COMPARING EFFICACY AND SAFETY OF LEFT ATRIAL APPENDAGE CLOSURE DEVICES: A NETWORK META-ANALYSIS OF RANDOMIZED CONTROLLED TRIALS
===============================================================================================================================

* John W Davis
* Steven L Mai
* Wissam Harmouch
* Jenna Reisler
* Elizabeth Davis

## Abstract

**Introduction** Atrial fibrillation-related stroke is a leading cause of morbidity and mortality. Options for prevention include left atrial appendage closure devices or oral anticoagulation. However, it remains unclear which option may be superior overall.

**Methods** We conducted a systematic review and network meta-analysis of all clinical trials comparing the WATCHMAN, Amplatzer Amulet (Amulet), and/or OACs. The primary outcomes of interest were any stroke and all-cause death. Safety outcomes included any thromboembolism, device embolization, and pericardial effusion. We calculated risk ratios and heterogeneity statistics for eacjowh comparison, and calculated the probability of intervention superiority where at least one comparison was significant.

**Results** There were 441 articles identified from the search, from which 5 eligible RCTs were identified (n=1,811). Compared to OACs (all warfarin), risk of stroke was non-significantly decreased with WATCHMAN (RR=0.90, 95% CI: 0.48, 1.72, I2=13.4%), but risk with Amulet was non-significantly lower than WATCHMAN (RR=0.95, 95% CI: 0.50, 1.81). However, observed risk of all-cause death was significantly lower with Amulet than OAC (RR=0.53, 95% CI: 0.33, 0.85, I2=0%) and trended towards significance versus WATCHMAN (RR=0.74, 95% CI: 0.55, 1.02, p=0.06). The P-score was 0.982, signifying a >98% probability Amulet was superior to all alternatives. Risk of thromboembolism was non-significantly increased with WATCHMAN (RR=2.04, 95% CI: 0.23, 18.4) and Amulet (RR=1.54, 95% CI: 0.11, 22.1), with head-to-head comparison favoring Amulet (RR=0.76, 95% CI: 0.17, 3.38, I2=0%). Versus WATCHMAN, device embolization risk was non-significantly elevated with Amulet (RR= 2.38, 95% CI: 0.67, 8.43, I2=0%). Finally, risk of pericardial effusion was significantly elevated with Amulet versus OACs (RR=27.0, 95% CI: 3.48, 210) and versus WATCHMAN (RR=2.11, 95% CI: 1.41, 3.17, I2=0%). The inverse P-score for Amulet (0.9995) indicated a very high probability Amulet was inferior to alternatives.

**Conclusion** While risk of some adverse events was greater with Amulet, we estimated >98% probability Amulet is superior to alternatives in risk of death. Pooled patient-level analyses are warranted.

## Introduction

Stroke is the second leading cause of death worldwide.1 Despite advancements in management, it remains a devastating disease frequently resulting in life-long disability and large healthcare burden.1 Additionally, the burden is projected to be growing, not only affecting more people, but younger populations as well.1 For these reasons, much focus has been turned to preventing both primary stroke and recurrent stroke. Atrial fibrillation (AF) represents one of the risk factors most commonly associated with ischemic stroke.2 Estimates suggest AF increases risk of stroke by 4- to 5-fold.3 Further, it has been reported that in cardioembolic stroke, the LAA is the thrombus origin for 91% of non-valvular AF patients and 15 to 38% of patients with cardiomyopathy but no AF.4,5 Therefore, the LAA represents a dangerous and common source of thrombi with or without presence of persistent arrhythmia.

Strokes related to LAA thrombi are also associated with worse disability and mortality than those caused by carotid disease, and thus requires aggressive intervention.6 While oral anticoagulation is highly effective in reducing AF-related stroke,7 many patients with AF have significant contraindications to anticoagulation,8 such as history of serious bleeding that outweighs the benefits of stroke prevention. Transcatheter left atrial appendage occlusion (LAAC) devices have hence been developed in an attempt to minimize the risk of cardioembolic stroke while maintaining patient safety. In 2015, the Food and Drug Administration (FDA) approved the WATCHMAN device, a singular disc designed to seal the LAA, after results from the PROTECT-AF,9 PREVAIL,10 and long-term follow up of the PROTECT-AF11 trials showed that the WATCHMAN device was non-inferior to warfarin in reducing risk of stroke. In the PROTECT-AF trial, WATCHMAN was non-inferior to warfarin therapy in reducing stroke, systemic embolism, and cardiovascular death (RR 0.62, 95% CI 0.35-1.25) with non-inferiority probability being greater than 99.9%.10 The PREVAIL trial also showed non-inferiority of WATCHMAN compared to warfarin regarding risk of ischemic stroke and systemic embolism (RR 1.60, 95% CI 0.50-4.20).10 Additionally, safety events were less with WATCHMAN compared to warfarin (4.2% vs 8.7%, P=0.004).10 Pericardial effusions requiring surgical intervention were noted to be less in the PREVAIL trial than in PROTECT-AF (0.4% vs 1.6%, P=0.03).10

In 2021, the Amplatzer Amulet (Amulet) device was approved after results from the Amulet IDE trial showed that the Amulet device was non-inferior to the WACTHMAN device when assessing outcomes of ischemic stroke and systemic embolism (2.8% vs 2.8%, P <0.001).12 Compared to WATCHMAN, Amulet deploys multiple discs to seal the LAA, which allows for periprocedural adjustment and possibly greater deployment success rates.12 Primary safety analysis between the two LAAC devices found that Amulet had a higher rate of procedural-related complications compared to WATCHMAN (4.5% vs 2.5%, HR=1.86, P=0.02).12 This increase in adverse effects stemmed primarily from the increase in pericardial effusions with 53.7% of the adverse events in the Amulet arm and 26.8% in the WATCHMAN arm being pericardial effusions.12 Device embolization was the second most common adverse event with 14.6% of the adverse events in the Amulet arm and 9.1% in the WATCHMAN arm being device embolization.12

However, the differences observed in this flagship trial can be augmented by including evidence from all available studies. Previous direct meta-analysis13 of LAAC devices combined observational and randomized evidence but still appeared underpowered to detect rarer events such as mortality. There have also been subsequent, head-to-head RCTs that may augment prior evidence. A network meta-analysis (NMA) allows for head-to-head comparison of interventions by evaluating indirect evidence across trials. In the present study, we aim to further elucidate the comparative safety and efficacy of LAAC devices, both head-to-head and against oral anti-coagulation (OAC).

## Methods

### The trials

PubMed, Cochrane Database, and Web of Science were searched for systematic reviews, meta-analyses and/or randomized clinical trials published in English on Watchman and/or Amplatzer (Amulet) devices. The search dates were from 1946 to May 2023. Methods were registered prospectively with Prospero (CRD# 42023428226; see supplement for search terms and strategy). We aimed to include all RCTs comparing either device to anticoagulation (either warfarin or direct oral anticoagulants, DOACs), or to the alternative device for subjects with AF.

There were no additional exclusion criteria for trial eligibility. At least 2 authors blinded to codification by other reviewers identified trials for final inclusion, then coded each for quality with a five-point Jadad quality score.14 Data on exposure, outcomes, comorbidities, and trial design were also extracted and collated for analysis. We also searched [clinicaltrials.gov](https://clinicaltrials.gov) for any additional information, and reported whichever source was more comprehensive for each outcome. After independent extraction, any disagreements were arbitrated by the principal investigator (JD). We made attempts to contact trial investigators for discrepancies.

### Exposure variable

Study arms were classified as ‘Watchman-exposed’, ‘Amulet-exposed’, or ‘Oral Anti-Coagulation’ (OAC)-exposed.

### Outcome variables

We first aimed to code for efficacy of the intervention – prevention of stroke (either ischemic, embolic, or hemorrhagic) and/or death. Adjudication as intervention-related or unrelated was ignored for all outcomes to avoid bias and because of variable definitions among trials. Trial definitions for each outcome were used because these data are study-level and cannot be reliably abstracted otherwise.

The first outcome was for any stroke (ischemic or hemorrhagic). The second was death from any cause. Next, we assessed adverse event outcomes. The first adverse event codified was risk of any thromboembolism. Next, we coded for risk of device embolization. Finally, we assessed risk of pericardial effusion, a known complication of the procedure.9,12 All events were counted regardless of whether trialists identified the event as ‘intervention-related’.

### Analysis

We largely followed methods from previous analyses15 in conducting this study. Published aggregate data from each trial were used. Prevalence estimates were calculated for each outcome by intervention arm. Pooled relative risk (RR) was estimated for each outcome, as follow-up time was approximately equal across each intra-study arm. Given that some events, such as device embolization, are obviously impossible with OAC controls, we used a 0.5 continuity correction to generate RRs where 0 cells were present.

When conducting the network meta-analysis (NMA), we utilized the Inverse Variance method with random effects in the ‘netmeta’ package16 to generate risk ratios for each outcome. We assessed the percentage of evidence utilized for each comparison that is ‘indirect’ (comparing interventions across trials) for each outcome. Risk of bias related to heterogeneity was assessed for all trials using standard measures (I2, Cochran’s Q)17,18 in addition to generating funnel plots. High risk of heterogeneity bias was defined as ≥75% I2.17,18 Because there was a small number of eligible trials, meta-regression and formal hypothesis tests for bias were not feasible. We qualitatively assessed the transitivity assumption – that there is no effect modification with baseline characteristics of subjects across trials – by confirming that age, comorbidities, and AF-related stroke risk were approximately equal at baseline. League tables providing RR estimates of all available comparisons were generated to display multiple comparisons.

Because NMA indirect evidence breaks randomization, causal language may not be appropriate.19 We therefore avoid doing so here. However, we used a frequentist tool (‘netrank’ command in the netmeta package)16,20 to quantify the probability that one intervention was superior to alternatives when at least one comparison was significantly different. A ‘P-score’, the inverse of a one-sided hypothesis test of superiority, was estimated for each outcome. A P-score of < 0.05 suggests there is less than a 5% probability that the intervention is superior to alternatives, whereas a P-score > 0.95 indicates a 95% probability the intervention is superior to alternatives.20 For ease of interpretation, we report inverse P-score (the probability of inferiority) where one intervention appears worse than alternatives.

We followed PRISMA guidelines for reporting and analyzing these studies. All analyses were conducted in R, Version 4.1.3.21

## Results

### Search Results

There were 441 articles that were initially returned in our search, of which 33 (7.5%) were either RCTs, systematic reviews, or meta-analysis that contained potentially eligible trials. From these, we identified 5 RCTs9,10,12,22,23 that met criteria and were abstracted for data. Three12,22,23 assessed both LAAC devices head-to-head, whereas the other 2 trials9,10 assessed WATCHMAN to OACs (both warfarin). Because we were interested in comparative efficacy of devices, trials that aggregated reports of WATCHMAN and Amulet (i.e., PRAGUE-17)24 versus OACs could not be included. There were no identified trials comparing Amulet to OACs. The PRISMA diagram (**Figure 1**) displays the search study flow.

### Study Quality and Subjects Enrolled

Summaries of study quality and characteristics are found in **Table 1**. Jadad scores for study quality varied. Because subjects and healthcare providers could not be blinded to WATCHMAN vs. OAC, the 2 earliest trials9,10 scored 3/5 for study quality. Two12,22 additional trials scored 4/5 possible points (Lakkireddy,12 lack of echo tech blinding; Mansour,22 time sequence-based randomization), and 1 trial23 scored 5/5 possible points for quality. Of note, the 2 trials assessing WATCHMAN vs. OAC excluded subjects with contraindications to warfarin, whereas the 3 trials assessing LAAC devices head-to-head did not. There were no other obvious differences in baseline comorbidities across interventions.

View this table:
[Table 1:](http://medrxiv.org/content/early/2024/04/01/2024.03.18.24304489/T1)

Table 1: Descriptive Statistics

Of the subjects included, 1,811 (55.4%) were randomized to WATCHMAN, 1,071 (32.8%) were randomized to Amulet, and 382 (11.7%) were randomized to OACs (N=3,264 subjects overall). Average treatment age for each intervention ranged from 71.7 to 76.5 years, and average CHADSVASC score ranged from 2.0-2.8,25 while the average HAS-BLED26 score (among the 3 head-to-head trials) ranged from 3.1-4.1. Follow-up time ranged from 45 days22 to 18 months.9,10,27 Full descriptive statistics can be found on eTable 2 in the ***Online Supplement***.

### Efficacy Outcomes

All 5 eligible trials reported risk of any stroke and were included in this analysis. Total event counts of all efficacy and safety outcomes can be found in the ***Online Supplement.*** The primary efficacy outcome results are displayed in **Table 2**. The observed risk of any stroke, when compared to OAC, was non-significantly lower in subjects randomized to WATCHMAN (RR=0.90, 95% CI: 0.48, 1.72) and Amulet (RR=0.86, 95% CI: 0.35, 2.14). Head-to-head device comparison suggested non-significantly lower risk of stroke with Amulet (RR=0.95, 95% CI: 0.50, 1.81). The overall I2 was 13.4% (Cochran’s Q p=0.33), indicating low risk of heterogeneity bias. Funnel plots did not show any meaningful deviation across trials (***Supplement***).

View this table:
[Table 2:](http://medrxiv.org/content/early/2024/04/01/2024.03.18.24304489/T2)

Table 2: Efficacy NMA League Table

Likewise, all 5 trials reported all-cause death and were included. The risk of death from any cause, however, significantly favored Amulet (RR=0.53, 95% CI: 0.33, 0.85) and non-significantly favored WATCHMAN (RR=0.70, 95% CI: 0.49, 1.02) over OAC. Head to head, Amulet was associated with borderline-significant lower risk of death than WATCHMAN (RR=0.74, 95% CI: 0.55, 1.02, p=0.06). The P-score for Amulet was 0.982, signifying a 98% probability that it is superior in risk of death versus alternative interventions, whereas the probability that OACs were inferior to both alternatives was approximately 98% (Inverse P-score 0.98). The I2 estimate was 0% (Cochran’s Q p=0.75), and funnel plots were unremarkable (***Supplement***).

### Safety Outcomes

#### Thromboembolism

The safety outcome results are displayed in **Table 3**. Four of the 5 trials reported risk of any thromboembolic event and were included. When compared to OACs, risk of any thromboembolism was non-significantly increased in both WATCHMAN (RR=2.04, 95% CI: 0.23, 18.4) and Amulet (RR=1.54, 95% CI: 0.11, 22.1). Head-to-head comparison suggested non-significantly lower risk of thromboembolism with Amulet (RR=0.76, 95% CI: 0.17, 3.38). Risk of heterogeneity bias was estimated to be low (I2=0%, Cochran’s Q p=0.95). Funnel plots were again unrevealing (***Supplement***).

View this table:
[Table 3:](http://medrxiv.org/content/early/2024/04/01/2024.03.18.24304489/T3)

Table 3: Safety NMA League Table

### Device Embolization

Because risk of device embolization is non-existent with OACs, we report only head-to-head risk ratios for this outcome. All 5/5 trials reported this event and were analyzed. Risk of device embolization was non-significantly higher with Amulet when compared to WATCHMAN (RR=2.38, 95% CI: 0.67, 8.43). Estimated risk of heterogeneity bias was low (I2=0%, Cochran’s Q p=0.91). Funnel plots were again unremarkable (***Supplement***).

### Pericardial Effusion (with or without tamponade)

Pericardial effusion, with or without tamponade, was reported in 4/5 trials. Trials varied in their reporting of these effusions. For example, some subcategorized effusions as ‘serious’ versus ‘non-serious’, ‘with tamponade’ or ‘without tamponade’, and ‘treatment-related’ or ‘treatment-unrelated’. Here, we report all cumulative event counts for each group.

When compared to OAC, risk of any pericardial effusion was, as expected, significantly greater with Amulet (RR=27.0, 95% CI: 3.48, 210) and significantly greater with WATCHMAN (RR=12.8, 95% CI: 1.72, 95.3). Head-to-head, risk of effusion over two-fold greater with Amulet than WATCHMAN (RR=2.11, 95% CI: 1.41, 3.17). The P-score suggested that the probability of increased risk with Amulet compared to alternatives was greater than 99% (Inverse P-score=0.9995), and the probability that OACs carried less risk was likewise greater than 99% (P-score=0.996). Risk of heterogeneity bias was again estimated to be low (I2=6.3%, Cochran’s Q p=0.34), and funnel plots revealed no meaningful deviations across trials (***Supplement***).

## Discussion

To our knowledge, this is the most comprehensive analysis of available RCT evidence on efficacy and safety of LAAC devices. While both WATCHMAN and Amulet serve the purpose of occluding the LAA, differences in design lead to differences in deployment protocol and subsequent efficacy and safety. Previous meta-analyses were either underpowered (lacking utilization of indirect evidence) or more confounded (including cohort designs). Here, we present compelling evidence that the Amulet may have a superior efficacy and safety profile to WATCHMAN.

It has already been established by head-to-head studies that successful deployment – defined as no significant device leak post-procedure – is approximately 2.1% more likely with Amulet versus Watchman (p=0.003 for superiority).12 However, in this study we use NMA to increase the statistical power to detect differences between interventions for underpowered secondary outcomes.20 Given that death was an underpowered ‘safety’ endpoint (though often included in efficacy composite endpoints) in these studies, it is unsurprising that associations were non-significant in individual trials. We estimated that risk of death was meaningfully lower with Amulet versus both WATCHMAN and OACs (RR=0.74, p=0.06 and 0.53, p=0.01) respectively.

The Amulet’s P-score for death, or probability that it is superior to all other interventions, exceeded 99%. In context, this is approximately equal to the level of certainty that pericardial effusion risk is greater with Amulet than with WATCHMAN or OACs. While the mechanism for this difference is not possible to confirm with study-level meta-analysis, it may be attributable to the greater success rate of LAA seal – and discontinuation of anticoagulation – in Amulet recipients.12,22,23 Given that the success rate excess is approximately 2%,12 these patients would be markedly less likely to experience major bleeding and/or stroke than in the case of unsuccessful deployment. Further, estimated risk of systemic embolism was non-significantly lower (RR=0.76) with Amulet versus WATCHMAN. This event reduction may also explain observed death risk reductions. Future works using patient-level data should confirm and expand these findings with multivariable regression to rule out confounding.

Operators may presently prefer WATCHMAN over Amulet due to risk of late pericardial effusion, which can result in life-threatening cardiac tamponade.12 Indeed, risk of any pericardial effusion, with or without tamponade, was more than two-fold greater with Amulet than WATCHMAN in this analysis. The Amulet, once deployed, typically has greater surface contact with the LAA which subsequently results in inflammation and effusion.12 The risk is exacerbated by operator inexperience with deploying Amulet, and notably no effusions occurred in subjects undergoing implant with an interventionalist with > 9 cases of experience.12 As operators become more experienced with Amulet deployment and as newer editions of Amulet emerge, it is expected that this risk would approach 0.

### Limitations

This study is not without limitations. As previously noted, NMAs use all available evidence to generate effect estimates for each intervention. Because these estimates are therefore non-randomized, we refrain from interpreting these results causally. This is especially true for Amulet versus OAC comparisons, which are based exclusively on indirect evidence. Previous direct meta-analysis13 compared only WATCHMAN and Amulet but 16/19 studies included were observational. Thus, it was more subject to bias, yet also did not include indirect RCT evidence and thereby lacked statistical power to properly evaluate outcomes such as death. The only OAC included in this study is warfarin, which is highly effective in preventing AF-related ischemic stroke28 but has an unfavorable safety profile when compared with Factor Xa inhibitors such as apixaban.7 This limits interpretation of comparative efficacy and safety of OACs, but does not impact estimates of device comparison. Further, our findings are consistent with recent real-world data,29 consistently were estimated to be at low risk for heterogeneity bias, and did not appear affected by any outlier results. The findings are substantial enough to warrant individual patient meta-analysis and possibly larger-scale, long-term head-to-head trials between WATCHMAN and novel editions of Amulet.13

Study-level differences in 1) exclusion criteria, 2) operator experience, and 3) event definitions could have meaningfully changed event rates and confounded our findings. For example, the exclusion of subjects with a prior history of major bleeding in the studies where subjects were randomized to OACs or WATCHMAN9,10 almost certainly decreased absolute risk of death and hemorrhagic stroke within the trial, while inclusion of inexperienced operators in the Amulet-IDE trial likely increased risk of pericardial effusion.29 While the first issue might limit power in detecting underlying effects, it would not change relative risk estimates. NMAs overcome this limitation by utilizing all available evidence, whereas traditional meta-analyses could have failed to detect the differences described here. The second issue suggests that adverse event rates attributable to the procedure in these trials are most generalizable in settings where practitioners have little experience with LAA closure. The issue of trial differences in event definitions was mitigated by our decision to not distinguish, for example, between ‘treatment-related’ vs. ‘treatment-unrelated’ events and other event subclassifications. While this approach protects against confounding from lack of blinding and/or definitional differences across studies, it provides less precise information on the seriousness of these events.

An individual patient data meta-analysis is the best method for ameliorating these limitations, and the logical next step in evaluating these devices’ comparative efficacy and safety. Abstracting study-level data can also lead to miscoding or misinterpretation of reported outcomes. We minimized this risk by having multiple study members code each article blindly, and strictly reporting trial-definition/reported outcomes rather than imposing a universal event definition. We minimized risk of confounding by qualitatively assessing funnel plots for skew, and quantitatively assessing risk of heterogeneity bias across trials. While these methods cannot rule out confounding, they strongly discourage the possibility that results would change with multivariable regression. The small number of trials precluded formal hypothesis tests for assessing bias, but given the low estimated risk of heterogeneity bias it is unlikely this limitation could impact confidence or interpretation of these findings.

## Conclusion

This study suggests a meaningful mortality benefit with Amulet compared to both WATCHMAN and OACs, despite a large increased risk of perioperative pericardial effusion in inexperienced operators. Overall, these findings favor use of Amulet in patients with clinical indication for LAA closure. Careful monitoring for perioperative, pericardial effusions is requisite for either device placement.

## Supporting information

Figure 1 [[supplements/304489_file02.docx]](pending:yes)

Online Supplement [[supplements/304489_file03.docx]](pending:yes)

## Data Availability

Data will be shared on reasonable request. Analyses, including code for R, are available at [rpubs.com](https://rpubs.com) for reproducibility. This page is live and may be supplemented after study publication.

## Footnotes

*   * Currently a medical student (4th year). Completed PhD in 2022.

*   slmai{at}utmb.edu, whharmou{at}utmb.edu, Jdreisle{at}utmb.edu, elidavis{at}utmb.edu

*   **Conflicts of Interest:** John Davis has received consulting honoraria from GE Healthcare and Novartis AG for work unrelated to any data or disease processes assessed in this manuscript. No other authors have perceived or real conflicts of interest to disclose.

*   **Data Sharing:** Data will be shared on reasonable request. Analyses, including code for R, are available at [rpubs.com](https://rpubs.com) for reproducibility. This page is ‘live’ and may be supplemented after study publication.

*   Some changes to rationale. Edited references for accuracy.

*   Received March 18, 2024.
*   Revision received April 1, 2024.
*   Accepted April 1, 2024.


*   © 2024, Posted by Cold Spring Harbor Laboratory

The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission.

## References

1.  1.Feigin VL, Owolabi MO, Feigin VL, et al. Pragmatic solutions to reduce the global burden of stroke: a World Stroke Organization–Lancet Neurology Commission. Lancet Neurol. 2023;22(12):1160–1206. doi:10.1016/S1474-4422(23)00277-6
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1474-4422(23)00277-6&link_type=DOI) 

2.  2.Virani SS, Alonso A, Benjamin EJ, et al. Heart Disease and Stroke Statistics-2020 Update: A Report From the American Heart Association. Circulation. 2020;141(9):e139-e596. doi:10.1161/CIR.0000000000000757
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIR.0000000000000757&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

3.  3.Saw J, Holmes DR, Cavalcante JL, et al. SCAI/HRS Expert Consensus Statement on Transcatheter Left Atrial Appendage Closure. JACC Cardiovasc Interv. 2023;16(11):1384–1400. doi:10.1016/j.jcin.2023.01.011
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jcin.2023.01.011&link_type=DOI) 

4.  4.Blackshear JL, Odell JA. Appendage obliteration to reduce stroke in cardiac surgical patients with atrial fibrillation. Ann Thorac Surg. 1996;61(2):755–759. doi:10.1016/0003-4975(95)00887-X
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/0003-4975(95)00887-X&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=8572814&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1996TU36200068&link_type=ISI) 

5.  5.Mahilmaran A, Nayar PG, Sudarsana G, Abraham K. Relationship of left atrial appendage function to left ventricular function. Indian Heart J. 2004;56(4):293–298.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15586736&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

6.  6.Naksuk N, Padmanabhan D, Yogeswaran V, Asirvatham SJ. Left Atrial Appendage: Embryology, Anatomy, Physiology, Arrhythmia and Therapeutic Intervention. JACC Clin Electrophysiol. 2016;2(4):403–412. doi:10.1016/j.jacep.2016.06.006
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiamNlcCI7czo1OiJyZXNpZCI7czo3OiIyLzQvNDAzIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDQvMDEvMjAyNC4wMy4xOC4yNDMwNDQ4OS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

7.  7.Avezum A, Lopes RD, Schulte PJ, et al. Apixaban in Comparison With Warfarin in Patients With Atrial Fibrillation and Valvular Heart Disease. Circulation. 2015;132(8):624–632. doi:10.1161/CIRCULATIONAHA.114.014807
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MTQ6ImNpcmN1bGF0aW9uYWhhIjtzOjU6InJlc2lkIjtzOjk6IjEzMi84LzYyNCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA0LzAxLzIwMjQuMDMuMTguMjQzMDQ0ODkuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

8.  8.Daimee UA, Wang Y, Masoudi FA, et al. Indications for Left Atrial Appendage Occlusion in the United States and Associated In-Hospital Outcomes: Results From the NCDR LAAO Registry. Circ Cardiovasc Qual Outcomes. 2022;15(8):e008418. doi:10.1161/CIRCOUTCOMES.121.008418
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCOUTCOMES.121.008418&link_type=DOI) 

9.  9.Holmes DR, Reddy VY, Turi ZG, et al. Percutaneous closure of the left atrial appendage versus warfarin therapy for prevention of stroke in patients with atrial fibrillation: a randomised non-inferiority trial. Lancet Lond Engl. 2009;374(9689):534–542. doi:10.1016/S0140-6736(09)61343-X
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(09)61343-X&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19683639&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000269019400021&link_type=ISI) 

10. 10.Holmes DR, Kar S, Price MJ, et al. Prospective randomized evaluation of the Watchman Left Atrial Appendage Closure device in patients with atrial fibrillation versus long-term warfarin therapy: the PREVAIL trial. J Am Coll Cardiol. 2014;64(1):1–12. doi:10.1016/j.jacc.2014.04.029
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6MzoiUERGIjtzOjExOiJqb3VybmFsQ29kZSI7czo0OiJhY2NqIjtzOjU6InJlc2lkIjtzOjY6IjY0LzEvMSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA0LzAxLzIwMjQuMDMuMTguMjQzMDQ0ODkuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

11. 11.Reddy VY, Sievert H, Halperin J, et al. Percutaneous left atrial appendage closure vs warfarin for atrial fibrillation: a randomized clinical trial. JAMA. 2014;312(19):1988–1998. doi:10.1001/jama.2014.15192
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2014.15192&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25399274&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000345450500013&link_type=ISI) 

12. 12.Lakkireddy D, Thaler D, Ellis CR, et al. Amplatzer Amulet Left Atrial Appendage Occluder Versus Watchman Device for Stroke Prophylaxis (Amulet IDE): A Randomized, Controlled Trial. Circulation. 2021;144(19):1543–1552. doi:10.1161/CIRCULATIONAHA.121.057063
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCULATIONAHA.121.057063&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

13. 13.Bing S, Chen RR. Clinical efficacy and safety comparison of Watchman device versus ACP/Amulet device for percutaneous left atrial appendage closure in patients with nonvalvular atrial fibrillation: A study-level meta-analysis of clinical trials. Clin Cardiol. 2023;46(2):117–125. doi:10.1002/clc.23956
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/clc.23956&link_type=DOI) 

14. 14.Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: Is blinding necessary? Control Clin Trials. 1996;17(1):1–12. doi:10.1016/0197-2456(95)00134-4
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/0197-2456(95)00134-4&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=8721797&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1996UA39800001&link_type=ISI) 

15. 15.Davis JW, Weller SC. Intensity of statin therapy and muscle symptoms: a network meta-analysis of 153 000 patients. BMJ Open. 2020:11(6):e043714. doi:10.1136/bmjopen-2020-043714
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1136/bmjopen-2020-043714&link_type=DOI) 

16. 16.Rücker G, Krahn U, König J, Efthimiou O, Schwarzer G. netmeta: Network Meta-Analysis using Frequentist Methods. Published online January 18, 2021. Accessed March 15, 2021. [https://CRAN.R-project.org/package=netmeta](https://CRAN.R-project.org/package=netmeta)
    
    

17. 17.Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539–1558. doi:10.1002/sim.1186
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/sim.1186&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12111919&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000176016900005&link_type=ISI) 

18. 18.Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557-560.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjEyOiIzMjcvNzQxNC81NTciO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wNC8wMS8yMDI0LjAzLjE4LjI0MzA0NDg5LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

19. 19.Davis JW, Weller SC. Intensity of statin therapy and muscle symptoms: a network meta-analysis of 153,000 patients. Published online 2020:413912 bytes. doi:10.5061/DRYAD.KPRR4XH2Q
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.5061/DRYAD.KPRR4XH2Q&link_type=DOI) 

20. 20.Rücker G, Schwarzer G. Ranking treatments in frequentist network meta-analysis works without resampling methods. BMC Med Res Methodol. 2015;15. doi:10.1186/s12874-015-0060-8
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s12874-015-0060-8&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26227148&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

21. 21.R Core Team. R: A language and environment for statistical computing. Published online 2022. [https://www.R-project.org/](https://www.R-project.org/).
    
    

22. 22.Mansour MJ, Harnay E, Al Ayouby A, et al. One year outcome and analysis of peri-device leak of left atrial appendage occlusion devices. J Interv Card Electrophysiol. 2022;64(1):27–34. doi:10.1007/s10840-021-01002-1
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10840-021-01002-1&link_type=DOI) 

23. 23.Galea R, De Marco F, Meneveau N, et al. Amulet or Watchman Device for Percutaneous Left Atrial Appendage Closure: Primary Results of the SWISS-APERO Randomized Clinical Trial. Circulation. 2022;145(10):724–738. doi:10.1161/CIRCULATIONAHA.121.057859
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCULATIONAHA.121.057859&link_type=DOI) 

24. 24.Osmancik P, Herman D, Neuzil P, et al. 4-Year Outcomes After Left Atrial Appendage Closure Versus Nonwarfarin Oral Anticoagulation for Atrial Fibrillation. J Am Coll Cardiol. 2022;79(1):1–14. doi:10.1016/j.jacc.2021.10.023
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jacc.2021.10.023&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

25. 25.Lip GYH, Nieuwlaat R, Pisters R, Lane DA, Crijns HJGM. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–272. doi:10.1378/chest.09-1584
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1378/chest.09-1584&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19762550&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000274612600006&link_type=ISI) 

26. 26.Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJGM, Lip GYH. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138(5):1093–1100. doi:10.1378/chest.10-0134
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1378/chest.10-0134&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20299623&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000284341700015&link_type=ISI) 

27. 27.Lakkireddy D, Thaler D, Ellis CR, et al. Amplatzer Amulet Left Atrial Appendage Occluder Versus Watchman Device for Stroke Prophylaxis (Amulet IDE): A Randomized, Controlled Trial. Circulation. 2021;144(19):1543–1552. doi:10.1161/CIRCULATIONAHA.121.057063
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCULATIONAHA.121.057063&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 

28. 28.Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators, Singer DE, Hughes RA, et al. The effect of low-dose warfarin on the risk of stroke in patients with nonrheumatic atrial fibrillation. N Engl J Med. 1990;323(22):1505–1511. doi:10.1056/NEJM199011293232201
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJM199011293232201&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=2233931&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F01%2F2024.03.18.24304489.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1990EK20900001&link_type=ISI) 

29. 29.Alkhouli M, Freeman JV, Ellis CR, et al. First Experience With Amulet in the United States. JACC Cardiovasc Interv. 2024;17(3):422–434. doi:10.1016/j.jcin.2023.11.027
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jcin.2023.11.027&link_type=DOI)