Length of stay following percutaneous left atrial appendage occlusion: Data from the prospective, multicenter Amplatzer Amulet Occluder Observational Study

Aims To evaluate factors influencing the length of stay in patients undergoing percutaneous left atrial appendage occlusion (LAAO). Methods and results Patient characteristics, procedural data and the occurrence of serious adverse events were analyzed from the AmplatzerTM AmuletTM Occluder Observational Study. Patients were divided into three groups: same day (S, 0day, n = 60, 5.6%) early (E, 1day, n = 526, 48.9%), regular (R, 2-3days, n = 338, 31.4%) and late (L, ≥4days, n = 152, 14.1%) discharge and followed up for 60 days. Procedure and device related SAE during the in-hospital stay (S: 0.0% vs. E: 1.0% vs. R: 2.1% vs. L: 23%, p<0.0001) were a major trigger for a prolonged in-hospital stay. Of the 37 subjects in the late discharge group with an SAE prior to discharge, cardiac or bleeding complications were the most common underlying conditions, occurring in 26 subjects. Multinomial logistic analysis only identified HAS-BLED score as an independent influencing factor (p = 0.04) for a late discharge. After 60 days, mortality tended to be greatest in the late discharge group (S: 0.0% vs. E: 1.0% vs. R: 1.2% vs. L: 3.3%, p = 0.1066). Conclusion Over half of the subjects receiving an Amplatzer Amulet occluder were discharged within 1 day of the implant procedure. Serious adverse events were a major trigger for a late discharge after LAAO. Increased HAS-BLED score was associated with a prolonged in-hospital stay.


Introduction
Atrial fibrillation (AF) is the most frequently occurring cardiac arrhythmia worldwide and represents a major cause for morbidity and mortality in health-care systems around the world [1,2]. Population-based studies [3,4] show that the incidence and prevalence of AF are steadily increasing with a life time risk for AF of approximately 23%. Since stroke is the most feared complications of AF, potentially leading to death or serious disability, stroke prevention is a substantial part of AF patient management. Symptom relief may be offered through antiarrhythmic drugs or catheter ablation techniques, but current treatments do not reliably prevent the occurrence of thromboembolic events. According to the guidelines a long-term anticoagulative strategy remains the first line therapy for stroke prevention with a Class I Level A recommendation, despite evident side effects and limited drug adherence [5,6]. However, percutaneous left atrial appendage occlusion (LAAO) has emerged as a favorable non-pharmacological stroke prevention. After long term follow-up several randomized trials showed reductions in major bleeding, hemorrhagic stroke, and mortality as compared to warfarin [7]. Despite positive clinical outcomes in trials of subjects tolerant of long-term anticoagulation, LAAO is currently only recommended for patients with AF who are unable to tolerate longterm anticoagulation use due to contraindications, high bleeding risk in general, low drug tolerance or low drug adherence [8] at a Class IIb Level B recommendation [6].
A large proportion of real-world LAAO patients present with significant comorbidities [9][10][11], which pose a special task to the peri-procedural LAAO team. Additionally, despite increasing operator experience [12,13], the procedure itself still poses a complication risk for a number of patients. All these issues may influence the length of stay (LoS), which is a crucial factor for patient comfort and cost in current restricted health care budgets. In this analysis, we aimed to determine the LoS in patients undergoing LAAO in a large world-wide cohort, identify potential risk factors for a late discharge, and investigate additive adverse events during a 60-day follow-up period.

Methods
Patients with non-valvular AF undergoing LAAO implant attempt within the Amplatzer TM Amulet TM Occluder Observational Study from 06/2015 to 09/2016 were included. The full study design can be found in the S1 File. Length of stay was not a pre-specified endpoint in the study, but we are investigating length of stay as part of a post-hoc sub-analysis. This project was presented and accepted by the scientific board. Participants were classified according to the LoS, with same day (S, 0day), early (E, 1day), regular (R, 2-3days) and late discharge (L, �4days after the index procedure) groups defined. Baseline, procedure-related characteristics and serious adverse events (SAE) during the in-hospital stay were analyzed and follow-up through 60 days reviewed. Fig 1 displays  (BARC) type 3 or greater), neurological deficits such as stroke or transient ischemic attack, and respiratory complications. All SAEs were reviewed by an independent clinical events committee which adjudicated relatedness to the implant procedure and device. SAE were grouped into cardiac (pericardial effusion, pericardial tamponade, device embolization, device thrombus, heart failure), bleeding (access-site hematoma, gastrointestinal bleeding, anemia and subdural hematoma), neurologic (ischemic stroke, seizure), respiratory (pneumoniae, exacerbated chronic obstructive lung disease, respiratory failure, pulmonary embolism) and other events (delirium, urinary retention, transesophageal echocardiography related event, air embolism, arteriovenous fistula and pseudoaneurysm).

Statistical analysis
Descriptive statistics summarized baseline and procedural characteristics. The Kruskal-Wallis test was used to identify differences in continuous variables between the same day, early, regular and late discharge groups. Fischer's Exact Test was used to identify differences in categorical variables. Multinomial logistic analyses were used to identify risk factors for a prolonged in-hospital stay. The rates of all-cause mortality, ischemic stroke, and major bleeding were calculated at 60-days post-LAAO using Kaplan-Meier method, with the Log-Rank test identifying differences between groups. A hierarchical model adding center effects was not appropriate due to a large number of centers enrolling patients and the sparseness of the data. A p-value <0.05 was considered statistically significant. SAS 9.4 (SAS Institute, Cary, North Carolina) was used for analysis, STATA/SE 16.0 (StataCorp, College Station, Texas) and Prism 8 (Graphpad Software Inc., San Diego, California) for graphing.

Ethical standards
This study has been approved by the appropriate ethics committees and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All persons gave their written informed consent prior to their inclusion in the study. Details that might disclose the identity of the subjects under study are omitted. A full list of institutional review boards, which approved the study, can be found in the S2 File.

Results
From the initial study population of 1088 patients, 12 were excluded because no discharge visit was performed: 8 with unsuccessful implant attempts (no SAEs), three who died prior to discharge, and one withdrawn after device embolization and surgical LAA ligation. The remaining participants were divided into a same day (0 days, n = 60) an early (1day, n = 526), regular (2-3days, n = 338) and a late discharge (�4 days after the index procedure, n = 152 Fig 2) group and further analyzed. The top three enrolling countries were Germany (n = 378, mean LoS 2.7 days), Italy (n = 178, mean LoS 3.0 days) and Spain (n = 105, mean LoS 3.0 days).
In the late discharge group, procedure-or device-related SAEs occurred in the highest numbers (S: 0.0% vs. E: 1.0% vs. R: 2.1% vs. L: 23.0%, p<0.0001) within seven days. Further procedural and in-hospital outcomes are shown in Table 2.
A sub-analysis of the late discharge group revealed that 37 subjects had a SAE (n = 46 events), of those, seven patients with more than one SAE during the hospital course. Of the 37 subjects with SAE prior to discharge, cardiac or bleeding complications were the most common underlying reason for a prolonged LoS (Fig 3), occurring in 26 subjects. Most SAEs (n = 41/46 events, 89.1%) of the late discharge group were adjudicated as related to the implant procedure and/or device. Fifteen of the 16 cardiac SAEs were procedure-or device-related, including 10 cases of pericardial effusion/tamponade. One cases of each AF, acute pulmonary edema, pleural effusion, device embolization, and thrombus on device were the remaining six procedure-or device-related cardiac SAEs. Univariable multinomial logistic analysis (length of stay categorized into 4 categories: 0 day, 1 day, 2-3 days, and �4 days) only identified HAS-BLED score as an independent influencing factor (p = 0.04). Results for gender (Female vs Male) and CHA 2 DS 2 -VASc score were not statistically significant (p = 0.08 and 0.09 respectively). As shown in Table 3, the odds ratio for late discharge relative to early discharge subjects is 1.271 for every one unit increase in HAS-BLED score, indicating an association between higher HAS-BLED score and longer length of stay following LAAO. As illustrated in Table 1, the HAS-BLED score was similar for patients with same day, early, or regular discharge durations (mean of 3.3), but elevated in the late discharge group (mean of 3.5). During a 60-day follow-up period, the overall mortality (S: 0.0% vs. E: 1.0% vs. R: 1.2% vs. L: 3.3%, p = 0.106, Fig 4A) was comparable. The rate of major bleeding events (S: 0.0% vs. E: 5.2% vs. R: 2.7% vs. L: 16.5%, p<0.0001, Fig 4B) and ischemic stroke (S: 0.0% vs. E: 0.4% vs. R: 0.3% vs. L: 4.0%, p<0.0001, Fig 4C) was the highest in the late discharge group.

Discussion
In the field of structural heart disease, percutaneous LAAO has emerged as a guideline recommended stroke prevention strategy performed with an increasing case load worldwide [14][15][16].
To our knowledge, a detailed LoS analysis, in a large number of patients from multiple geographies, has not yet been performed. Our study shows that 1) SAE are a major trigger for a prolonged LoS 2) the most common underlying reasons are cardiac and bleeding complications and 3) only the HAS-BLED score is independently associated with a longer in-hospital stay.

SAE as a major trigger for a prolonged LoS
LoS analyses of LAAO patients were previously performed in small numbers and mostly concentrate on the safety and feasibility of early or same-day discharge concepts [17][18][19] since almost all health-care systems have to cope with budget-restrictions. These concepts are applicable for low risk patients with a high likelihood for an uneventful procedure. Although infrequent, SAEs associated with LAAO procedures or devices may be associated with increased mortality and morbidity [20]. Patients undergoing LAAO usually present with significant comorbidities, which may contribute to the occurrence of intraprocedural events. The incidence of peri-procedural SAEs in this global registry is in line with results from other recent LAAO registries, which is a positive result for a rather "early" experience with percutaneous LAAO [21,22]. For the first time, we report a strong correlation of SAEs with prolonged LoS. The peri-procedural phase specifically demands attention. Prompt identification and close monitoring of SAE during that period is crucial for patient safety and comfort and may reduce the LoS and associated resources and expenses. While bleeding and cardiac complications are potentially dangerous, this study suggests that proper care can be provided during the in-hospital setting. These complications may be in part due to the high intrinsic bleeding risk of the LAAO population studied and an extended LoS may be required to deal with the complications. However, even after discharge, patients with a prolonged intra-hospital LoS tended to experienced more SAEs, indicating that a late discharge may define an at-risk patient group, requiring closer monitoring after the initial, interventional phase.
Despite SAEs as a major trigger for a prolonged LoS, we see a difference in LoS between the United Kingdom and Nordic states compared to the rest of Europe (i.e., Germany, Italy, France and Spain). This might be attributed to country-based differences in reimbursement, hospital market competition and procedural volume, but cannot be proven with the current data set.

HAS-BLED score as a predictive factor for prolonged LoS after LAAO
The HAS-BLED score was developed as a practical risk score to predict major bleeding events in patients on oral anticoagulation and hence, enables a benefit-risk estimation in patients with atrial fibrillation [23,24]. It has also been used in percutaneous coronary intervention studies to predict major bleeding events [25] and survival in patients without AF [26]. The score combines modifiable and unmodifiable pre-conditions of the patient and clinicians should revisit the HAS-BLED score to re-confirm correctable risk factors to avoid bleeding and, as our study could show, a prolonged length of stay in patients undergoing percutaneous LAAO. For example, if uncontrolled hypertension is present, the blood pressure level should be improved prior to device implantation. In case of warfarin intake and labile INR, patients may be switched to unfractionated heparine prior to the procedure to avoid inadequate bloodthinning. Those lifestyle modifications and/or drug adjustments may delay the procedure but have the potential to significantly improve outcome. In case of non-modifiable risk factors leading to a high HAS-BLES score, the implanting team has to be aware of an increased risk for major complications and subsequently a prolonged LoS after device implantation.

Same day discharge versus early discharge
Looking at the same day discharge and early discharge groups we see no differences. Therefore, it seems reasonable to base the decision upon whether same day or early discharge is performed on patient wish and social environment. A successful procedure without any SAE often occurred for patients discharged within either time frame and may be the primary factor for allowing same day or early discharge.

Limitations
This is a prospective, multicenter-global trial with over 1000 participants. However, it is not randomized and only one specific device for LAAO was utilized. The study was initially not powered to determine risk factors for the LoS after LAAO. Discharge policy may vary by institution and country in aspects that are unknown or that we are unable to adjust for. A detailed cost analysis in the participating countries is not able to be performed with the study dataset. The study refers to an "early" experience with percutaneous LAAO (2015-2016) with a single device and may not be representative for current practice.

Conclusion
Over half of the subjects receiving an Amplatzer Amulet occluder were discharged within 1 day of the implant procedure. HAS-BLED score was the only independent risk factor associated with a prolonged in-hospital stay. SAE are a major trigger for a late discharge after LAAO, with rates of major bleeding and ischemic stroke significantly greater in the late discharge group. The 60-day mortality rate did not differ between patients grouped by length of stay.