Background
Atrial fibrillation (AF) can present with severe symptoms, thromboembolic events, and hemodynamic instability that leads to morbidity and mortality and frequent hospitalizations [
1‐
4]. Patients who are highly symptomatic from AF are candidates for a rhythm control strategy. Rhythm control strategies have traditionally focused on anti-arrhythmic drugs (AADs), although studies have shown that these drugs have numerous side effects and many patients do not durably maintain sinus rhythm [
1,
5,
6]. Catheter ablation (CA) has emerged as a viable alternative to AADs that may better maintain sinus rhythm or reduce AF burden.
Emerging evidence has suggested that CA may be more effective than AADs for improving symptoms of AF, although the effect of CA on hard patient outcomes including hospitalization, adverse events, and mortality is unclear. Several randomized controlled trials have sought to assess the effectiveness of CA as a first-line therapy for AF and found CA resulted in decreased AF burden and improved subjective quality of life compared to AADs [
7‐
10]. A recent retrospective study of younger, commercially-insured patients found that CA resulted in fewer hospitalizations for AF and heart failure [
11]. Additionally, there are several ongoing clinical trials to assess the safety and efficacy of early ablation for AF in patients who have not been treated with AADs (NCT03118518, NCT02686749).
These studies have not addressed the optimal timing of CA, although a prospective study showed that shorter duration between AF diagnosis and CA reduced rate of AF recurrence and adverse cardiac remodeling, using NT-proBNP and left atrial size as surrogate measures [
12]. In particular, patients who underwent CA within 1 year of AF diagnosis (referred to as “early CA” for study purposes), had the greatest chance of long-term maintenance of sinus rhythm. Yet, recent consensus guidelines continue to recommend trial of Class I or Class III anti-arrhythmic prior to CA for AF [
13]. Additionally, many US payer-based guidelines mandate treatment with an antiarrhythmic drug prior to referral for catheter ablation, which may delay time to CA and reduce its long-term effectiveness.
It is not well known how this information has translated to real-world practice, particularly with regard to how the frequency of early CA is changing over time, regional differences in practice, and patient specific factors that lead to referral for early CA. This study assesses practice patterns for treatment of atrial fibrillation within a young nationally-representative and commercially-insured population.
Methods
Study population
A retrospective observational study was conducted using medical and prescription claims data from the IBM MarketScan® Commercial Database. The Commercial database includes a nationally-representative, Health Information Portability and Accountability Act of 1996 (HIPAA) compliant sample of patients with employer-sponsored private health insurance [
14]. The study was exempt from Institutional Review Board approval at Beth Israel Deaconess Medical Center.
Patients were identified for inclusion in the study using International Classification of Diseases, 9th revision and 10th revision, Clinical Modification (ICD-9/ICD-10) diagnostic codes for AF (427.21, I48.X). Patients age 20 to 64 with at least two different visits either in the inpatient or outpatient setting with a primary diagnosis of AF within 3 months from January 1, 2010 to September 30, 2016 were included. The use of 2 different visits was used in order to increase the specificity of the AF diagnosis. The date of first AF diagnosis was considered as ‘index AF diagnosis’. Patients needed to be continuously enrolled for at least 12 months pre-index and 12 months post-index period to be included. Because the goal was to diagnose only those patients with new onset atrial fibrillation, patients were excluded if they had any diagnosis of AF in the pre-index period or if they had filled an AAD in the pre-index period. The following AADs were identified: amiodarone, disopyramide, dofetilide, dronedarone, flecainide, quinidine, propafenone, and sotalol.
Covariates and outcomes
Patient demographics included age, sex, region (Northeast, North Central, South, and West), insurance type (comprehensive, Exclusive Provider Organization (EPO) or Health Maintenance Organization (HMO), Point of Service (POS) with capitation, Preferred Provider Organization (PPO), Consumer-Drive Health Plan (CDHP) or High-Deductible Health Plan (HDHP)). Patient clinical characteristics included Charlson Comorbidity Index (CCI), CHA
2DS
2-VASc Score, and previously defined Elixhauser comorbidities [
15‐
17]. The primary outcome of interest was CA within the first year after AF diagnosis, identified using the following ICD-9/ICD-10 codes (ICD-93734; ICD-1002553ZZ, 02563ZZ, 02573ZZ, 02583ZZ, 025K3ZZ, 025L3ZZ, 025M3ZZ, 025S3ZZ, 025T3ZZ) and CPT codes (93,651, 93,656). Additional outcome variables included number of AADs trialed, anticoagulants used, and Direct Current Cardioversions (DCCV). The following oral anticoagulants were included: warfarin, apixaban, dabigatran, rivaroxaban, and edoxaban.
Statistical analysis
We first examined rates of CA within the first year of index for each year of the study, as well as preceding AAD use, and examined temporal trends using an XYZ test. We then developed a multivariable logistic regression model to determine factors associated with CA within the first year. Covariates included patient demographics, CCI Score, CHA2DS2-VASc score, and comorbidities. Results are presented as odds ratios with 95% confidence intervals. All analyses were performed with SAS for Windows, version 9.4 at a 2-tailed significance of P < 0.05.
Discussion
A number of studies have demonstrated the benefits of early referral for CA in selected patient populations. Single-center studies and randomized-controlled trials have shown that early ablation can lead to improvements in adverse cardiac remodeling, greater success in maintaining sinus rhythm, and decreased need for repeat CA [
12,
18]. Studies have shown that young patients may benefit the most from early ablation, however most observational and retrospective studies have focused on patients older than 65 [
19‐
21]. Our study addresses trends in catheter ablation within a young population and specifically identifies predictors of referral for early ablation.
Despite recognition of the benefits of an early rhythm control strategy in some patients, only 30% of patients were trialed on an anti-arrhythmic, and an even smaller portion, 7%, were referred for early CA. This may reflect low arrhythmia burden of newly-diagnosed AF within a young population, or a missed opportunity to avoid long-term sequelae of undertreated AF. Nonetheless, the odds of being referred for early ablation increased by 2.2 times over the study period. This suggests the increases in CA shown in other studies are not simply being driven by increased referral of older patients or those who have already been trialed on AADs. Although we were unable to directly measure AF burden, patients diagnosed in the inpatient setting were 12% more likely to be referred for CA, suggesting severity of AF symptoms were a predictor of early CA.
As AF subtype was recorded after implementation of ICD-10, we investigated whether this affected ablation strategy. We would expect that patients with paroxysmal AF or persistent AF of short duration would be most likely to benefit from ablation and therefore would be referred at higher rates [
12,
22‐
24]. Yet, the rates of ablation were similar, likely due to inclusion and exclusion criteria that minimized the number of patients with long-standing persistent AF or chronic AF and coding integrity. Overall, the ability to generalize outcomes by AF subtype is limited given a large proportion of patients in the post ICD-10 era are classified as having unspecified AF.
Forty-two percent of patients proceeded to ablation without trialing an AAD in 2010, which decreased to 36% in 2016. The increase in AAD use prior to ablation likely reflects increased recognition of the benefits of rhythm control in a young population along with insurance mandates to trial an AAD prior to ablation. As there is increasing recognition that many patients do not have durable responses to treatment with AADs, proceeding with early ablation can lead to improved outcomes and avoid likelihood of repeat ablation [
25]. Overall, these findings are compatible given that many younger patients have a short duration of treatment with AADs prior to ablation [
20].
Overall, healthier patients were most likely to be referred for ablation, both reflected in CHA
2DS
2-VASc score and CCI. Patients with lower CHA
2DS
2-VASc scores were more likely to undergo ablation, which could reflect decreased perceived procedural risk or the desire to discontinue anticoagulation after ablation. Further work is required to understand whether patients are discontinuing anticoagulation after ablation. Similar to results found in other studies, fewer patients are treated with anticoagulants than is recommended by guidelines. Only 46% of patients with a CHA
2DS
2-VASc score of 1–2 and 60% of patients with a CHA
2DS
2-VASc score of 3 or higher filled a prescription for an anticoagulant in the year after diagnosis. While most comorbidities were negatively associated with early CA, patients with cardiomyopathy were more likely to be referred for CA. Presumably some of these patients had tachycardia-induced cardiomyopathy and benefited from maintenance of normal sinus rhythm or decreased burden of AF, as was shown in the CASTLE-AF Trial [
24].
We found significant differences in practice patterns unrelated to patient demographics. Patients located within the Northeast and North Central regions were significantly less likely to be referred for early CA. This mirrors geographic variation seen in other studies of CA as well as cardiac devices more generally [
20,
26,
27]. A study of Medicare patients found that ablation was more likely in the South and West, although there were more specific differences within referral regions [
20]. Even within Europe, studies have shown geographic variation in CA utilization [
28]. Further investigation is required to understand the drivers of these differences, which may include physician training networks affecting referral threshold, hospital incentives and payment methods, or patient preferences. Insurance type is also a significant predictor of early ablation. Patients with PPO or high-deductible health plans are more likely to undergo early ablation. Potential explanations include decreased barrier to early electrophysiology referral, different out-of-pocket expense, or fewer barriers for trialing alternative treatments prior to proceeding to ablation.
Our study includes a large, nationally-representative sample of commercially insured patients. Our study focuses on a non-elderly adult cohort, which coincides with our understanding of who is likely to benefit from ablation. Given non-uniform adoption of CA, there are opportunities to explore the causes of these differences to ensure more uniform adoption of early ablation.
Study limitations
Our study has several limitations. Despite its large size, there is year-to-year variation in total patient encounters due to changes in agreements with the commercial vendor. This limits the ability to characterize absolute procedural volume over time, but still permits understanding of likelihood of undergoing CA. Furthermore, there were differences in geographic representation within the dataset due to vendor agreements, with patients in the South and North Central regions more frequently represented. We cannot draw conclusions about the overall prevalence of ablation within these regions, but the factors associated with early ablation should not be affected. Additionally, we were unable to determine whether ablation was successful in maintaining normal sinus rhythm given we used a claims-based dataset, although future work will explore surrogates of successful ablation including repeat ablation and hospitalizations.
Our reliance on diagnosis codes for AF phenotyping and the underlying criteria for incident AF identification could have influenced the study results. While the type of AF (paroxysmal, persistent, permanent) may relate to the benefit of early ablation, this distinction was only implemented in ICD-10 and does not appear reliable or generalizable given the majority of patients were coded as having unspecified AF. Nonetheless, we would expect this distinction to be less relevant within the first year of diagnosis. As our study includes a younger population, the findings cannot be extended to an older population.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.