Patient characteristics associated with using transcatheter left atrial appendage occlusion vs. oral anticoagulants for atrial fibrillation

Abstract

Background:

Transcatheter left atrial appendage occlusion (LAAO) is an alternative to oral anticoagulants (OACs) for stroke prevention in patients with atrial fibrillation (AF), but the predictors of LAAO use in routine care are unclear. We aimed to assess the utilization trends of LAAO, compare change in characteristics of LAAO users vs. OACs since its marketing.

Methods:

Using the US Medicare claims database (03/15/2015-12/31/2020), we identified patients with AF , >=65 years, CHA₂DS₂-VASc score >=2 (men) or >=3 (women), with either first implantation of an LAAO device or initiation of OACs, including apixaban, dabigatran, rivaroxaban, edoxaban, or warfarin. Patient characteristics, measured during 365 days before the first LAAO or OAC use date, were compared using logistic regression.

Results:

There were 30,058 LAAO recipients (mean age 77.74 years, female 42.1%) and 792,600 OAC initiators (mean age 78.48, female 53.3%). In 2020, patients had higher odds of initiating LAAO use than in 2015 (0.52 vs. 9.32%; adjusted odds ratio [aOR] 13.64, 12.56-14.81). Old age (i.e., >85 vs. 65-75 years old; aOR 0.84, 0.80-0.88), female sex (aOR 0.74, 0.71-0.76), black race (aOR 0.63, 0.58-0.68, vs. white race), and Medicaid eligibility (aOR 0.61, 0.58-0.64) were associated with lower odds of receiving LAAO. Among clinical characteristics, frailty, cancer, fractures, and venous thromboembolism were associated with lowers odds of LAAO use while history of intracranial and extracranial bleeding, coagulopathy, and falls were associated with higher odds of receiving LAAO.

Conclusions:

Among AF patients receiving stroke-preventive therapy, LAAO use increased rapidly from 2015 to 2020 and was positively associated with the risk factors for OAC complications but negatively associated with old age, advanced frailty, and cancer. Black race and female sex were associated with a lower likelihood of receiving LAAO.

Introduction

Atrial fibrillation (AF) is associated with a 5-fold increased risk of ischemic stroke ¹. Oral anticoagulants (OACs) can reduce this risk by ~70%, but their utilization has been suboptimal and nearly 50%^2,3 of patients with guideline-based indications for anticoagulants^4,5 are not anticoagulated^4,5. The commonly cited reasons for withholding OAC include recurrent bleeding events, high risk of falling, advanced dementia, and end-stage liver disease ^2,6,7. Even for those receiving OACs for stroke prevention, adherence to OACs has been poor. A study of 66,090 individuals with AF newly starting a direct oral anticoagulant (DOAC) found that only 31.6% persisted in taking the medication at 12 months ⁸. Poor adherence to OACs has been shown to compromise the intended stroke prevention effectiveness ⁹. Therefore, identifying effective alternative stroke prevention strategies for AF is warranted for those at high risk of developing bleeding complications.

Because ~90% of thrombus formation in AF occurs in the left atrial appendage, clinial trials have shown transcatheter left atrial appendage occlusion (LAAO) could achieve comparable stroke prevention efficacy in AF when compared to oral anticoagulants^10,11. The landmark trial, PROTECT AF, reported a hazard ratio (HR) of 0·62 (95% confidence interval [CI], 0·35–1·25) for the composite efficacy outcome (stroke, systemic embolism, or death)¹⁰. The earlier concern that LAAO was associated with higher rates of procedural complications were mitigated by a subsequent trial, PREVAIL¹¹. As a result, LAAO was approved in the US in 2015 for stroke prevention in AF patients at high risk for complications from OACs ¹². The stroke prevention effect after LAAO is not dependent on medication adherence, which may be appealing in light of poor adherence with OACs ¹³. The uptake of transcatheter LAAO has been growing rapidly ¹⁴. However, there is very little data on how clinicians actually select patients for transcatheter LAAO and whether this has evolved over time. Based on US Medicare claims data from 2015 to 2020, we aimed to describe the temporal trend of transcatheter LAAO vs. OAC utilization and identify the patient features associated with use of LAAO and whether these features have changed over time.

Methods

Data Source

Because of the sensitive nature of the data collected for this study, requests to access the dataset from qualified researchers trained in human subject confidentiality protocols may be sent to CMS/ResDAC at “resdac@umn.edu”. This study was approved by the Institutional Review Board of the Brigham and Women’s Hospital, Boston, Massachusetts. We analyzed claims data from Medicare fee-for-service Parts A (inpatient), B (outpatient), and D (pharmacy claims). Medicare is a federally funded insurance program that covers legal residents in the U.S. aged 65 years and older and individuals with certain disabilities ¹⁵. The Medicare claims data contain information on demographics, enrollment start- and end-dates, dispensed medications and performed procedures, and medical diagnoses.

Study Population

Among adults aged >= 65 years with non-valvular AF and a CHA₂DS₂-VASc stroke risk score >=2 in males and >=3 in females, we established a comparative cohort of new users of a transcatheter LAAO device or an OAC (warfarin, dabigatran, rivaroxaban, apixaban, and edoxaban) from 03/15/2015 to 12/31/2020. The cohort entry (index) date was the LAAO procedure or OAC dispensing date; the 365-day period prior to and including the index date was the baseline assessment period (BAP). The OAC new user group was required not to have used any OACs or transcatheter or surgical LAAO in the BAP. The transcatheter LAAO new user group was required not to have used transcatheter or surgical LAAO but was allowed to have used an OAC in the BAP. This is because transcatheter LAAO is often considered among those who have failed or developed a complication from taking an OAC. OAC use was determined using Medicare dispensing data, and LAAO was assessed using procedure codes in the Medicare claims data. We further applied the following criteria in the BAP: 1) diagnoses of AF, 2) at least 365-day enrollment in medical and pharmacy coverage, 3) not missing age or gender information, 4) no valvular heart diseases (see Appendix in the Supplementary Materials for detailed protocol and Supplementary Table S1 for the definitions of each condition).

Measurement of Baseline Covariates

In the BAP, we assessed patient demographics, comorbidities, prescription drug use, and healthcare utilization (see Supplementary Tables S2, S3, S4 and S5 for details). We assessed proxy socioeconomic status (SES) based on zip code of residence¹⁶ and concomitant Medicaid (covering an economically disadvantaged population) eligibility (i.e., “dual eligibles”), as a proxy for low SES status. We also computed the CHA₂DS₂-VASc stroke risk score^17,18, the HAS-BLED bleeding risk score¹⁹, and a combined comorbidity score ²⁰. Frailty was measured using a claims-based frailty index (CFI) validated against the clinical measurement of frailty parameters ^21-24 Based on literature, we used CFI <0.15 to define older adults as robust, 0.15–0.24 as prefrail, 0.25–0.34 as mildly frail, and >=0.35 as moderate-to-severely frail ^21-24.

Statistical Analysis

We assessed the univariate time trend of patient characteristics by the Cochran-Armitage Trend Test. We used multivariable logistic regression to estimate the odds ratio of receiving LAAO vs. OACs with 95% confidence interval (CI), adjusting for 69 above-listed covariates. Missing indicator method was used to handle missing data. In order to assess multivariable-adjusted time-trend of patient demographic factors and comorbidities, we included the interaction terms of calendar time (before vs. after 2017 2^nd quarter (Q2), the mid-point of the study period) with age groups, gender, race, SES status, and combined comorbidity score in a logistic regression. The statistical analyses were conducted using STATA 17 (StataCorp LLC. 2021., College Station, TX) and SAS 9.4 (SAS Institute Inc., Cary, NC)

Results

Study population:

In this cohort of AF patients, 30,058 underwent LAAO (mean age 77.74 years, female 42.1%) and 792,600 began use of an OAC (mean age 78.48, female 53.3%, see Supplementary Figure S1 for cohort formation process). The mean combined comorbidity score was higher in the LAAO cohort (6.21 [SD=3.41]) than in the OAC cohort (5.40 [3.69]). Indicators of both prior stroke and prior bleeding risk were higher in the LAAO cohort. The mean CHA₂DS₂-VASc score was 5.10 (1.58) in the LAAO cohort vs. 4.84 (1.68) in the OAC cohort. The mean HAS-BLED score was 3.14 (0.88) in the LAAO cohort versus 2.94 (0.92) in the OAC cohort.

Time trend in use of LAAO and demographic factors of LAAO recipients:

Among patients with AF, the number of OAC users decreased over time (147,013 in 2015 vs. 71,460 in 2020), whereas there was a markedly increasing trend of utilization of LAAO (766 in 2015 and 7,346 in 2020, adjusted odds ratio [aOR]: 13.64, 12.56-14.81, comparing 2020 to 2015 in terms of odds of getting LAAO vs. OACs, Table 1 and Figure 1). In 2020, LAAO accounted for 9.32% of all patients receiving either LAAO or OACs. In multivariable-adjusted models, compared to individuals aged 65-75 years old, those aged 76-85 had a higher odds (aOR1.27, 1.24-1.31) but those older than 85 had a lower odds of receiving LAAO than OACs (aOR 0.84, 0.80-0.88). Female sex was associated with lower odds of receiving LAAO than OACs (aOR: 0.74, 0.71-0.76). There was significant US regional variation in the relative use of LAAO vs. OACs, with the highest LAAO utilization in the West, followed by the South, Midwest, and Northeast (Table 1).

Table 1.

Patient demographics and time trend in LAAO^* vs. OAC^† users for AF^‡ stroke prevention

Variables	LAAO, N (%)	OAC, N (%)	Crude OR§ (95% CI)	Adjusted OR∥ (95% CI)
Age
65-75 years	12267 (40.8)	326225 (41.2)	Ref	Ref
76-85 years	14596 (48.6)	322593 (40.7)	1.20 (1.17, 1.23)	1.27 (1.24, 1.31)
>=86 years	3195 (10.6)	143782 (18.1)	0.59 (0.57, 0.61)	0.84 (0.80, 0.88)
Gender
Male	17411 (57.9)	370237 (46.7)	Ref	Ref
Female	12647 (42.1)	422363 (53.3)	0.64 (0.62, 0.65)	0.74 (0.71, 0.76)
Race
White	28130 (93.6)	714799 (90.2)	Ref	Ref
Black	832 (2.8)	39368 (5.0)	0.54 (0.50, 0.58)	0.63 (0.58, 0.68)
Other	1096 (3.6)	38433 (4.8)	0.72 (0.68, 0.77)	0.73 (0.69, 0.79)
Region
Northeast	4519 (15.0)	118134 (14.9)	Ref	Ref
Midwest	5868 (19.5)	145584 (18.4)	1.05 (1.01, 1.10)	1.19 (1.14, 1.25)
South	10970 (36.5)	244095 (30.8)	1.17 (1.13, 1.22)	1.31 (1.25, 1.36)
West	5863 (19.5)	103189 (13.0)	1.49 (1.43, 1.55)	1.78 (1.70, 1.86)
Missing	2838 (9.4)	181598 (22.9)	0.41 (0.39, 0.43)	0.58 (0.52, 0.65)
Year
2015	766 (2.5)	147013 (18.5)	Ref	Ref
2016	2350 (7.8)	142809 (18.0)	3.16 (2.91, 3.43)	2.76 (2.53, 3.01)
2017	4235 (14.1)	138517 (17.5)	5.87 (5.43, 6.34)	4.78 (4.40, 5.20)
2018	6448 (21.5)	149949 (18.9)	8.25 (7.65, 8.90)	6.44 (5.93, 6.99)
2019	8913 (29.7)	142852 (18.0)	11.97 (11.12, 12.90)	8.87 (8.18, 9.62)
2020	7346 (24.4)	71460 (9.0)	19.73 (18.30, 21.27)	13.64 (12.56, 14.81)

^*LAAO= transcatheter left atrial appendage occlusion

^†OAC= oral anticoagulants

^‡AF= atrial fibrillation

^§OR=odds ratio

^∥aOR=adjusted OR, adjusting for patient demographics, comorbidities, prescription drug use, healthcare utilization, socioeconomic status¹⁶, and claims-based frailty index ^21-24(see Supplementary Tables S2-S5 for covariate definition and Supplementary Table S6 for estimates for all the covariates in the model), CI=Confidence interval, Ref=Reference.

Figure 1. Time trend of LAAO users among patients receiving stroke prevention treatment (LAAO or OAC) for atrial fibrillation.

Abbreviations: LAOO = Transcatheter left atrial appendage occlusion; OAC = Oral anticoagulants.

Race and SES status:

In multivariable-adjusted models, black race was associated with lower odds of receiving LAAO compared to OAC (aOR: 0.63, 0.58-0.68, compared to white race; Table 1) and the negative association was persistently observed in both the first half of the study (2015 to 2017, aOR: 0.76, 0.66-0.88), and the second half of the study (2018 to 2020, aOR: 0.58, 0.42-0.79). Medicaid dual eligibility (a proxy for low SES) was negatively associated with use of LAAO utilization (aOR 0.61, 0.58-0.64). Compared to those with low SES score in the first half of the study period, those with high SES score had a modestly higher odds of receiving LAAO (aOR 1.16, 1.06-1.27) but this association did not continue into the second half of the study (aOR 0.94, 0.77-1.13. p for interaction <0.001, Table 2).

Table 2.

Socioeconomic status in LAAO^* vs. OAC^† users for AF^‡ stroke prevention

Variables	LAAO, N (%)	OAC, N (%)	Crude OR§ (95% CI)	Adjusted OR∥ (95% CI)
Medicaid dual eligibility #	2842 (9.5)	144385 (18.2)	0.47 (0.45, 0.49)	0.61 (0.58, 0.64)
In the 1st half of study period **
Low SES score‡‡	613 (8.34)	40566 (9.47)	Ref	Ref
Low-intermediate SES score	531 (7.22)	30732 (7.17)	1.14 (1.02, 1.29)	1.06 (0.94, 1.20)
High-intermediate SES score	861 (11.71)	46666 (10.89)	1.22 (1.10, 1.36)	1.10 (0.99, 1.23)
High SES score	355 (48.36)	161910 (37.80)	1.45 (1.33, 1.58)	1.16 (1.06, 1.27)
Missing SES score	1791 (24.36)	148465 (34.66)	0.80 (0.73, 0.88)	1.50 (1.31, 1.72)
In the 2nd half of study period ††
Low SES score	2817 (12.41)	47572 (13.06)	Ref	Ref
Low-intermediate SES score	2226 (9.80)	35608 (9.78)	1.05 (0.83, 1.35)	0.95 (0.74, 1.24)
High-intermediate SES score	3468 (15.27)	53144 (14.59)	1.10 (0.99, 1.37)	0.98 (0.77, 1.23)
High SES score	12587 (55.43)	182511 (50.10)	1.16 (0.97, 1.39)	0.94 (0.77, 1.13)
Missing SES score	1609 (7.09)	45426 (12.47)	0.60 (0.49, 0.74)	0.93 (0.72, 1.20)

^*LAAO= transcatheter left atrial appendage occlusion

^†OAC= oral anticoagulants

^‡AF= atrial fibrillation

^§OR=odds ratio

^∥aOR=adjusted OR, adjusting for patient demographics, comorbidities, prescription drug use, healthcare utilization, socioeconomic status ¹⁶, and a claims-based frailty index ^21-24 (see Supplementary Tables S2-S5 for covariate definition and Supplementary Table S6 for estimates for all the covariates in the model), CI=Confidence interval, Ref=Reference.

^#Medicaid dual eligibility, which overrepresents the economically disadvantaged population

^**1^st half of the study=2015-2017

^††2^nd half of the study=2018-2020

^‡‡SES= socioeconomic status, SES score was calculated based on zip code of residence, low=0-49, low intermediate=50-52, high-intermediate=53-56, high=57-100 ¹⁶.

Time trend in patient features:

From 2015 to 2020, among LAAO users, there was a significant increasing trend in the mean combined comorbidity score (from 5.04 to 6.26, p<0.001), the proportion of having a low SES score (from 6.01 to 13.15%, p<0.001), and age >85 years (from 7.83 to 10.52%, p<0.001). While females were less likely to receive LAAO in both the first and second halves of the study, the female proportion increased from 2015 to 2020 (from 35.77 to 42.17%, p=0.046; Figure 2). The small percentage of black patients receiving LAAO persisted over the study period, 3.0% in 2015 vs. 2.4% in 2020.

Figure 2. Time trend of key demographic and comorbidity factors among LAAO users.

Abbreviations: LAOO = Transcatheter left atrial appendage occlusion; SES = Socioeconomic status; CCS = Combined comorbidity score.

The association between frailty and comorbidities and LAAO:

In multivariable-adjusted models, strong positive predictors for LAAO use included prior intracranial bleeding (aOR: 2.83, 2.64-3.03), gastrointestinal (GI) bleeding (aOR: 2.19, 2.04-2.36), other extracranial bleeding (aOR: 1.49, 1.42-1.57), coagulopathy (aOR: 2.76, 2.64-2.88), or history of falling (aOR: 1.56, 1.50-1.61). Congestive heart failure, hypertension, chronic kidney disease, peptic ulcer disease, obesity, and anemia were also positively associated with LAAO use. Strong negative predictors of LAAO use included moderate-to-severe frailty (aOR: 0.14, 0.13-0.16), cancer (aOR: 0.76, 0.74- 0.78), deep vein thrombosis (DVT, aOR: 0.47, 0.44-0.49), pulmonary embolism (PE, aOR: 0.30, 0.27-0.33), arterial embolism (aOR 0.65, 0.59-0.72), hip fracture (aOR: 0.30, 0.27-0.33),vertebral fracture (aOR: 0.72, 0.67-0.78), acute kidney injury (aOR: 0.43, 0.41-0.44), liver dysfunction (aOR, 0.80, 0.77,−0.83), history of coronary bypass surgery (aOR: 0.43, 0.38-0.47) and receipt of prescription antiplatelet agents [clopidogrel (aOR: 0.66, 0.63-0.68), prasugrel (aOR 0.71, 0.58-0.86), or ticagrelor (aOR: 0.70, 0.62-0.79)]. Among indices of healthcare utilization, number of hospitalizations [aOR 2.07, 2.04-2.10] and prior cardiologist visit [aOR 2.50, 2.32-2.70] were positively associated with LAAO use, while number of prior emergency room visits were negatively associated with LAAO use (aOR 0.71, 0.71-0.72; Table 3). See Table S6 for all coefficients of multivariable model.

Table 3.

Selected patient comorbidities in LAAO^* vs. OAC^† users at study entry

Variables	LAAO, N (%)	OAC, N (%)	Crude OR‡ (95% CI)	Adjusted OR§ (95% CI)
Common risk factors for OAC complications
Intracranial bleeding	2887 (9.6)	17922 (2.3)	4.59 (4.40, 4.79)	2.83 (2.64, 3.03)
GI Bleeding	2775 (9.2)	10528 (1.3)	7.56 (7.23, 7.89)	2.19 (2.04, 2.36)
Other extracranial bleeding	7710 (25.7)	65866 (8.3)	3.81 (3.70, 3.91)	1.49 (1.42, 1.57)
Coagulopathy	4182 (13.9)	27529 (3.5)	4.49 (4.34, 4.65)	2.76 (2.64, 2.88)
Fall	6792 (22.6)	131149 (16.5)	1.47 (1.43, 1.51)	1.56 (1.50, 1.61)
Cardiovascular comorbidities
Ischemic stroke	10371 (34.5)	201876 (25.5)	1.54 (1.50, 1.57)	1.10 (1.05, 1.16)
Congestive heart failure	16343 (54.4)	359617 (45.4)	1.43 (1.40, 1.47)	1.09 (1.05, 1.12)
Hypertension	29450 (98.0)	754229 (95.2)	2.46 (2.27, 2.67)	2.02 (1.85, 2.20)
Diabetes	13801 (45.9)	323426 (40.8)	1.23 (1.20, 1.26)	1.01 (0.98, 1.05)
CABG	478 (1.6)	20030 (2.5)	0.62 (0.57, 0.68)	0.43 (0.38, 0.47)
Other comorbidities
Cancer	12344 (41.1)	277886 (35.1)	1.29 (1.26, 1.32)	0.76 (0.74, 0.78)
DVT	1670 (5.6)	72984 (9.2)	0.58 (0.55, 0.61)	0.47 (0.44, 0.49)
PE	652 (2.2)	41334 (5.2)	0.40 (0.37, 0.44)	0.30 (0.27, 0.33)
Arterial embolism	564 (1.9)	23480 (3.0)	0.63 (0.58, 0.68)	0.65 (0.59, 0.72)
Hip fracture	529 (1.8)	29242 (3.7)	0.47 (0.43, 0.51)	0.30 (0.27, 0.33)
Pelvis fracture	199 (0.7)	5869 (0.7)	0.89 (0.77, 1.03)	1.01 (0.86, 1.19)
Vertebral fracture	1105 (3.7)	28187 (3.6)	1.04 (0.97, 1.10)	0.72 (0.67, 0.78)
Acute kidney injury	7420 (24.7)	198109 (25.0)	0.98 (0.96, 1.01)	0.43 (0.41, 0.44)
Liver dysfunction	4111 (13.7)	83757 (10.6)	1.34 (1.30, 1.39)	0.80 (0.77, 0.83)
Dementia	2493 (8.3)	95767 (12.1)	0.66 (0.63, 0.69)	0.97 (0.92, 1.02)
Chronic kidney disease	12206 (40.6)	259402 (32.7)	1.41 (1.37, 1.44)	1.17 (1.14, 1.21)
Peptic ulcer disease	3270 (10.9)	98071 (12.4)	0.86 (0.83, 0.90)	1.18 (1.13, 1.24)
Obesity	12215 (40.6)	237249 (29.9)	1.60 (1.57, 1.64)	1.21 (1.18, 1.25)
Anemia	17955 (59.7)	341717 (43.1)	1.96 (1.91, 2.00)	1.28 (1.24, 1.31)
Frailty
Frailty: Robust∥	3329 (11.1)	103618 (13.1)	Ref	Ref
Frailty: Prefrail∥	20622 (68.6)	501627 (63.3)	1.28 (1.23, 1.33)	0.81 (0.78, 0.85)
Frailty: Mildly frail∥	5709 (19.0)	160823 (20.3)	1.10 (1.06, 1.15)	0.41 (0.39, 0.44)
Frailty: Moderate-to-severe frail∥	398 (1.3)	26532 (3.3)	0.47 (0.42, 0.52)	0.14 (0.13, 0.16)
Medication use
Clopidogrel	4733 (15.7)	129365 (16.3)	0.96 (0.93, 0.99)	0.66 (0.63, 0.68)
Prasugrel	128 (0.4)	3498 (0.4)	0.96 (0.81, 1.15)	0.71 (0.58, 0.86)
Ticagrelor	342 (1.1)	7729 (1.0)	1.17 (1.05, 1.30)	0.70 (0.62, 0.79)
Other antiplatelets	244 (0.8)	9611 (1.2)	0.67 (0.59, 0.76)	0.64 (0.55, 0.74)
Beta-blockers	20689 (68.8)	560972 (70.8)	0.91 (0.89, 0.93)	0.79 (0.76, 0.81)
Calcium channel blockers	8059 (26.8)	239731 (30.2)	0.84 (0.82, 0.87)	0.80 (0.77, 0.82)
Healthcare utilization, mean (sd)
# of hospitalization#	2.04 (1.35)	1.17 (1.27)	1.39 (1.38, 1.40)	2.07 (2.04, 2.10)
# of ED visit#	1.56 (1.91)	1.42 (1.94)	1.03 (1.02, 1.03)	0.71 (0.71, 0.72)
# of physician visit#	19.00 (9.84)	12.64 (8.51)	1.06 (1.06, 1.06)	1.06 (1.06, 1.06)
# of cardiologist visit#	0.98 (0.15)	0.92 (0.26)	3.28 (3.05, 3.53)	2.50 (2.32, 2.70)

^*LAAO= transcatheter left atrial appendage occlusion

^†OAC= oral anticoagulants

^‡OR=odds ratio

^§aOR=adjusted OR, adjusting for patient demographics, comorbidities, prescription drug use, healthcare utilization, socioeconomic status ¹⁶, and a claims-based frailty index (CFI) ^21-24 (see Supplementary Tables S2-S5 for covariate definition and Supplementary Table S6 for estimates for all the covariates in the model), CI=Confidence interval, Ref=Reference.

^∥Frailty was assessed based on a CFI validated against clinical measures of frailty, with CFI <0.15 as robust, 0.15–0.24 as prefrail, 0.25–0.34 as mildly frail, and >=0.35 as moderate-to-severely frail state ^21-24.

^#Healthcare utilization was measured in the 365 days before the cohort entry.

Discussion

Among US Medicare beneficiaries, the uptake of LAAO has grown rapidly from 2015 to 2020 with moderate regional variation. From comprising just 0.52% of AF patients receiving stroke preventive therapy in 2015, LAAO use grew to 9.32% in 2020. Age 76-85 was associated with increased use of LAAO vs. those aged 65-75. However, those older than 85 were less likely to receive LAAO. Women and black patients were also less likely to receive LAAO. The risk factors for OAC complications, including a history of intracranial hemorrhage , major bleeding, coagulopathy, and risk of falling were strong predictors of receiving LAAO. In contrast, frailty and cancer were negatively associated with LAAO use. AF patients with a history of venous thromboembolism (VTE), many of whom require OAC for their VTE, were less likely to receive LAAO. It is possible that the association of cancer and fractures with VTE may have favored the use of OAC rather than LAAO as stroke-preventive therapy.

To our knowledge, this is the first population-based study comparing the characteristics of patients taking OACs vs. receiving LAAO reflecting clinical practice in the US. By virtue of our use of Medicare data, the mean age of our study population (78 years) was substantially older than the ages in the randomized trial populations (72-74 ^10,25) and the registry populations (73-76)^26,27. The mean CHA₂DS₂-VASc score of our study population (4.85) was higher than the trial populations (3.4-3.8) ^10,11, and the registry populations (4.5-4.6)^26,27. The outcome data from different studies and settings should always be interpreted in light of these differences in demographics and comorbidity profiles. While patients enrolled in registries received care outside of an experimental setting, they were all LAAO users with no long-term OAC users to serve as a reference group. Our comparison with OAC users provides a head-to-head comparison of patients receiving LAAO vs. OAC in usual clinical care.

We used the US Medicare claims database because it is the largest health insurance payor for older adults, covering 49% of the US population aged 65 years or older. This national study is representative of the US elderly population with high-risk features (e.g., risk of bleeding complications from OACs) for whom LAAO is considered a reasonable alternative. A major advantage of LAAO over OAC is that its stroke prevention effect does not depend on medication adherence. In contrast, OACs must be taken daily and persistence deteriorates over time ²⁸ and poor adherence is a significant threat to its effectiveness ⁹. Based on registry data in the early phase (2015-2017), the growth of transcatheter LAAO uptake has been reported to be at a rate of 2 to 3-fold yearly ¹⁴. Our more contemporary analysis demonstrated a 13-fold increased odds of receiving of LAAO vs. OAC use in 2020 vs. 2015. The observed decrease in the absolute number of LAAO and OAC users as well as the increased proportion of LAAO among all OAC-indicated populations could be potentially explained by the reduction in clinical events during 2020 due to the SARS-2 Coronavirus 2019 (COVID-19) pandemic ²⁹.

Race and socioeconomic status (SES) were both significantly associated with LAAO use. Compared to those with a low SES score, those with a high SES score had an increased odds of receiving LAAO only in the first half of the study period (2015 to 2017) but not in the later period (2018 to 2020). After adjusting for demographics, SES, comorbidities, and healthcare utilization, black race remained negatively associated with LAAO use compared to white race. Unlike SES, racial disparities persisted over the course of our study (aOR was 0.58 to 0.76 throughout the study period comparing black vs. white race). This is consistent with a recent study showing zip codes with a higher proportion of Black or Hispanic patients had lower LAAO rates³⁰. Another study reported that non-White patients had worse outcomes following LAAO implantation³¹. One hypothesis underlying such differential outcomes is that LAAO for non-White patients is more likely to be performed at centers with lower volume and experience^32-36, which warrant future investigation for confirmation. The trends in racial disparaty of LAAO utilization need to be closely monitored and addressed.

Female sex was negatively associated with LAAO use. Prior studies have reported that women are more prone to major complications after cardiac implantable electronic device implantation, which may be explained by some physiologic or anatomic differences (e.g., smaller and thinner vessels, smaller chest cavities, etc. ^37,38). While these added risks or challenges associated with other devices may not be generalizable to LAAO, it is conceivable they could lead to providers’ hesitancy. A recent registry study demonstrated significantly higher rates of in-hospital adverse events associated with LAAO for women compared to men ³⁹. We also noted the female proportion among LAAO users has gradually increased from 2015 to 2020. It will be important to assess if this increasing trend continues as providers of LAAO gain more experience.

According to Medicare reimbursement guidelines, LAAO is indicated when a patient is deemed not suitable for long-term OAC ⁴⁰. The Centers for Medicare & Medicaid Services (CMS) require a shared decision-making discussion with a healthcare provider to confirm that the risk of long-term OAC is substantial, most often due to having high-risk features, such as a history of intracranial hemorrhage, fall, major bleeding, or coagulopathy ^2,6,41,42. These factors were all found to be strong predictors for LAAO use in our analysis. We noted 54.2% among LAAO users did not have any claims code for our pre-specified high-risk factors, which may be partly explained by the fact that the high-risk conditions migh be under-coded in the claims data and the cases captured in the claims data tend to be more severe ones. Our study also found that individuals aged 76-85 years had a higher propensity of receiving LAAO than OACs when compared to those aged 65-75 years old, which can be explained by their higher perceived risk of bleeding risk for OAC use compared to the younger group. In contrast, those older than 85 years were found to be less likely to use LAAO compared to those aged 65-75 years old, most likely reflecting hesitancy in using a new, invasive procedure in patients with an advanced age and complex co-morbidities^43,44. Frailty was identified as a strong negative predictor of LAAO use in our analysis. Patients with high frailty may increase physicians’ hesitancy in performing an invasive procedure because frailty may increase procedural complications^45,46. In contrast, while frailty is positively associated with a higher risk of falling⁴⁷, history of falls represent a high-risk feature for long-term OAC use and was demonstrated to be a positive predictor of LAAO use in our analysis.

Our study has limitations. First, using insurance claims data, we cannot directly measure clinical variables (e.g., laboratory parameters or lifestyle factors) that may affect the analysis. As a result, unmeasured confounding remains possible, and the observed associations may not have a causal interpretation. Second, we applied a claims-based frailty index to account for frailty in our analysis. While this index was previously validated against the clinical measurement of frailty parameters ^21,23, it remains a proxy of frailty. Although it is ideal to conduct a clinical frailty assessment, such evaluation in a large national study population is infeasible. Likewise, SES status was approximated by Medicare-Medicaid dual eligibility and a zip code-based SES score. While the SES score was validated against relevant direct measurements of SES, the possibility of misclassification needs to be acknowledged. Next, our findings may not be generalizable to a population with different backgrounds, such as patients with commercial insurance or those aged younger than 65 years. Lastly, comparing the clincal endpoints and complication rates of LAAO vs. OAC users is beyond the scope of the current study. Future studies are warrented to inform the optimal choice between the two treatment modalities for stroke prevention in patients with AF.

In conclusion, based on a nationally representative US elderly population, we demonstrated that use of LAAO is rapidly growing. Features associated with the risk of OAC complications, including history of major bleeding, fall, and coagulopathy made it more likely to receive LAAO. In contrast, age >85 years, high frailty, and risk factors for VTE were independent negative correlates of LAAO use. Still, the proportion of old patients and those with higher comorbidity scores increased in the second half of our study period, perhaps reflecting more confidence in using LAAO in more medically complex patients. While higher SES correlated with receipt of LAAO early in our study period, SES was no longer associated with receipt of LAAO in the second half of the study. After adjusting for multiple potential selection factors, we found that black race and female sex made it less likely that patients would receive LAAO vs. OACs.

What is Known

• Though use of oral anticoagulants (OAC) reduces the risk of ischemic stroke associated with atrial fibrillation, their use and adherence is not optimal.

• The use of transcatheter left atrial appendage occlusion (LAAO) in patients with atrial fibrillation was approved in 2015 in the US and its use appears to be growing over time.

What the Study Adds

• The uptake of LAAO increased by 13-fold from 2015 to 2020.

• LAAO use was positively associated with the risk factors for OAC complications, but black race, female sex, old age, frailty, and low socio-economic status make it less likely to receive LAAO.

Non-standard Abbreviations and Acronyms

AF

Atrial fibrillation

OAC

Oral anticoagulants

DOAC

Direct oral anticoagulants

LAAO

Left atrial appendage occlusion

CHA₂DS₂-VASc

Congestive heart failure, hypertension, age >= 75 (doubled), diabetes, stroke (doubled), vascular disease, age 65 to 74, and sex (female category)

BAP

Baseline assessment period

SES

Socioeconomic status

HAS-BLED

Hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile INR, elderly, drugs/alcohol concomitantly

CFI

Claims-based frailty index

CI

Confidence interval

VTE

Venous thromboembolism

GI

Gastrointestinal

aOR

Adjusted odds ratio

COVID-19

Coronavirus disease 2019

ICD

Implantable defibrillators

CMS

Centers for Medicare & Medicaid Services

US

United States