2024 Volume 47 Issue 7 Pages 1383-1391
The global prevalence of atrial fibrillation (AF) is rising, paralleling increased life expectancy. Early rhythm control benefits AF management. However, in low-risk, often asymptomatic, AF patients, anticoagulant monotherapy is the selected treatment, aligning with current guidelines. However, early AF progression in these low-risk individuals is not well-understood. Thus, this study aims to: 1) determine the proportion of low-risk AF patients who worsen within a year of initial AF diagnosis and 2) identify risk factors such treatment transitions. We analyzed data from 18623 AF patients, spanning January 2005 to June 2017. Low-risk patients were those on anticoagulant monotherapy ± rate control, following the JCS/JHRS 2020 Guideline on Pharmacotherapy of Cardiac Arrhythmias. We defined 2 patterns of treatment transition for 1) initiating ablation or antiarrhythmic drug therapy and 2) solely using antiarrhythmic drugs. This retrospective cohort study was employed a 12-month study, following a 6-month screening period. We included 1874 patients for all rhythm control (analysis 1) and 1503 for only medication-based control (analysis 2). The primary endpoint, treatment transition of AF under monotherapy, occurred in 28.4% of patients in analysis 1 and 10.8% in analysis 2. Risk factors common to both scenarios were male gender, baseline rate control drug use, and rivaroxaban selection, as identified by multiple logistic regression. These findings suggest a higher AF treatment transition trend in patients starting rivaroxaban, calling for further research. The study highlights the importance of informed early rhythm control initiation decisions in clinical settings.
Atrial fibrillation (AF), a potent risk factor for stroke, heart failure, and high mortality, is reported to be present in approximately 1% of the total population in Japan.1) Atrial fibrillation patients present a high risk of stroke without adequate therapy. Globally, the cost burden for patients with AF is estimated at 40 billion dollars. The prevalence of AF is projected to rise, reaching approximately 18 million individuals in Europe by 2060 and between 6 and 12 million in the U.S.A. by 2050.2–5)
The standard treatment for AF involves (1) anticoagulant therapy and/or rate control and (2) rhythm control, which may include medication or catheter ablation.6–9) The proportion of the patients who were selected for anticoagulation monotherapy was reported to be approximately 47–52%.10,11) On the other hand, the proportion for rate control or rhythm control in addition to anticoagulant therapy was reported at 32.3 and 9.4%, respectively.11) Notably, the low percentage of patients undergoing rate or rhythm control therapy can be attributed to the fact that approximately 70% of individuals with AF do not exhibit symptoms of arrhythmia or tachycardia.11) The symptoms caused by AF lead to discomfort, including palpitations, chest pain, and dizziness. These symptoms significantly lower patients’ QOL and reduce their ability to concentrate.12–15) Despite the success of anticoagulant therapy in achieving proper anticoagulation, annual stroke reports still approximate 1.5%.16) Importantly, catheter ablation has a chance to cure AF in approximately 70% of AF patients.17) Although early rhythm control reduces mortality, a significant proportion of patients with AF do not opt for this approach. This is often owing to the lack of overt symptoms in early stages of AF or to patients’ personal decisions to avoid intensive treatment based on their own lifestyle or reasons for hesitation, as previously discussed.18) Moreover, the progression from paroxysmal to nonparoxysmal or from persistent to permanent AF has been observed at a rate of 4.9–5.8 per 100 person-years, regardless of whether the patients are symptomatic or asymptomatic.19) These observations are highlighting the critical need to focus on asymptomatic individuals and needs for providing information on the requirement for early rhythm control in patients with AF who have chosen anticoagulant monotherapy. However, the risk factors for AF treatment transition in patients who initially receive only anticoagulant therapy without rhythm control remain uncertain. Therefore, our study aims to determine the rate of treatment transition within 1 year among patients with low-risk AF at the initial diagnosis, utilizing extensive claims data from Japan.
We utilized the Japan Medical Data Center (JMDC) database, which contains anonymized data of patients. The cumulative dataset comprises approximately 16 million subjects (inpatients, outpatients, and pharmacy claims) from approximately 90% of all hospitals in Japan. All patients in the JMDC database have “social insurance” that covers the working members and their families.20,21) As of July 2023, the database represented approximately 13% of the Japanese population.
Patients’ Identification ProcessThis study sought to ascertain the risk of treatment transition in low-risk AF patients diagnosed for the first time. Consequently, we categorized low-risk patients with AF based on their treatment strategy, specifically between step 3 (anticoagulant monotherapy) and step 4 (anticoagulant ± rate control) as delineated by the “5-steps treatment of AF” in the “JCS/JHRS 2020 Guideline on Pharmacotherapy of Cardiac Arrhythmias,” last updated on October 13, 2023.22)
In our study, we analyzed a cohort of 18623 patients with a history of receiving warfarin (WF) or direct oral anticoagulants (DOACs) in association with AF. These patients were among approximately three million social insurance participants identified from the JMDC database between January 2005 and June 2017. We excluded several patient groups from this cohort: first, those diagnosed with AF within a 6-month screening period to obtain new users of anticoagulants and a subsequent 12-month observation period to calculate the percentage in the study cohort (10294 individuals); second, patients who underwent catheter ablation, heart valve surgery, coronary artery disease operations, or pacemaker-related procedures during the screening period (168 patients), to eliminate those without a primary diagnosis of AF; third, individuals with a history of deep vein thrombosis (International Classification of Diseases-10 codes: ICD10 code for S72) as indicated by ICD-10 codes, or those who had undergone surgeries related to artificial joints (excluding elbow joints) and hip joint procedures during the screening period (4 patients); fourth, patients who received early rhythm control during the observation period (1963 individuals), to exclude those with a more advanced AF treatment strategy; fifth, those under 40 years of age (543 patients) to avoid AF with genetic background23); and sixth, those with medium or high CHADS2 scores (2009 patients) to focus on low-risk AF patients (Supplementary Tables 5, 6).24,25)
To assess the extent of AF treatment transition in the early stages among low-risk patients, we conducted two specific analyses. Analysis 1 included 1874 patients, and Analysis 2, after excluding patients who underwent catheter ablation during the observation period (371 patients) from analysis 1, comprised 1503 patients. These analyses are depicted in Fig. 1.
WF, warfarin; DOAC, direct oral anticoagulants; DVT, deep-vein thrombosis.
The primary endpoint of this study was to assess the proportion of treatment transitions in AF patients within 12 months in both analyses 1 and 2. The secondary endpoint aimed to elucidate the risk factors for AF treatment transition in the study patients within one year from the initial AF diagnosis, particularly focusing on low-risk AF patients in both analyses 1 and 2.
Definitions of This StudyThe definition for index month was the same calendar month at the time of starting anticoagulant therapy. We defined early rhythm control as patients who underwent treatment with antiarrhythmic drugs or ablation within the same calendar month at the initiation of anticoagulant drugs. The treatment transition of AF was defined according to the previously mentioned 5-step treatment strategy, which involves advancing to step 5—required rhythm control—from step 3 or 4 in the absence of a history of rhythm control among the study patients (Supplementary Table 2). To assess the robustness of this approach, we established two definitions for the treatment transition of AF. In analysis 1, treatment transition was defined as the introduction of an antiarrhythmic drug and/or catheter ablation in patients (Supplementary Fig. 2). Conversely, for analysis 2, a more stringent criterion was applied by introducing an antiarrhythmic drug after excluding patients who underwent catheter ablation therapy (Supplementary Fig. 1). This distinction in analysis 2 aimed to mitigate the potential overestimation of AF treatment transition rates, which could arise from conflating different scenarios of catheter ablation therapy. These scenarios include patients awaiting catheter ablation due to lifestyle considerations or physician advisement based on patient preferences, such as hesitation or fear towards the procedure.22) To avoid overestimation, we have shown the results for the proportion for AF treatment transition as providing the percentage between 2 scenarios in this study.
The definition of initiating antiarrhythmic drug therapy involves two or more sequential prescriptions of an antiarrhythmic drug as a new user, or repeated prescriptions with an interval of 90 d or more, as in scenarios such as “pill-in-the-pocket” for first-time patients.26) Catheter ablation and other interventions were identified using standardized medical codes in Japan (Supplementary Table 1, Medical Practice). Anticoagulant drugs were categorized by amalgamating the name of each drug with its corresponding WHO-Anatomical Therapeutic Chemical (ATC) code, namely warfarin, apixaban, dabigatran, rivaroxaban, and edoxaban (Supplementary Table 1, Anticoagulant). Baseline comorbidities were classified as hypertensive diseases, diabetes mellitus, stroke, chronic kidney disease, heart failure, dyslipidemia, transient ischemic attack, sleep apnea syndrome, chronic obstructive pulmonary disease, hyperthyroidism, and asthma (Supplementary Table 1, Medical History). The definition of antiarrhythmic drugs was based on ATC codes and includes groups I and III in the Vaughan–Williams Classification (Supplementary Table 2). The Charlson Comorbidity Index score assessment was conducted in line with the specified reference (Supplementary Table 3).27)
Statistical AnalysisUnivariable analyses, including the Mann–Whitney U-test and chi-square test, were conducted. Additionally, logistic regression analysis was employed as a multivariable analysis to compare the groups “with treatment transition” and “without treatment transition” over the 12-month observation periods. These analyses were conducted using baseline demographics data from 1874 patients (analysis 1) and 1503 patients (analysis 2).
The risk factors for AF treatment transition were assessed using rate control (beta blocker, calcium channel blocker), baseline age (older individuals are at increased risk),1,28) sex (males are at elevated risk),28,29) hypertension,30–32) heart failure,33,34) coronary artery disease,31,35,36) diabetes mellitus,37,38) and baseline comorbidities clinically related to AF and Charlson comorbidity index.27,39)
Data are expressed as medians with ranges or means ± standard deviations. Data analysis was performed using JMP® 17 (SAS Institute Inc., Cary, NC, U.S.A.).
Ethics ApprovalThe commercially available JMDC database used in this study comprises anonymized information processed based on Japan’s Personal Information Protection Law, and individual informed consent is not required for the provision and use of this information. In addition, according to the ethical guidelines for clinical research in Japan, review by an ethics committee is not required for studies using anonymized information. Therefore, no informed consent was obtained for this study because the patients’ data were anonymized before being accessed.
In our investigation, the patient backgrounds between analyses 1 and 2 were not observed to have notable differences. Specifically, in analysis 2, which entailed a robust assessment of treatment transition, warfarin emerged as the predominant anticoagulant, prescribed to 48% of the eligible participants (male/female ratio: 1131/372, average age: 56.2 ± 8.3 years). Following warfarin, the usage rates of other anticoagulants were as follows: rivaroxaban at 20%, dabigatran at 12%, and apixaban and edoxaban each at 10% (as detailed in Table 1). Regarding the Charlson comorbidity index, in analysis 2, the distribution was 47, 41, 8, and 4% for low, medium, high, and very high, respectively. This distribution was similar in analysis 1, with proportions of 48, 42, 7, and 3%, respectively. The prevalent comorbidities identified were hypertensive diseases in 502 individuals (33.4%), diabetes mellitus in 85 individuals (5.6%), stroke in 1 individual (0.1%), chronic kidney disease in 15 individuals (1.0%), heart failure in 61 individuals (4.1%), and dyslipidemia in 255 individuals (17.0%). Finally, the proportion of patients receiving rate control medication at the index month was 37% in analysis 1 and 36% in analysis 2, as shown in Table 1.
| Analysis 1 | Analysis 2 | |
|---|---|---|
| Number (male/female) | 1874 (1467/407) | 1503 (1131/372) |
| Age, year [mean (S.D.)] | 55.7 [8.1] | 56.2 [8.3] |
| Anticoagulant [n] (%) | ||
| Warfarin | 774 (41%) | 714 (48%) |
| Apixaban | 240 (13%) | 151 (10%) |
| Dabigatran | 288 (15%) | 187 (12%) |
| Rivaroxaban | 413 (22%) | 307 (20%) |
| Edoxaban | 159 (9%) | 144 (10%) |
| Charlson comorbidity index (n) (%) | ||
| Low | 892 (48%) | 711 (47%) |
| Medium | 793 (42%) | 623 (41%) |
| High | 132 (7%) | 113 (8%) |
| Very high | 57 (3%) | 56 (4%) |
| Comorbidities (n) (%) | ||
| Hypertensive diseases (I10–15) | 577 (30.7%) | 502 (33.4%) |
| Diabetes mellitus (E10–14) | 99 (5.3%) | 85 (5.6%) |
| Stroke (I64) | 1 (0.05%) | 1 (0.1%) |
| Chronic kidney disease (N189) | 16 (0.9%) | 15 (1.0%) |
| Heart failure (I509) | 99 (5.3%) | 61 (4.1%) |
| Dyslipidemia (E785) | 309 (16.5%) | 255 (17.0%) |
| No. of patients with rate control (n) (%) | 697 (37%) | 547 (36%) |
S.D., standard deviation; n, number.
In analysis 1, 533 out of 1874 patients (28.4%) were estimated to have suffered treatment transitions of AF during a 12-month observation period, representing the highest estimation of treatment transition frequency. Conversely, analysis 2 found a lower estimation, with 10.8% (162/1503) of AF patients experiencing treatment transitions over the same 12-month period.
Regarding the background characteristics in analysis 1, univariable analysis revealed significant differences between patients with and without treatment transitions. Those experiencing treatment transitions were predominantly younger adults (mean age 54 ± 8 years versus 56 ± 8 years, p < 0.0001), male (89.7 versus 73.8%, p < 0.0001), and more likely to have been prescribed rate control medications at baseline (43.9 versus 34.5%, p = 0.0002). Additionally, they had a lower Charlson Comorbidity Index (median 1 [range 0–5] versus 1 [range 0–10], p < 0.0001) and were less frequently on warfarin (22.7 versus 48.7%, p < 0.0001), but more frequently on apixaban (20.5 versus 9.8%, p < 0.0001) compared to those without treatment transitions, as shown in Table 2. Analysis 2 demonstrated a similar trend between groups with and without treatment transitions, as indicated in Table 3.
| Total (n = 1874) | With treatment transition (including catheter ablation) (n = 533) | Without treatment transition (n = 1341) | p-Value | |
|---|---|---|---|---|
| Age, mean (S.D.) | 56 [8] | 54 [8] | 56 [8] | <0.0001 |
| Sex | ||||
| Female | 407 | 55 (10.3%) | 352 (26.2%) | – |
| Male | 1467 | 478 (89.7%) | 989 (73.8%) | <0.0001 |
| Rate control | ||||
| Without | 1177 | 299 (56.1%) | 878 (65.5%) | – |
| With | 697 | 234 (43.9%) | 463 (34.5%) | 0.0002 |
| Anticoagulant | ||||
| Warfarin | 774 | 121 (22.7%) | 653 (48.7%) | <0.0001 |
| Apixaban | 240 | 109 (20.5%) | 131 (9.8%) | <0.0001 |
| Dabigatran | 288 | 128 (24.0%) | 160 (11.9%) | <0.0001 |
| Rivaroxaban | 413 | 155 (29.0%) | 258 (19.2%) | <0.0001 |
| Edoxaban | 159 | 20 (3.8%) | 139 (10.4%) | <0.0001 |
| Charlson comorbidity index, median (range) | 1 [0–10] | 1 [0–5] | 1 [0–10] | <0.0001 |
S.D., standard deviation; n, number.
| Total (n = 1503) | With treatment transition (n = 162) | Without treatment transition (n = 1341) | p-Value | |
|---|---|---|---|---|
| Age, mean (S.D.) | 56 [8] | 55 [8] | 56 [8] | 0.4728 |
| Sex | ||||
| Female | 372 | 20 (12.3%) | 352 (26.2%) | – |
| Male | 1131 | 142 (87.7%) | 989 (73.8%) | <0.0001 |
| Rate control | ||||
| Without | 956 | 78 (48.2%) | 878 (65.5%) | – |
| With | 547 | 84 (51.8%) | 463 (34.5%) | <0.0001 |
| Anticoagulant | ||||
| Warfarin | 714 | 61 (37.7%) | 653 (48.7%) | <0.0001 |
| Apixaban | 151 | 20 (12.3%) | 131 (9.8%) | <0.0001 |
| Dabigatran | 187 | 27 (16.7%) | 160 (11.9%) | <0.0001 |
| Rivaroxaban | 307 | 49 (30.2%) | 258 (19.2%) | <0.0001 |
| Edoxaban | 144 | 5 (3.1%) | 139 (10.4%) | <0.0001 |
| Charlson comorbidity index, median (range) | 1 [0–10] | 1 [0–4] | 1 [0–10] | 0.0135 |
S.D., standard deviation; n, number.
The secondary endpoint, which focused on identifying risk factors for initiating rhythm control intervention in AF patients, was assessed using logistic regression analysis. This analysis revealed the following odds ratios (ORs) and 95% confidence intervals (95% CIs) for various factors: age, with an OR of 0.96 per additional year (95% CI: 0.95–0.97, p < 0.0001); male sex compared to female sex, showing an OR of 2.38 (95% CI: 1.73–3.28, p < 0.0001); patients under rate control therapy, with an OR of 1.29 (95% CI: 1.03–1.61, p = 0.0235); and those treated with different anticoagulants, specifically apixaban versus warfarin (OR: 4.66, 95% CI: 3.34–6.50, p < 0.0001), dabigatran versus warfarin (OR: 4.27, 95% CI: 3.11–5.85, p < 0.0001), and rivaroxaban versus warfarin (OR: 3.21, 95% CI: 2.40–4.29, p < 0.0001). Edoxaban, compared with warfarin, showed an OR of 0.87 (95% CI: 0.51–1.46, p = 0.5838). Furthermore, a lower score on the Charlson Comorbidity Index was associated with reduced odds (OR: 0.89, 95% CI: 0.81–0.99, p = 0.0277) in analysis 1 (Fig. 2). In analysis 2, we consistently identified predictive factors, with notable associations including male sex over female sex (OR: 2.04, 95% CI: 1.25–3.34, p = 0.0045), patients receiving rate control therapy (OR: 1.77, 95% CI: 1.26–2.47, p = 0.0009), and the preference for rivaroxaban over warfarin (OR: 1.77, 95% CI: 1.16–2.69, p = 0.008) (Fig. 3).
CI: confidence interval.
CI: confidence interval.
In this study, we elucidated the proportion of AF treatment transition, which ranged from 10 to 28%, patients who commenced anticoagulant monotherapy or a combination of anticoagulant and rate control medications within a 12-month period. This was determined using two distinct scenarios analyzed in analysis 1 and analysis 2. Furthermore, the risk factors for AF treatment transition in both scenarios included being male, concurrent administration of rate control medication, and initiation with rivaroxaban as anticoagulant therapy.
According to current guidelines, catheter ablation is the primary choice for paroxysmal or symptomatic AF treatment, provided the patient is physically suitable for the procedure. However, some patients with AF especially in asymptomatic AF opt against catheter ablation as their initial treatment due to several reasons: 1) the requirement for hospital admission for several days, 2) discomfort associated with undergoing intensive treatment through catheter ablation, and 3) apprehension regarding the surgical aspect of the procedure.40,41) Although previous studies have reported that early timing of rhythm control is associated with lower mortality, a notable 87% of AF patients exhibited hesitation.42) In certain cases, even when physicians recommend initiating catheter ablation at the time of AF diagnosis, patient-related factors such as hesitation or fear may delay the procedure. Consequently, the percentage of AF treatment transition reported in this study might be overestimated, considering the inclusion of cases with delayed catheter ablation initiation. Conversely, the percentages in analysis 2 might be underestimated, as some AF patients who genuinely required catheter ablation were excluded. Nonetheless, this study successfully highlighted two distinct scenarios regarding the progression of the disease within one year of initiating treatment in low-risk AF patients. This finding is significantly deemed important for future considerations regarding the appropriateness of proactive early rhythm control interventions.
In the context of our study, we noted that risk factors for AF, especially in male patients, are predominantly associated with the rate control approach. Notably, it is well-established that the progression of AF tends to be more rapid in males than in females.43) Furthermore, patients with advanced stages of AF, including those with persistent AF, frequently undergo rate control strategies.44,45) Therefore, these observations reinforce that our results are in alignment with previously identified risk factors for low-risk AF individuals in our study setting.
In this study, rivaroxaban was consistently observed as a risk factor for AF treatment transition in both analyses (Figs. 2, 3). The inhibitory effects of Xa inhibitors on inflammation and fibrillation have been observed in patients with AF in vitro.46) Specifically, protease-activated receptors (PAR) are known to augment inflammatory cytokines, such as interleukin 6 and tumor necrosis factor-α, in response to the activation by coagulation factor Xa.47,48) Notably, one of the mechanisms implicated in AF involves PAR, which modulates collagen in vessels,49,50) and inflammation that induces cardiac remodeling.51–53) Importantly, edoxaban has demonstrated anti-inflammatory effects in myocardial tissues of AF mice,50) exhibiting more pronounced effects compared to rivaroxaban. Furthermore, the administration of Xa inhibitors has been shown to reduce inflammation and fibrillosis.50) We observed significant risks for AF treatment transition in patients who were started with apixaban or dabigatran in analysis 1 solely. This suggests that apixaban and dabigatran have potential risks for treatment transition in AF patients. Notably, apixaban and dabigatran were reported to have higher anti-inflammation activity in the myocardium, which is one of the potential mechanisms relating to anti-arrhythmia, compared to rivaroxaban in an in vitro study.46,52) The consistent trend between the findings of the above-mentioned in vitro study and our in vivo results could be partially explained by differences in the magnitudes of anti-inflammation activity in the myocardium owing to DOACs. Therefore, Xa inhibitors may offer preventative benefits against AF, although rivaroxaban exhibited a lower efficacy, which could explain the higher proportion of AF treatment transition observed during the one-year study period in our research. On the other hand, future research needs to assess reasons underlying the lower risks for treatment transition on using warfarin.
LimitationsThe study’s limitations are partly attributable to the dataset used. We categorized treatment transitions based on a transition from step 3 or 4 to step 5, following the criteria outlined in the “JCS/JHRS 2020 Guideline on Pharmacotherapy of Cardiac Arrhythmias in 2020” (last update on October 13, 2023).22) These treatment transitions were divided into two groups based on specific scenarios. A notable limitation was the unavailability of physicians’ detailed descriptions in the medical records. To mitigate this limitation, we conducted two separate analyses in our study, presenting the percentages along with their respective ranges.
In this research, we identified the proportion of AF treatment transitions over a 12-month period, highlighting the ranges and pinpointing risk factors at the initiation of anticoagulant therapy. Notably, the trend of higher AF treatment transition rates was observed in patients who commenced treatment with rivaroxaban, warranting further investigation in the near future. Importantly, our study underscores the potential for medical staff to make informed decisions about recommending early rhythm control in clinical settings.
This work was supported by a Grant from Showa University Translational Research in 2023.
All authors meet the ICMJE recommendations. HF and KM contributed to the study conception, drafted the manuscript, and collected the raw data; HF created and confirmed the definitions of the analysis; YK and KM conducted data management. KM provided clinical interpretation and provided the final approval. All authors participated in the discussions during the manuscript preparation. All authors have agreed to publish the final version of the manuscript.
JMDC did not intervene in the implementation of the data analyzed in this study; KM received honorarium fees for presentations from JMDC; KM received travel reimbursement from Abbvie to attend their conference. Department of Hospital Pharmaceutics, School of Pharmacy, Showa University received funding from Ono. with a contract research project according to the collaborative research agreement.
As a potential conflict of interest, the Department of Hospital Pharmaceutics, School of Pharmacy, Showa University received Grants from Nippon-Kayaku, Ono, Shionogi, Bayer, Daiichi Sankyo, Eisai, Mochida, and Taiho; KM received honorarium fees for Eisai, Nippon-Kayaku, Sawai, and Abbvie; The other authors declare no conflict of interest.
This article contains supplementary materials.
