dose-response relationship we observed between the total number of SMurfs and adverse outcomes demonstrates that the cumulative impact of these risk factors is more prognostically powerful than any single component. This finding reinforces the idea that a patient's long-term trajectory is shaped by their overall burden of atherosclerotic risk, shifting the focus from managing isolated conditions to assessing a patient's comprehensive risk profile.

Our findings should be contextualized by the known ethnic differences in AF. Asian patients often exhibit a higher intrinsic risk of ischemic stroke than Western counterparts with similar CHA₂DS₂-VASc scores³⁷ and may experience a greater prognostic impact from certain SMurfs like hypertension.³⁸ Concurrently, Asian populations have a demonstrably higher risk of intracranial hemorrhage with OACs.³⁹ This narrower therapeutic window between ischemia and bleeding underscores the heightened importance of managing the “residual atherosclerotic risk” in this population. Conceptually, while SMurfs are classically recognized as drivers of atherosclerotic disease, their prognostic impact in anticoagulated AF patients likely operates through dual pathways. First, a high cumulative SMurf burden accelerates coronary and vascular progression, which are major independent contributors to all-cause mortality in this cohort. Second, chronic metabolic stress from hypertension, diabetes, and smoking exacerbates systemic inflammation, endothelial dysfunction, and atrial structural remodeling.⁴⁰ These maladaptive changes promote a sustained pro-thrombotic milieu that persists despite oral anticoagulation, thereby driving the significant residual risk of ischemic stroke observed in our high-burden groups. By controlling the SMurf burden, clinicians can mitigate the underlying drive for atherothrombotic events, a strategy of particular importance for Asian patients. Our study provides crucial, large-scale data on the dose-response impact of SMurfs, offering a strong evidence base for prioritizing integrated care in this high-risk group.

A key methodological consideration was our decision to adjust for baseline SMurf-related medications. While our primary rationale was to control for confounding by treating medication use as an indicator of disease severity, these therapies can also act as mediators on the causal pathway. Our extensive sensitivity analyses provided clarity on this issue. The results showed that excluding these protective medications from the model strengthened the association between SMurf burden and mortality, as would be expected. This consistency across different adjustment strategies confirms that the association between a higher SMurf burden and worse outcomes is highly robust and is not dependent on the specific statistical model used.

The primary clinical implication of our study is the urgent need to move beyond an anticoagulation-only focus towards a holistic, integrated care model for AF patients. Our findings provide strong, real-world evidence from a large Asian cohort supporting the “C” (Cardiovascular and comorbidity optimization) pillar of the “Atrial Fibrillation Better Care” (ABC) pathway. By quantifying the substantial “residual risk” driven by the cumulative SMurf burden, our study complements the full spectrum of existing evidence from randomized trials like mAFA-II to observational studies and meta-analyses. While these studies have collectively shown what can be achieved with active management, our study shows what is at stake without it, reinforcing the importance of the ABC framework in routine practice. Streamlining this approach requires practical strategies. Key steps should include: (1) establishing multidisciplinary care teams involving cardiologists, primary care physicians, and other specialists; (2) implementing structured protocols for systematic risk factor assessment and management in AF clinics; and (3) empowering patients through education on lifestyle and treatment adherence. Leveraging digital health tools can further support these efforts. Adopting such an integrated model is crucial for addressing the patient's total cardiovascular risk and improving long-term outcomes.

This study has some limitations. Firstly, the research primarily utilized data from the THMDP, predominantly based on the Chinese population. The findings may not be generalizable to other population. Secondly, the medical history and medication data relied on ICD codes from electronic health records, which may have limited sensitivity and specificity, potentially introducing misclassification bias. Specifically, hyperlipidemia was defined primarily by baseline lipid levels to reflect the physiological status. However, this approach may underestimate the prevalence of hyperlipidemia among patients with well-controlled lipids on statins. Similarly, smoking status may be subject to under-reporting. Thirdly, the diagnosis of AF was also based on ICD codes without detailed classification of different types of AF. Fourthly, the study is retrospective and observational, therefore, causation cannot be determined. Furthermore, the observed hazard ratios represent a modest relative risk increase, suggesting that the findings should be interpreted as robust associations rather than definitive causality. Lastly, despite efforts to control for confounding factors, the study remains susceptible to unmeasured confounders. Furthermore, we were unable to assess patient adherence to OACs or account for dynamic treatment adjustments (eg, discontinuation or switching of therapies) during the follow-up period, which could influence outcomes. Our analysis also focused on the cumulative number of SMurfs, without investigating potential synergistic or interactive effects between specific combinations of these risk factors. Finally, our sensitivity analyses of concomitant medications were based on baseline use and could not capture subsequent changes in these treatments over time.

Conclusion

Our study demonstrates that in Asian patients with non-valvular AF on OAC therapy, a higher cumulative burden of SMurfs is associated with increased risks of all-cause mortality and ischemic stroke. These findings highlight the substantial residual risk conferred by baseline comorbidities and underscore the potential importance of implementing proactive, multifactorial risk management strategies. While our observational data cannot prove the effectiveness of such interventions, they provide a strong rationale for future trials and for prioritizing comprehensive risk factor control in clinical practice.

Supplemental Material

sj-docx-1-cat-10.1177_10760296261442055 - Supplemental material for Standard Modifiable Cardiovascular Risk Factors and Prognostic Outcomes in Non-Valvular Atrial Fibrillation Patients on Oral Anticoagulants

Supplemental material, sj-docx-1-cat-10.1177_10760296261442055 for Standard Modifiable Cardiovascular Risk Factors and Prognostic Outcomes in Non-Valvular Atrial Fibrillation Patients on Oral Anticoagulants by Linhao Liu, Hangkuan Liu, Pengfei Sun, Geru Aa, Yuanyuan Liu, Zhiqiang Zhang, Qing Yang and Xin Zhou in Clinical and Applied Thrombosis/Hemostasis

Footnotes

Nonstandard Abbreviations and Acronyms

Acknowledgments

We sincerely appreciate all participants of this study for their invaluable contributions, which have been instrumental in advancing this research.

ORCID iDs

Linhao Liu

Hangkuan Liu

Geru Aa

Xin Zhou

Author Contributions

XZ and QY contributed to the conception and design of the study. LL and HL contributed to manuscript drafting, data acquisition, analysis and interpretation of the data. PS, GA, YL, and ZZ contributed to the critical revision of the manuscript for important intellectual content. All gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Key Research and Development Program of China (No. 2022YFF0706000), National Natural Science Foundation of China (82320108001, 82270349, 82321001, 72274133 and 72304205) and Tianjin Key Medical Discipline (Specialty) Construction Project (Grant No. TJYXZDXK-069C).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data analyzed in this study were derived from the Tianjin Health Medical Data Platform (THMDP) and were rigorously de-identified prior to use by the research team to protect patient privacy. However, due to institutional policies of the THMDP, ethical approval stipulations, and overarching privacy safeguards designed to prevent any potential re-identification, the individual-level de-identified dataset underlying this article cannot be made publicly available.

Supplemental Material

Supplemental material for this article is available online.

References

Van Gelder

Rienstra

Bunting

, et al. 2024 ESC guidelines for the management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS). Eur Heart J. 2024;45(36):3314-3414. doi:10.1093/eurheartj/ehae176

Joglar

Chung

Armbruster

, et al. 2023 ACC/AHA/ACCP/HRS guideline for the diagnosis and management of atrial fibrillation: a report of the American College of Cardiology/American Heart Association Joint Committee on clinical practice guidelines. J Am Coll Cardiol. 2024;83(1):109-279. doi:10.1016/j.jacc.2023.08.017

Lip

GYH

. The ABC pathway: an integrated approach to improve AF management. Nat Rev Cardiol. 2017;14(11):627-628. doi:10.1038/nrcardio.2017.153

Guo

Lane

Wang

, et al. Mobile health technology to improve care for patients with atrial fibrillation. J Am Coll Cardiol. 2020;75(13):1523-1534. doi:10.1016/j.jacc.2020.01.052

Chu

Zhang

Gong

, et al. Telemedicine-based integrated management of atrial fibrillation in village clinics: a cluster randomized trial. Nat Med. 2025;31(4):1276-1285. doi:10.1038/s41591-025-03511-2

Romiti

Proietti

Vitolo

, et al. Clinical complexity and impact of the ABC (Atrial fibrillation Better Care) pathway in patients with atrial fibrillation: a report from the ESC-EHRA EURObservational Research Programme in AF General Long-Term Registry. BMC Med. 2022;20(1):326. doi:10.1186/s12916-022-02526-7

Romiti

Pastori

Rivera-Caravaca

, et al. Adherence to the ‘atrial fibrillation better care’ pathway in patients with atrial fibrillation: impact on clinical outcomes-a systematic review and meta-analysis of 285,000 patients. Thromb Haemost. 2022;122(3):406-414. doi:10.1055/a-1515-9630

Hindricks

Potpara

Dagres

, et al. 2020 ESC guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): the Task Force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) Developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. Eur Heart J. 2021;42(5):373-498. doi:10.1093/eurheartj/ehaa612

Gao

Yang

, et al. Number of standard modifiable risk factors and mortality in patients with first-presentation ST-segment elevation myocardial infarction: insights from China Acute Myocardial Infarction registry. BMC Med. 2022;20(1):217. doi:10.1186/s12916-022-02418-w

10.

Khot

Bajzer

, et al. Prevalence of conventional risk factors in patients with coronary heart disease. JAMA. 2003;290(7):898-904. doi:10.1001/jama.290.7.898

11.

Yusuf

Hawken

Ounpuu

, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364(9438):937-952. doi:10.1016/S0140-6736(04)17018-9

12.

Yogasundaram

Dover

Hawkins

, et al. Trends in uptake and adherence to oral anticoagulation for patients with incident atrial fibrillation at high stroke risk across health care settings. J Am Heart Assoc. 2022;11(15):e024868. doi:10.1161/JAHA.121.024868

13.

Levey

Stevens

Schmid

, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604-612. doi:10.7326/0003-4819-150-9-200905050-00006

14.

Fang

Lyu

, et al. China multi-disciplinary expert consensus on diagnosis and treatment of hyperuricemia and related diseases (2023 edition). Chin J Pract Intern Med. 2023;43(6):461-480. [in Chinese]. DOI:10.19538/j.nk2023060106

15.

do Nascimento

Dos Santos

de Moraes Batista

, et al. Cause-specific mortality prediction in older residents of São Paulo, Brazil: a machine learning approach. Age Ageing. 2021;50(5):1692-1698. doi:10.1093/ageing/afab067

16.

Nakamaru

Nishimura

Kumamaru

, et al. Trajectories of frailty and clinical outcomes in older adults with atrial fibrillation: insights from the Shizuoka Kokuho database. Circ Cardiovasc Qual Outcomes. 2024;17(8):e010642. doi:10.1161/CIRCOUTCOMES.123.010642

17.

Corazza

Formagnana

Lenti

. Bringing complexity into clinical practice: an internistic approach. Eur J Intern Med. 2019;61:9-14. doi:10.1016/j.ejim.2018.11.009

18.

Nadarajah

Nakao

, et al. Temporal trends and patterns in atrial fibrillation incidence: a population-based study of 3.4 million individuals. Lancet Reg Health Eur. 2022;17:100386. doi:10.1016/j.lanepe.2022.100386

19.

Romiti

Corica

Mei

, et al. Patterns of comorbidities in patients with atrial fibrillation and impact on management and long-term prognosis: an analysis from the prospective global GLORIA-AF registry. BMC Med. 2024;22(1):151. doi:10.1186/s12916-024-03373-4

20.

Romiti

Proietti

Corica

, et al. Implications of clinical risk phenotypes on the management and natural history of atrial fibrillation: a report from the GLORIA-AF. J Am Heart Assoc. 2023;12(20):e030565. doi:10.1161/JAHA.123.030565

21.

Jani

Nicholl

McQueenie

, et al. Multimorbidity and co-morbidity in atrial fibrillation and effects on survival: findings from UK Biobank cohort. Europace. 2018;20(FI_3):f329-f336. doi:10.1093/europace/eux322

22.

Kotalczyk

Guo

Wang

Lip

GYH

. Impact of multimorbidity and polypharmacy on clinical outcomes of elderly Chinese patients with atrial fibrillation. J Clin Med. 2022;11(5):1370. doi:10.3390/jcm11051370

23.

Andersson

Nayor

Tsao

Levy

Vasan

. Framingham heart study: JACC focus seminar, 1/8. J Am Coll Cardiol. 2021;77(21):2680-2692. doi:10.1016/j.jacc.2021.01.059

24.

Bhat

Gan

GCH

Chen

HHL

, et al. Prognostic implications of traditional and non-traditional cardiovascular risk factor profiles in patients with non-valvular atrial fibrillation. Eur J Clin Invest. 2022;52(9):e13799. doi:10.1111/eci.13799

25.

Wallace

Theou

Kirkland

, et al. Accumulation of non-traditional risk factors for coronary heart disease is associated with incident coronary heart disease hospitalization and death. PLoS One. 2014;9(3):e90475. doi:10.1371/journal.pone.0090475

26.

Basnet

Naeem

Sharma

, et al. Atrial fibrillation ablation in patients with chronic kidney disease: a review of literature. Cureus. 2023;15(10):e46545. doi:10.7759/cureus.46545

27.

Gadde

Kalluru

Cherukuri

Chikatimalla

Dasaradhan

Koneti

. Atrial fibrillation in chronic kidney disease: an overview. Cureus. 2022;14(8):e27753. doi:10.7759/cureus.27753

28.

Suwanwongse

Shabarek

. Does atrial fibrillation increase the risk of developing end-stage renal disease in patients with chronic kidney disease? Cureus. 2020;12(2):e6908. doi:10.7759/cureus.6908

29.

Pinho-Gomes

Azevedo

Copland

, et al. Blood pressure-lowering treatment for the prevention of cardiovascular events in patients with atrial fibrillation: an individual participant data meta-analysis. PLoS Med. 2021;18(6):e1003599. doi:10.1371/journal.pmed.1003599

30.

Pathak

Elliott

Lau

, et al. Aggressive risk factor reduction study for atrial fibrillation implications for ablation outcomes: the ARREST-AF randomized clinical trial. JAMA Cardiol. 2025;10(12):e254007. doi:10.1001/jamacardio.2025.4007

31.

Khunti

Kosiborod

Ray

. Legacy benefits of blood glucose, blood pressure and lipid control in individuals with diabetes and cardiovascular disease: time to overcome multifactorial therapeutic inertia? Diabetes Obes Metab. 2018;20(6):1337-1341. doi:10.1111/dom.13243

32.

Reddy

Zhang

Natarajan

. Epigenetic mechanisms in diabetic complications and metabolic memory. Diabetologia. 2015;58(3):443-455. doi:10.1007/s00125-014-3462-y

33.

Douros

Dell'Aniello

OHY

, et al. Sulfonylureas as second line drugs in type 2 diabetes and the risk of cardiovascular and hypoglycaemic events: population based cohort study. Br Med J. 2018;362:k2693. doi:10.1136/bmj.k2693

34.

Albertsen

Rasmussen

Lane

, et al. The impact of smoking on thromboembolism and mortality in patients with incident atrial fibrillation: insights from the Danish Diet, Cancer, and Health study. Chest. 2014;145(3):559-566. doi:10.1378/chest.13-1740

35.

Choi

Chang

Kim

, et al. Association of smoking cessation after atrial fibrillation diagnosis on the risk of cardiovascular disease: a cohort study of South Korean men. BMC Public Health. 2020;20(1):168. doi:10.1186/s12889-020-8275-y

36.

Lee

Choi

Jung

Han

K-D

Lip

GYH

. Smoking cessation after diagnosis of new-onset atrial fibrillation and the risk of stroke and death. J Clin Med. 2021;10(11):2238. doi:10.3390/jcm10112238

37.

Chao

Liu

Wang

, et al.

Should atrial fibrillation patients with 1 additional risk factor of the CHA2DS2-VASc score (beyond sex) receive oral anticoagulation?

J Am Coll Cardiol. 2015;65(7):635-642. doi:10.1016/j.jacc.2014.11.046

38.

Ueshima

Sekikawa

Miura

, et al. Cardiovascular disease and risk factors in Asia: a selected review. Circulation. 2008;118(25):2702-2709. doi:10.1161/CIRCULATIONAHA.108.790048

39.

Shen

Yao

Brar

Jorgensen

Chen

. Racial/ethnic differences in the risk of intracranial hemorrhage among patients with atrial fibrillation. J Am Coll Cardiol. 2007;50(4):309-315. doi:10.1016/j.jacc.2007.01.098

40.

Goette

Kalman

Aguinaga

, et al. EHRA/HRS/APHRS/SOLAECE expert consensus on atrial cardiomyopathies: definition, characterization, and clinical implication. Europace. 2016;18(10):1455-1490. doi:10.1093/europace/euw161