Atrial fibrillation (AF) is the most common cardiac arrhythmia in clinical practice, significantly increasing the risk of mortality and thromboembolic events, especially ischemic stroke (IS) and systemic embolism (SE) (1, 2). Non-valvular atrial fibrillation (NVAF), defined as AF without mechanical prosthetic heart valves or moderate-to-severe mitral stenosis (3, 4), further amplifies stroke risk, accounting for 20%–30% of all IS worldwide (1, 5). Given the substantial individual and societal implications of stroke, comprehensive management of NVAF is a key component of IS prevention (6).

The United States (U.S.) and European guidelines recommend oral anticoagulants (OAC)—including vitamin K antagonists like Warfarin and direct oral anticoagulants (DOACs, such as dabigatran, rivaroxaban, apixaban, and edoxaban)—to reduce IS and SE risks in patients with NVAF. Although their efficacy and safety have been demonstrated by rigorous randomized clinical trials (RCTs), real-world implementation of these agents frequently encounters hurdles with their notable underutilization (7–10). While recent U.S. studies note an uptrend in OAC use, with 33%–65% of AF patients receiving OACs, substantial variations persist across different populations (11–13). This inconsistency underscores the need for further investigation into OAC utilization patterns among U.S. patients with NVAF.

Moreover, literature regarding the optimal timing for OAC initiation relative to stroke risk in NVAF patients is limited. Although individual studies have reported mixed results (14–17), a recent systematic review and meta-analysis on patients with acute ischemic stroke found that early OAC initiation was associated with a reduced risk of composite outcomes, including ischemic stroke recurrence, intracranial hemorrhage, major bleeding, and all-cause mortality (18). This correlation remains unexplored in patients without prior strokes. In addition, to our knowledge, no prior research has investigated this association in a real-world setting.

This study aimed to describe OAC usage patterns within the first-year post diagnosis of NVAF among patients with no prior OAC use. Additionally, we examined the relationship between the timing of OAC initiation and the one-year risk of all-cause and stroke/SE-specific hospitalizations for patients starting OAC within this timeframe.

Between January 1, 2017, and March 31, 2021, we identified 93,251,105 patients aged ≥18. Of these, 23,148 adults with newly diagnosed NVAF were included (Supplementary Table S3). 11,059 (47.8%) received OAC within one year of diagnosis (OAC users), while 12,089 (52.2%) did not (Non-OAC users). The lowest OAC initiation rate (44.5%) was in 2017, and the highest (52.3%) in 2021.

The mean age was 70 ± 14 years (Table 1, median = 70; 62.7% ≥ 65 years), with 46.9% females, and 80.0% white. The major healthcare coverage was Medicare (63.4%), followed by commercial (21.0%), Medicaid (13.2%), and other, uninsured, or unknown (2.4%). The median CCI was 3. Common comorbidities included coronary artery disease (44.2%) and congestive heart failure (39.7%). The median CHA₂DS₂-VASc score was 4, with 77.8% of males and 85.4% of females having CHA₂DS₂-VASc scores of ≥2 and ≥3, respectively. The median HAS-BLED score was 3, with 60.4% patients having HAS-BLED score ≥3, and 74.1% of them diagnosed at an inpatient setting. Most patients were admitted from non-healthcare facilities (77.4%), came in for an emergency visit (63.8%), and were discharged to home (64.1%). The characteristics of hospitals, where patients received their NVAF diagnosis, varied (Table 2). Most patients were treated in urban (86.7%), non-teaching hospitals (54.7%), with bed size of 1–299 (38.8%).

When stratified by OAC users and non-OAC users, age, sex, and race distributions were similar between the two groups. Compared to non-OAC users, a lower proportion of OAC users were Hispanic or Latino and on Medicaid, while a higher percentage had commercial payers (p-value < 0.0001). In terms of comorbidities, OAC users generally had a higher prevalence of cardiovascular disease and risk factors than non-OAC users, such as coronary artery disease (46.2% vs. 42.4%), congestive heart failure (45.0% vs. 34.9%), diabetes (40.8% vs. 35.5%), obesity (40.2% vs. 31.4%), vascular disease history (36.2% vs. 34.2%), cerebrovascular disease (24.9% vs. 22.3%), myocardial infarction (21.0% vs. 19.5%), and stroke/SE (13.0% vs. 9.2%). In contrast, non-OAC users had a higher prevalence of malignancy (14.3% vs. 11.5%) and dementia (14.7% vs. 10.7%) than OAC users. The CHA₂DS₂-VASc Score was higher among OAC users, while the HAS-BLED score was 3 in both groups. A larger proportion of OAC users vs. non-OAC users (were diagnosed with AF in the inpatient setting (79.4% vs. 69.4, respectively, p-value < 0.0001), with higher emergency admissions of (68.2% vs. 59.8%, respectively), and higher elective admissions (18.4% vs. 27.0%, respectively, p-value < 0.0001). Compared to non-OAC users, more OAC users were discharged to home (p-value < 0.0001). Non-OAC users were more frequently discharged to hospice or were documented as expired than OAC users (p-value < 0.0001).

Among OAC users, 74.9% initiated OAC during the index visit, with the remaining 25.1% initiating during the one-year follow-up (Table 3). When stratified by diagnosis visit care setting, 84.3% of patients in an inpatient setting initiated OAC during the index visit, compared to 39.0% of patients in an outpatient setting. Among patients initiating OAC during follow-up, the median time to OAC initiation following diagnosis was 94 days, with no significant differences between those diagnosed in inpatient and those in outpatient settings. Apixaban was the most commonly used OAC (62.2%), followed by rivaroxaban (23.2%), warfarin (15.8%), and dabigatran (3.1%). Throughout this follow-up period, a total of 3,695 (33.4%) OAC users experienced all-cause hospitalization, of which 177 were due to stroke/SE (Table 4). Among those initiating OAC at the index visit, the 12-month risk was 29.2% for all-cause hospitalization and 1.3% for hospitalization due to stroke/SE. In contrast, among those initiating OAC during follow-up, 1,274 (45.9%) and 73 (2.6%) experienced all-cause and stroke/SE hospitalization, respectively. The adjusted odds ratio for all-cause hospitalization was 0.35 (95% CI: 0.32–0.39, p-value < 0.0001) and, for hospitalization due to stroke/SE, it was 0.34 (95% CI: 0.24–0.47, p-value < 0.0001), comparing those initiating OAC at the index visit to those during the follow-up period. When stratified by diagnostic visit care setting, both all-cause hospitalization and hospitalization due to stroke/SE remained lower for those initiating at the index visit compared to during the follow-up period (Inpatient: adjusted OR (95% CI): all-cause hospitalization 0.31 (0.27–0.35); hospitalization due to stroke/SE 0.36 (0.25–0.52); Outpatient: adjusted OR (95% CI): all-cause hospitalization 0.51 (0.42–0.63); hospitalization due to stroke/SE 0.18 (0.05–0.61)). The observed trend of a lower hospitalization rate in those initiating OAC at the index visit compared to during follow-up remained after excluding warfarin users (Supplementary Table S4), stratifying by stroke history (Supplementary Table S5), including additional variables in the model (Supplementary Table S6), and among patients with at least one refill (Supplementary Table S7). Among OAC users, the primary diagnoses leading to all-cause hospitalizations during the follow-up period included atrial fibrillation (20.0%), sepsis (12.3%), hypertensive heart and chronic kidney disease (9.4%), hypertensive heart disease (7.3%), and acute kidney failure (4.5%), as detailed in Table 5.

Using real-world data, this study found that, among patients newly diagnosed with NVAF, 47.8% initiated OAC initiation within one year. It appears that patients with malignancy and dementia were associated with lower odds of initiating OAC within one year post diagnosis. Additionally, for patients who initiated OAC within one year post AF diagnosis, an early initiation was associated with lower risk of all-cause and stroke/SE-specific hospitalization.

The recommendation for using OAC in AF patients has been well-established for years. Both the 2019 American Heart Association/American College of Cardiology/Heart Rhythm Society (4) and the 2020 European Society of Cardiology guidelines (29) have similar recommendations for managing AF, emphasizing the use of the CHA₂DS₂-VASc score for stroke risk stratification. For patients with higher scores (≥2 for men, ≥3 for women), there is a consensus recommendation to initiate oral anticoagulant (OAC) therapy to mitigate the stroke risk, while for those at intermediate risk (1 for men, 2 for women), OAC prescription remains a consideration. For low-risk patients (0 for men, 1 for women), it is reasonable to omit anticoagulant therapy, given their relatively low risk of stroke. Additionally, both guidelines highlight the importance of balancing potential stroke benefits against the elevated bleeding risk associated with OAC use. Despite these guidelines, OAC use in the AF population remains suboptimal. One study demonstrated a modest increasing trend in OAC initiation from 2010 through 2020 in the Medicare Advantage population, with a peak of 32.9% in 2020 for newly diagnosed AF patients with elevated stroke risk (CHA₂DS₂-VASc Score ≥2 for men, CHA₂DS₂-VASc Score ≥3 for women) (11). Another study within the fee-for-service Medicare population between 2013 and 2017 showed a 48.7% OAC utilization rate in patients with a CHA₂DS₂-VASc Score of ≥2 (12). Furthermore, a study from 2011–2020 in adults aged over 40 with AF across 88 US health systems reported a slight increase in OAC usage over the years, from 56.3%–64.7%. (13) Our study found only 47.8% OAC-naïve NVAF patients initiated OAC within one year following the AF diagnosis, with the rate of OAC initiation among patients with elevated stroke risk (CHA₂DS₂-VASc Score ≥2 for men, CHA₂DS₂-VASc Score ≥3 for women) at 50.0%. These findings suggest variation in OAC usage rates across different populations. In addition, despite the general upward trend observed in recent years, these findings highlight the need for enhanced strategies to improve adherence to guideline-recommended OAC use in AF patient, aiming for optimal stroke risk management. Notably, significant disparities in OAC prescription rates have been observed among high-risk AF patients managed by primary care clinicians within the same regional health system (30). Implementing targeted interventions, such as an email-based educational initiative directed at primary care clinicians managing high-risk AF patients, has been shown to effectively increase OAC prescription rates and improve overall stroke prevention in this population (30).

In addition, some patients with low CHA₂DS₂-VASc scores were also found to be taking OAC (36.1% for men with a CHA₂DS₂-VASc score of 0, and 31.9% for women with a CHA₂DS₂-VASc score of 1). It is possible that CHA₂DS₂-VASc scores are underreported in this study. Our look-back period was 365 days, which may have led to missing vascular events that occurred earlier. Additionally, CHA₂DS₂-VASc scores were assessed at baseline and could have increased during the follow-up period, during which OAC might have been initiated. Despite these factors, the current finding aligns with previous studies that also reported higher-than-expected OAC use in populations with low CHA₂DS₂-VASc scores. A study in Australia reported that 25.1% of AF patients with low stroke risk (CHA₂DS₂-VASc score of 0 for males and 1 for females) had a record of OAC prescription within 60 days of AF diagnosis (31). Another study in the US using the PINNACLE registry found OAC use in 31.1% of AF patients with a CHA₂DS₂-VASc score of 0 and 34.6% with a CHA₂DS₂-VASc score of 1 (32). Further research is necessary to better understand benefits and potential risks of OAC use in patients with low CHA₂DS₂-VASc scores.

Studies have investigated demographic and clinical characteristics influencing the OAC use, reporting that factors such as older age, female, Black race, and certain comorbidities including dementia, frailty, anemia, bleeding were associated with decreased odds of OAC initiation (11, 12). In contrast, histories of cardiovascular diseases, including stroke, SE, and obesity, were associated with increased likelihood of OAC prescription. Interestingly, there was no substantial difference in CHA₂DS₂-VASc score and HAS-BLED scores between OAC users and non-users. The current study aligns with these observations. Despite similar demographic characteristics between OAC users and non-users, a distinct comorbidity profile was observed: OAC users showed a higher prevalence of cardiovascular disease and risk factors, whereas non-OAC users had a higher prevalence of malignancy and dementia. The CHA₂DS₂-VASc score was slightly higher among OAC users, with no notable difference in HAS-BLED scores between the groups. These insights underline the complexity of decision-making in anticoagulant therapy and highlight the necessity for a nuanced approach that considers individual patient profiles beyond risk scores alone. This could be pivotal in improving OAC utilization where it is most needed, potentially optimizing therapeutic outcomes in atrial fibrillation management.

The current study emphasized that initiating OAC early might be associated with a decreased risk of all-cause hospitalization and hospitalization due to stroke/SE. Although few studies have explored the association between the timing of OAC initiation and hospitalization risk, this observation is in line with findings from a 3-year follow-up study, which demonstrated that OAC use was associated with a 7% lower risk of all-cause hospitalization compared to non-use among Medicare beneficiaries with AF (33). Although studies specifically examining NVAF patients without prior stroke are limited, there is substantial literature assessing OAC timing for those with NVAF and concurrent stroke (14–17). These studies largely indicated similar safety and effectiveness regardless of whether OAC therapy initiates earlier or later. A meta-analysis comprising six cohort studies and two randomized controlled trials reported no significant difference in the safety and efficacy of OACs initiated within one to two weeks following an acute ischemic stroke (16). Consistent with these findings, the Early vs. Late Initiation of Direct Oral Anticoagulants in Post-ischemic Stroke Patients with Atrial Fibrillation (ELAN) trial showed no significant difference in the combined outcomes, including recurrent ischemic stroke, systemic embolism, major bleeding, symptomatic intracranial hemorrhage, or vascular death, within 30 days of starting early vs. later direct oral anticoagulant (DOAC) administration in patients with NVAF and acute ischemic stroke (OR 0.70, 95% CI 0.44–1.14) (14). Interestingly, a lower risk was observed at the 90-day interval (OR 0.65, 95% CI 0.42–0.99) (14). Other studies, including the TIMING study, a registry-based randomized controlled noninferiority trial, have similarly found no substantial association between the timing of OAC initiation and stroke outcomes following an acute ischemic stroke (15, 17). In the current study, stratified by stroke history, NVAF patients would benefit from initiating OAC earlier regarding stroke/SE with or without stroke history. These collective insights suggest that while the early initiation of OACs might not necessarily amplify adverse outcomes, its advantages, especially in patients without prior stroke, are yet to be conclusively determined. This highlights an evident gap in knowledge and the pressing need for further investigations into the timing of OAC initiation in NVAF patients without a stroke history.

It is noteworthy that the majority of hospitalizations during the follow-up period were not due to stroke or SE. The most prevalent primary diagnosis remains to be atrial fibrillation, followed by sepsis, heart and kidney diseases, and pulmonary diseases. This pattern suggested that patients with NVAF may have a heightened risk of comorbidities and other underlying factors contributing to their overall likelihood of hospitalization. These observations highlight the complexity of the NVAF population and emphasize the urgent need for comprehensive management strategies to effectively mitigate their stroke risk.

This study is not without limitations. Due to its retrospective observational design, it can identify associations but cannot establish causality. There may be residual confounding even after adjustment, such as the subjective nature of anticoagulant selection by physicians, who decide both when and what specific type of OAC to use. Additionally, there is a possibility that patients who started OAC earlier were more health-conscious, which could skew results. Additionally, the reliance on administrative data introduces potential constraints, such as the possibility of coding errors and data incompleteness. For example, HAS-BLED score has labile INR, which is not available in the data, and while NVAF has different forms, their characteristics (e.g., diagnostic modalities and whether it is clinical or subclinical, or with or without symptoms) are not available in this data. Our study utilizes hospital administrative data and closed claims data from a real-world database, which serve as proxies for OAC utilization but may not accurately reflect actual usage or the exact date when the patient started taking the OAC. However, sensitivity analysis among patients with at least one refill of OAC showed similar results, suggesting that the findings are robust among patients likely to be actually taking the OAC. Moreover, this study does not assess OAC discontinuation patterns, such as cessation after successful AF ablation, which may influence outcomes. Despite these limitations, this study, using real-world data, described the recent trends in OAC utilization among newly diagnosed NVAF patients. Moreover, it is the first study to investigate the association between the timing of OAC initiation and the rates of all-cause and stroke/SE-specific hospitalizations, using real-world data. Future studies are needed to further investigate drug adherence, different types of NVAF and their characteristics, and OAC dosing to better assess their impact on the association between OAC and hospitalization in the NVAF population. Future studies should further explore drug adherence, NVAF subtypes and characteristics, and OAC dosing to better understand their impact on hospitalization risks. Additionally, research should evaluate whether healthcare interventions, such as closer follow-up and increased awareness, modify this association. Lastly, assessing follow-up period characteristics, such as fall risk and hemorrhagic events, could provide further insights into patient outcomes across treatment groups.

In conclusion, this study demonstrated the underutilization of OAC among patients with NVAF, despite existing guideline recommendations. Furthermore, our findings suggest that early initiation of OAC therapy could be associated with decreased rates of hospitalization. The findings underscored a compelling need for personalized, comprehensive management strategies for NVAF patients, taking into account not only stroke risk but also their comorbidity profile. The observed correlation between the timing of OAC initiation and reduced hospitalization rates warrants further investigation, especially in patients without stroke history.