Impact of chronic kidney disease on clinical outcomes among patients with atrial fibrillation following percutaneous left atrial appendage occlusion: a multicentre cohort study Authors: Ahmed Mazen Amin, Ameer Awashra, Ubaid Khan, Christopher V DeSimone Published: 2026 ABSTRACT AbstractAimsAtrial fibrillation (AF) and chronic kidney disease (CKD) often coexist, and both increase the risk of thromboembolism and bleeding. We aim to investigate the impact of CKD in patients with AF undergoing left atrial appendage occlusion (LAAO).Methods and resultsWe conducted a retrospective analysis using the TriNetX database, from April 2015 to February 2023. Patients aged ≥ 18 years with AF undergoing their first LAAO were included. We performed a comparative analysis of patients with CKD/end-stage renal disease (ESRD) vs. those without. Propensity Score Matching (PSM) (1:1) was employed based on baseline characteristics. The primary outcomes were ischaemic stroke/transient ischaemic attack (TIA)/systemic embolism (SE) and major bleeding over a three-year follow-up. We included 10 689 patients, of whom 4140 had CKD/ESRD, and 6549 had no CKD. PSM resulted in 3085 patients in each cohort. CKD/ESRD was significantly associated with a higher risk of major bleeding [HR: 1.179, 95% CI (1.080, 1.288), P < 0.001], acute myocardial infarction, acute heart failure, acute kidney injury, cardiac-related haemodynamic instability (CRHI), hospitalizations, and all-cause mortality (P < 0.001), with no significant difference in ischaemic stroke/TIA/SE [HR: 1.019, 95% CI (0.909, 1.141), P = 0.750].ConclusionOur study suggests that CKD/ESRD was significantly associated with a higher risk of major bleeding, acute myocardial infarction, acute heart failure, acute kidney injury, CRHI, hospitalizations, and all-cause mortality, with no significant difference in ischaemic stroke/TIA/SE among patients with AF undergoing LAAO.Keywords: Atrial fibrillation, Left atrial appendage occlusion, TriNetX Introduction Atrial fibrillation (AF) is recognized as the most prevalent sustained cardiac arrhythmia in adults, with an estimated prevalence ranging from 2% to 4%.1 Stroke emerges as the most debilitating complication associated with AF, making its prevention a primary objective in the management of AF.2 The global prevalence of chronic kidney disease (CKD) is estimated at 13.4%.3 It has been shown that there is a bidirectional relationship between CKD and AF.4 Notably, the prevalence of AF escalates in accordance with the degree of renal dysfunction.5 CKD is characterized by a prothrombotic state, which is an independent risk factor for stroke.6 Patients with CKD also experience an increased bleeding risk, with a paradoxical rise in those with CKD who have a high thromboembolic risk.4 Specifically, for each 10 mL/min/1.73 m2 increase in estimated Glomerular Filtration Rate (eGFR), there is a corresponding 9% decrease in haemorrhagic risk.7 Left atrial appendage occlusion (LAAO) has emerged as a nonpharmacological strategy to mitigate thromboembolic events in patients with AF.8,9 Approximately 90% of left atrial thrombi are found in the left atrial appendage.10,11 In patients with CKD, the incidence of left atrial appendage thrombus is higher and increases with worsening renal dysfunction, likely due to increased prothrombotic tendencies.12 Some studies previously assessed the impact of CKD on patients undergoing LAAO.13,14 However, this study aims to conduct a multi-centre U.S. retrospective cohort analysis using a large database to thoroughly investigate the impact of CKD in patients with AF undergoing LAAO. Methods Data source and study design We conducted a retrospective cohort analysis using the TriNetX, LLC platform. TriNetX is a federated network-based platform that provides aggregated, de-identified data from electronic medical records (EMRs) of large healthcare organizations (HCOs) predominantly based in the US, encompassing inpatient, outpatient, and ambulatory care for insured and uninsured patients. Our analysis utilized data from the HCOs within the TriNetX US Collaborative Network, which included 67 HCOs at the time of data extraction. Ethical approval was not required because the platform's patient and HCO data are de-identified per Section §164.514 of the HIPAA Privacy Rule. Data extraction occurred in March 2026 and was based on the following coding systems: CPT for procedures, ICD-CM/PCS for diagnoses and procedures, RxNorm VA/ATC for medications, TNX/LOINC for laboratory data, and HL7V3.0 for visit information. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and adhered to the ethical and reporting standards established by TriNetX under their ‘Publication Guidelines’.15,16 More information about the TriNetX design and analysis is reported in Supplementary material online, Methods. Participants eligibility Patients were included if they met the following criteria: 1) aged ≥ 18 years, 2) had a documented record of AF, and 3) underwent their first LAAO between 1 April 2015 (date of Food and Drug Administration approval for use of the LAAO device in clinical practice), and 28 February 2023. Our exclusion criteria are 1) overlap in eGFR values between the CKD/end-stage renal disease (ESRD) group and the no CKD group, 2) surgical LAAO, 3) surgical left atrial appendage exclusion, 4) presence of prosthetic heart valve, 5) any heart valve replacement, 6) rheumatic mitral stenosis, 7) heart assist device, 8) intracardiac thrombosis, 9) total knee/hip arthroplasty, and 10) kidney transplant. Patients were divided in the primary analysis into two cohorts: 1) patients with CKD/ESRD and 2) patients with no CKD. eGFRs were extracted from TriNetX using the MDRD equation. We defined ‘no CKD’ as patients who met at least one of the following: (1) any instance of eGFR ≥ 60 mL/min/1.73m2 within one day to 30 days before the LAAO procedure, or (2) no documented ICD-10 CKD diagnosis code; and who had no instance of eGFR < 60 mL/min/1.73m2 from one day up to 30 days before LAAO procedure and no record of a dialysis code; CKD stage I: eGFR ≥ 90 and CKD stage II: eGFR 60–89 were treated as no CKD because these early CKD stages have previously been shown to have similar prognoses for cardiovascular outcomes.17 We defined ‘CKD’ as patients who met at least one of the following: (1) eGFR between 15 and 59 mL/min/1.73m2 or < 15 mL/min/1.73m2 but not on dialysis (CKD stage III: eGFR 30–59; CKD stage IV: eGFR 15–29; CKD stage V: eGFR <15 not on dialysis), or (2) any previous documented ICD-10 CKD diagnosis code; and who had no ICD-10 N18.6 code, no dialysis record, and no instance of eGFR ≥ 60 mL/min/1.73m2 within one day to 30 days before the LAAO procedure. We defined ‘ESRD’ by the presence of any record of ICD-10 N18.6 code or a dialysis code. Detailed criteria with codes for selecting the participants are provided in Supplementary material online, Table S1. Baseline characteristics and propensity score matching Baseline characteristics were recorded from the year preceding the first LAAO date to one day before the first LAAO, encompassing demographic, clinical, and procedural variables, and were extracted from the TriNetX database (see Supplementary material online, Table S2). Propensity score matching (PSM) was performed at a 1:1 ratio between the CKD/ESRD and the no CKD cohorts. The nearest-neighbour matching algorithm was used with a calliper of 0.1 pooled standard deviations. The matching quality was evaluated using standardized mean differences (SMDs), with an SMD < 0.1 indicating successful covariate balance. Additionally, density curves were used to illustrate the balance between the cohorts before and after matching (see Supplementary material online, Figure S1). Study endpoints The primary outcomes were (1) ischaemic stroke/transient ischaemic attack (TIA)/systemic embolism (SE) and (2) major bleeding (a composite endpoint of gastrointestinal bleeding (GIB), intracranial haemorrhage, critical site bleeding (including ocular, retroperitoneal, pericardial, pleural, peritoneal, and articular/joint spaces), procedural bleeding, any event requiring blood transfusion, or resulting in precipitous drop in haematocrit or acute posthaemorrhagic anaemia). Secondary outcomes included acute myocardial infarction (AMI), acute heart failure (AHF), acute kidney injury (AKI), venous thromboembolism (VTE), GIB, pericardial effusion/tamponade, cardiac-related haemodynamic instability (CRHI), all-cause hospitalizations, and all-cause mortality. To prevent reverse causality, ischaemic stroke/TIA/SE, major bleeding, AMI, VTE, GIB, and all-cause hospitalizations were assessed beginning from the seventh day after the LAAO. However, AKI, AHF, pericardial effusion/tamponade, CRHI, and all-cause mortality were evaluated from the same day of the LAAO procedure. Each outcome was assessed at a three-year follow-up, both before and after PSM. Moreover, peri-procedural analysis was conducted to assess all outcomes from the same day of the LAAO through 30 days after the LAAO, before and after PSM. Additionally, we have assessed the prescription of oral anticoagulants from the same day of the LAAO up to three-year follow-up, and conducted another analysis up to 30 days after the LAAO, both before and after PSM. It included at least one prescription of any direct oral anticoagulant (DOAC), at least one prescription of apixaban, at least one prescription of rivaroxaban, at least one prescription of dabigatran, and at least one prescription of warfarin. Detailed outcome definitions and corresponding codes are provided in Supplementary material online, Table S3. Subgroup analysis We conducted a subgroup analysis based on CKD stages: 1) CKD, 2) ESRD, and 3) No CKD; excluding patients with overlapping eGFR values between the groups, one day up to 30 days before the LAAO procedure. Statistical methods All statistical analyses were conducted using the TriNetX platform. The characteristics of the two groups were presented as mean ± standard deviation (SD) for continuous variables and as frequencies and percentages for dichotomous outcomes, employing an independent-sample Student’s t-test or the Chi-square test, respectively. For outcome analysis, the odds ratio (OR) with a 95% confidence interval (95% CI) was calculated to compare the odds between cohorts. Survival analysis was performed by depicting the Kaplan-Meier and estimating time-to-event probabilities. The log-rank test was applied to assess differences in survival distributions between groups, while Cox proportional hazards regression was used to calculate hazard ratios (HRs) with 95% CI. A two-sided P-value of less than 0.05 was considered statistically significant. The proportional hazard assumption was tested using the generalized Schoenfeld approach built into the TriNetX platform. A multivariable Cox proportional hazards model was used to assess the independent association of CKD/ESRD vs. no CKD on primary outcomes, adjusting for baseline confounders in the unmatched groups. Moreover, we performed a competing risk analysis using Cumulative Incidence Functions (CIF) via the Aalen-Johansen estimator. To assess the impact of treating competing risks (all-cause mortality) as non-informative censoring, we calculated the absolute difference between the Kaplan-Meier (1-survival) estimates and the Aalen-Johansen CIF along the follow-up. This allowed us to assess the potential overestimation bias inherent in traditional survival analysis for all outcomes. Data visualization was performed in R. An E-sensitivity analysis was conducted through the ‘Evalue’ R package to account for the potential influence of undetected confounders for the primary and significant secondary outcomes.18 Results Study population and patient characteristics A total of 10 689 patients met the study's inclusion criteria, with 4140 classified as CKD/ESRD and 6549 as No CKD. After 1:1 PSM based on baseline characteristics, 3085 patients remained in each cohort for the final analysis (Figure 1). After PSM, the cohorts were well-balanced in all baseline characteristics (SMD < 0.1), including age (76.2 ± 7.5 years vs. 76.3 ± 7.4 years, P = 0.666), proportion of males (58.0% vs. 59.2%, P = 0.326), prevalence of persistent AF (36.3% vs. 36.4%, P = 0.937), and heart failure (41.3% vs. 41.5%, P = 0.877). More details about the baseline characteristics are outlined in Tables 1 and Supplementary material online, Table S4. Baseline characteristics according to CKD stages vs. no CKD are reported in Supplementary material online, Table S5. CONSORT flow diagram of the selection process. Table 1Baseline characteristics of the study cohort before and after propensity score matchingCharacteristicCKD/ESRDn = 3085No CKDn = 3085 P-valueaSMD Demographics Age at Index, years, Mean (SD)76.2 (7.5)76.3 (7.4)0.6660.01 Male, n (%)1788 (58.0)1826 (59.2)0.3260.03Race, n (%) Black or African American193 (6.3)189 (6.1)0.8330.01 White2629 (85.2)2625 (85.1)0.886< 0.01 Hispanic or Latino, n (%)77 (2.5)81 (2.6)0.7470.01 Vital Signs BMI ≥ 30 Kg/m2, n (%)1393 (45.2)1396 (45.3)0.939< 0.01 Systolic Blood Pressure, mm [Hg], Mean (SD)131.4 (20.3)131.6 (19.6)0.7530.01 Diastolic Blood Pressure, mm [Hg], Mean (SD)71.3 (11.7)72.2 (11.7)0.004**0.08 Heart Rate, Mean (SD)71.6 (14.3)71.5 (13.7)0.6850.01 Comorbidities Persistent AF, n (%)1121 (36.3)1124 (36.4)0.937< 0.01 Cardiovascular Procedures, n (%)b2601 (84.3)2645 (85.7)0.1160.04 Essential Hypertension, n (%)2338 (75.8)2358 (76.4)0.5500.02 Heart Failure, n (%)1274 (41.3)1280 (41.5)0.877< 0.01 Diabetes Mellitus, n (%)1044 (33.8)1041 (33.7)0.936< 0.01 Dyslipidemia, n (%)2004 (65.0)2006 (65.0)0.957< 0.01 Ischaemic Heart Disease, n (%)1596 (51.7)1584 (51.3)0.7600.01 Thyroid Disorders, n (%)606 (19.6)604 (19.6)0.949< 0.01 Chronic Obstructive Pulmonary Disease, n (%)755 (24.5)791 (25.6)0.2900.03 Obstructive Sleep Apnea, n (%)650 (21.1)644 (20.9)0.851< 0.01 Liver Disease, n (%)229 (7.4)232 (7.5)0.885< 0.01 Peptic Ulcer, n (%)47 (1.5)54 (1.8)0.4820.02 Peripheral Vascular Disease, n (%)323 (10.5)343 (11.1)0.4120.02 Cerebral Infarction, n (%)419 (13.6)432 (14.0)0.6310.01 Neoplasms, n (%)782 (25.3)777 (25.2)0.884< 0.01 Medications Beta-Blockers, n (%)1912 (62.0)1890 (61.3)0.5650.01 Calcium Channel Blockers, n (%)1179 (38.2)1202 (39.0)0.5480.02 Diuretics, n (%)1535 (49.8)1556 (50.4)0.5930.01 ACE Inhibitors, n (%)643 (20.8)652 (21.1)0.7780.01 ARBs, n (%)717 (23.2)719 (23.3)0.952< 0.01 Digoxin, n (%)198 (6.4)208 (6.7)0.6080.01 Insulin, n (%)641 (20.8)613 (19.9)0.3760.02 Metformin, n (%)295 (9.6)285 (9.2)0.6630.01 Sulfonylureas, n (%)159 (5.2)151 (4.9)0.6410.01 Anti-Hypertensives, n (%)549 (17.8)549 (17.8)1.000< 0.01 Vasoprotectives, n (%)2148 (69.6)2164 (70.1)0.6570.01 Glucocorticoids, n (%)1230 (39.9)1238 (40.1)0.8350.01 NSAIDs, n (%)302 (9.8)302 (9.8)1.000< 0.01 Proton Pump Inhibitors, n (%)1301 (42.2)1317 (42.7)0.6800.01 H2-Receptor Antagonists, n (%)372 (12.1)382 (12.4)0.6970.01 Organic Nitrates, n (%)675 (21.9)659 (21.4)0.6210.01 Anticoagulants, n (%)2147 (69.6)2126 (68.9)0.5620.01 Heparin, n (%)838 (27.2)836 (27.1)0.954< 0.01 Enoxaparin, n (%)360 (11.7)369 (12.0)0.7230.01 Bivaleriudin, n (%)18 (0.6)22 (0.7)0.5260.02 Apixaban, n (%)1231 (39.9)1210 (39.2)0.5850.01 Rivaroxaban, n (%)448 (14.5)432 (14.0)0.5600.01 Dabigatran, n (%)67 (2.2)78 (2.5)0.3550.02 Warfarin, n (%)495 (16.0)493 (16.0)0.945< 0.01 Platelets Aggregation Inhibitors, n (%)1211 (39.3)1219 (39.5)0.8350.01 Aspirin, n (%)1071 (34.7)1095 (35.5)0.5220.02 Clopidogrel, n (%)439 (14.2)426 (13.8)0.6340.01 Ticagrelor, n (%)29 (0.9)30 (1.0)0.896< 0.01 Anti-Arrhythmic Drugs Class I/III, n (%)1821 (59.0)1839 (59.6)0.6410.01 Antilipemic Agents, n (%)1627 (52.7)1620 (52.5)0.858< 0.01 Laboratory GFR, mL/min, Mean (SD)c43.4 (14.5)76.4 (17.5)< 0.001***2.05 Creatinine, mg/dL, Mean, (SD)c1.8 (3.0)0.9 (0.2)< 0.001***0.41 Hgb < 10 g/dL, n (%)d932 (30.2)916 (29.7)0.6570.01 Total Cholesterol, mg/dL, Mean (SD)144.1 (40.4)143.7 (40.4)0.8210.01 LDL-C, mg/dL, Mean (SD)73.6 (35.8)73.6 (32.2)0.973< 0.01 HDL-C, mg/dL, Mean (SD)44.0 (18.7)44.0 (19.8)0.966< 0.01 Triglycerides, mg/dL, Mean (SD)121.6 (72.3)116.9 (71.0)0.1090.07 LVEF ≤ 40%, n (%)59 (1.9)59 (1.9)1.000< 0.01 CRP, mg/L, Mean (SD)31.2 (52.1)29.3 (44.9)0.6430.04 BNP ≥ 150 pg/mL, n (%)460 (14.9)465 (15.1)0.858< 0.01 NT-proBNP ≥ 450 pg/mL, n (%)339 (11.0)322 (10.4)0.4840.02 INR, Mean (SD)1.4 (0.6)1.4 (0.6)0.6830.01CKD: chronic kidney disease; ESRD: end-stage renal disease; SMD: standardized mean difference; n: number; SD: standard deviation; AF: atrial fibrillation; BMI: body mass index; ACE: angiotensin-converting enzyme; ARBs: angiotensin II receptor blockers; NSAIDs: nonsteroidal anti-inflammatory drugs; GFR: glomerular filtration rate; Hgb: haemoglobin; LDL: low-density lipoprotein; HDL: high-density lipoprotein; HBA1c: haemoglobin A1C; LVEF: left ventricular ejection fraction; CRP: C-reactive protein; NT-proBNP: n-terminal prohormone of brain natriuretic peptide; BNP: B-Type Natriuretic Peptide; and INR: international normalized ratio. a*P < 0.05; **P < 0.01; ***P < 0.001 bCardiovascular procedures include cardiography, echocardiography, catheterization, cardiac devices, and electrophysiological procedures. cThese variables were not included in the propensity score matching model. dHaemoglobin (g/dL), Mean ± SD: CKD/ESRD group (12.2 ± 2.0) vs. No CKD group (12.7 ± 2.0), P < 0.001. Baseline characteristics of the study cohort before and after propensity score matching CKD: chronic kidney disease; ESRD: end-stage renal disease; SMD: standardized mean difference; n: number; SD: standard deviation; AF: atrial fibrillation; BMI: body mass index; ACE: angiotensin-converting enzyme; ARBs: angiotensin II receptor blockers; NSAIDs: nonsteroidal anti-inflammatory drugs; GFR: glomerular filtration rate; Hgb: haemoglobin; LDL: low-density lipoprotein; HDL: high-density lipoprotein; HBA1c: haemoglobin A1C; LVEF: left ventricular ejection fraction; CRP: C-reactive protein; NT-proBNP: n-terminal prohormone of brain natriuretic peptide; BNP: B-Type Natriuretic Peptide; and INR: international normalized ratio. a*P < 0.05; **P < 0.01; ***P < 0.001 bCardiovascular procedures include cardiography, echocardiography, catheterization, cardiac devices, and electrophysiological procedures. cThese variables were not included in the propensity score matching model. dHaemoglobin (g/dL), Mean ± SD: CKD/ESRD group (12.2 ± 2.0) vs. No CKD group (12.7 ± 2.0), P < 0.001. Primary outcomes. After PSM, CKD/ESRD was significantly associated with a 18% higher risk of major bleeding [HR: 1.179, 95% CI (1.080, 1.288), P < 0.001] compared to no CKD, with no significant difference between CKD/ESRD vs. no CKD in ischaemic stroke/TIA/SE [HR: 1.019, 95% CI (0.909, 1.141), P = 0.750] (Table 2 and Figures 2 and 3). Outcomes result before PSM are reported in Supplementary material online, Table S6. Our multivariable Cox proportional hazards model showed that CKD/ESRD was significantly associated with a 22% higher risk of major bleeding [aHR: 1.215, 95% CI (1.125, 1.311), P < 0.001] compared to no CKD. However, there was no significant difference between CKD/ESRD vs. no CKD in ischaemic stroke/TIA/SE [aHR: 1.005, 95% CI (0.910, 1.110), P = 0.924]. Table 2Clinical outcomes of the study cohort after propensity score matchingOutcome n Event nRisk [%]ORa95 CIa P-valuebThree-years survival [%]HRa95 CIa P-valueb Primary Outcomes Ischaemic Stroke/TIA/SE c No CKD308559319.2——78.76—— CKD/ESRD308559519.31.004[0.885, 1.140]0.94978.241.019[0.909, 1.141]0.750 Major Bleeding c No CKD308593130.2——67.00—— CKD/ESRD3085106134.41.213[1.090, 1.350]< 0.001***61.891.179[1.080, 1.288]< 0.001*** Secondary Outcomes Acute Myocardial Infarction c No CKD30852528.2——90.46—— CKD/ESRD308535111.41.443[1.217, 1.711]< 0.001***86.371.452[1.235, 1.708]< 0.001*** Acute Heart Failure d No CKD308556818.4——79.24—— CKD/ESRD308576024.61.449[1.282, 1.637]< 0.001***71.741.411[1.265, 1.573]< 0.001*** Acute Kidney Injury d No CKD308541713.5——84.43—— CKD/ESRD308593130.22.765[2.431, 3.145]< 0.001***65.152.543[2.266, 2.855]< 0.001*** Venous Thromboembolism c No CKD30852106.8——92.07—— CKD/ESRD30852207.11.051[0.864, 1.279]0.61791.401.075[0.890, 1.299]0.452 Gastrointestinal Bleeding c No CKD308555718.1——80.23—— CKD/ESRD308568622.21.298[1.145, 1.471]< 0.001***75.271.276[1.141, 1.427]< 0.001*** Pericardial Effusion/Tamponade d No CKD308538412.4——87.14—— CKD/ESRD308542013.61.109[0.956, 1.286]0.17385.621.099[0.957, 1.262]0.175 Cardiac-Related Haemodynamic Instability d No CKD3085742.4——97.29—— CKD/ESRD30851484.82.050[1.544, 2.723]< 0.001***94.302.064[1.561, 2.728]< 0.001*** All-Cause Hospitalization c No CKD308599432.2——63.39—— CKD/ESRD3085116737.81.280[1.153, 1.421]< 0.001***56.571.245[1.144, 1.355]< 0.001*** All-Cause Mortality d No CKD308543614.1——83.94—— CKD/ESRD308563820.71.584[1.386, 1.811]< 0.001***76.661.526[1.351, 1.724]< 0.001*** aOR = Odds Ratio, HR = Hazard Ratio, CI = Confidence Interval b*P < 0.05; **P < 0.01; ***P < 0.001 cOutcomes reported starting from the seventh day after the index event dOutcomes reported starting from the same day of the index eventTIA: Transient Ischaemic Attack and SE: Systemic Embolism. Clinical outcomes of the study cohort after propensity score matching aOR = Odds Ratio, HR = Hazard Ratio, CI = Confidence Interval b*P < 0.05; **P < 0.01; ***P < 0.001 cOutcomes reported starting from the seventh day after the index event dOutcomes reported starting from the same day of the index event TIA: Transient Ischaemic Attack and SE: Systemic Embolism. A) forest plot of summary of outcomes after propensity score matching and B) bar graph of rates of study outcomes after propensity score matching. CKD: Chronic Kidney Disease, ESRD: End-Stage Renal Disease, TIA: Transient Ischemic Attack, SE: Systemic Embolism, and CRHI: Cardiac-Related Hemodynamic Instability. Kaplan-meier curves after propensity score matching. A) Ischemic Stroke/Transient Ischemic Attack/Systemic Embolism, and B) Major Bleeding. Compared to no CKD, CKD and ESRD were significantly associated with a 17% higher risk of major bleeding [aHR: 1.173, 95% CI (1.083, 1.270), P < 0.001] and a 25% higher risk of major bleeding [aHR: 1.247, 95% CI (1.052, 1.478), P = 0.011], respectively. However, there was no significant difference between CKD vs. no CKD and ESRD vs. no CKD in ischaemic stroke/TIA/SE [aHR: 1.007, 95% CI (0.908, 1.116), P = 0.893] and [aHR: 0.925, 95% CI (0.720, 1.188), P = 0.540], respectively (see Supplementary material online, Figures S2 and S3). Detailed results from the multivariable Cox proportional hazards model are presented in Supplementary material online, Tables S7 and S8. Secondary outcomes. After PSM, CKD/ESRD was significantly associated with a 45% higher risk of AMI [HR: 1.452, 95% CI (1.235, 1.708), P < 0.001], a 41% higher risk of AHF [HR: 1.411, 95% CI (1.265, 1.573), P < 0.001], more than two-fold higher risk of AKI [HR: 2.543, 95% CI (2.266, 2.855), P < 0.001], a 28% higher risk of GIB [HR: 1.276, 95% CI (1.141, 1.427), P < 0.001], more than two-fold higher risk of CRHI [HR: 2.064, 95% CI (1.561, 2.728), P < 0.001], a 25% higher risk of all-cause hospitalizations [HR: 1.245, 95% CI (1.144, 1.355), P < 0.001], and a 53% higher risk of all-cause mortality [HR: 1.526, 95% CI (1.351, 1.724), P < 0.001], with no significant difference in VTE and pericardial effusion/tamponade (Table 2 and Supplementary material online, Figures S4-S12). All outcomes according to CKD stages before PSM are reported in Supplementary material online, Table S9. Differences in oral anticoagulants prescriptions during the follow-up period before and after PSM are reported in Supplementary material online, Tables S10 and S11. Competing risk analyses and overestimation bias estimation for all outcomes are presented in Supplementary material online, Figures S13-S22. Peri-procedural analysis for all outcomes results before and after PSM are reported in Supplementary material online, Tables S12 and S13. All proportionality P-values for all outcomes are reported in Supplementary material online, Tables S14 and S15. E-sensitivity analysis. Supplementary material online, Table S16 reports the E-values. The E-value for ischaemic stroke/SE is 1.13, and major bleeding is 1.5, indicating that the association may be sensitive to unmeasured confounding. Discussion Our analysis showed that CKD did not significantly affect ischaemic stroke/TIA/SE but was linked to higher risks of adverse events, including major bleeding, GIB, AMI, AHF, AKI, CRHI, all-cause hospitalizations, and all-cause mortality. Patients with AF and CKD show a significantly elevated risk for both stroke and bleeding events when compared to patients without CKD.19 In patients with CKD, the coagulation cascade is sustainedly activated, with elevated levels of homocysteine and heightened platelet activation, all contributing to a heightened risk of thrombosis.20 The relationship between kidney function and the risk of incident stroke is inverse and stepwise.21 Despite the higher thrombotic vulnerability of patients with CKD/ESRD, our matched analysis showed no significant difference in ischaemic stroke/TIA/SE between the two groups after LAAO at three-year follow-up. This is consistent with Munir et al., who reported similar 1-year ischaemic stroke and SE rates across non-CKD, CKD, and kidney failure with replacement therapy (KFRT) groups in the U.S. NCDR LAAO Registry after LAAO,14 and with Della Rocca et al., who found that LAAO preserved its stroke-preventive efficacy across renal function categories.13 Thus, LAAO may reduce appendage-related thromboembolic risk in CKD/ESRD, while residual strokes may arise from non-appendage sources such as aortic or valvular atherosclerosis and small-vessel cerebrovascular disease.22 Our study showed that CKD/ESRD was associated with a higher risk of both major bleeding and GIB after LAAO at three-year follow-up. Our findings were consistent with Munir et al., who demonstrated a stepwise increase in 1-year major bleeding and peri-procedural GIB with worsening CKD severity (both P < 0.0001).14 CKD is recognized as a significant risk factor for perioperative complications, particularly due to an increased likelihood of bleeding.23 This bleeding diathesis is attributed to two key aspects of the underlying pathophysiology associated with CKD: (1) the accumulation of uraemic toxins, which leads to platelet dysfunction, and (2) the presence of anaemia, which further compromises normal coagulation processes.23,24 This could elucidate the increased risk of major bleeding and GIB as kidney function deteriorates. Our analysis showed that CKD/ESRD was significantly associated with a higher risk of AKI compared to no CKD after LAAO. Sedaghata et al. identified the contrast volume-to-GFR ratio as a key predictor of AKI,25 while Brockmeyer et al. found contrast-induced AKI to be the most common adverse event in CKD patients undergoing LAAO, despite similar mean contrast volumes between CKD and non-CKD groups.26 These observations have motivated contrast-sparing strategies. Rodriguez et al. demonstrated the feasibility of zero-contrast LAAO.27 However, there was no significant difference between CKD/ESRD and no CKD in VTE and pericardial effusion/tamponade at three-year follow-up. Likewise, Munir et al. observed comparable 1-year event rates, with pericardial effusion and tamponade occurring in about 0.8% of both CKD and non-CKD groups, and systemic thromboembolism being exceedingly rare.14 These stable results are expected, as such complications are more closely related to procedural technique, device handling, vascular access, and peri-procedural care than to renal function itself. Patients with CKD and ESRD remain clinically vulnerable even after successful cardiovascular procedures, with higher morbidity and mortality reported.28–30 Our findings are consistent with Tanaka et al., who reported significantly higher all-cause and cardiovascular mortality in haemodialysis patients who underwent LAAO.31 Similarly, López-Tejero et al. found higher 5-year mortality in advanced CKD who underwent LAAO,32 and Munir et al. noted increased 1-year mortality among CKD and KFRT cohorts.14 The higher mortality after LAAO in CKD/ESRD patients likely reflects their broader cardio-renal burden rather than a device-related effect, driven by uraemic cardiomyopathy, vascular calcification, accelerated atherosclerosis, inflammation, malnutrition, and dialysis-related haemodynamic stress33–37 Clinical implications and future research From a clinical standpoint, these findings indicate that patients with CKD/ESRD undergoing percutaneous LAAO should be managed as a high-risk group, particularly for bleeding, consistent with the ARC-HBR framework that identifies severe renal impairment as a major criterion.38 Preventive care should include pre-procedural volume optimization, renal-protective hydration, avoidance of nephrotoxins, and contrast-sparing techniques to reduce the risk of AKI.39,40 Procedural safety may also be improved through ultrasound-guided vascular access and careful transseptal puncture under intracardiac or transoesophageal echocardiographic guidance.41 After implantation, antithrombotic therapy should be individualized rather than routine, balancing device-related thrombus prevention against the elevated bleeding risk, with consideration of simplified or abbreviated regimens when appropriate.41 These recommendations align with modern AF guidelines that support LAAO in patients unsuitable for long-term anticoagulation, with an emphasis on shared decision-making and multidisciplinary care.42 Despite improvements in LAAO devices and procedural safety, the optimal post-LAAO strategy in CKD/ESRD remains uncertain. Dedicated randomized trials and prospective registries are needed to define kidney-function–specific antithrombotic approaches, standardize bleeding and renal safety outcomes, and determine whether renal-risk–adapted LAAO pathways can preserve stroke prevention while reducing non-stroke complications in this vulnerable population. Strengths and limitations This study used the large TriNetX U.S. Collaborative Network, providing a broad and representative real-world population. PSM helped reduce confounding and ensured balanced comparison between CKD/ESRD and non-CKD groups. Stratified analyses across CKD stages offered insight into outcome gradients by renal severity. The use of E-value sensitivity analysis to test for unmeasured confounding further strengthened the reliability of the findings. However, there were several limitations. 1) The retrospective design prevents firm causal conclusions, and residual confounding may remain. 2) TriNetX relies on ICD, CPT, and RxNorm coding, which can lead to misclassification and limit procedural detail (e.g. LAAO device type, imaging findings). 3) CKD staging according to eGFR values was based on a single eGFR from one day to 30 days before the LAAO procedure, which may not reflect long-term kidney trends. 4) Antithrombotic therapy data were limited to prescription records, without adherence details. 5) Unmeasured factors such as frailty, nutritional status, or operator experience were unavailable. 6) Due to technical limitations in TriNetX, we could not apply a Fine-Gray Model. Instead, we used CIFs using the Aalen-Johansen estimator to account for the competing risk of all-cause mortality, and assessed the overestimation bias inherent in the traditional survival analysis for all outcomes. Conclusion Our study suggests that CKD/ESRD was significantly associated with a higher risk of major bleeding, AMI, AHF, AKI, CRHI, all-cause hospitalizations, and all-cause mortality, with no significant difference in ischaemic stroke/TIA/SE among patients with AF undergoing LAAO. Therefore, these patients should be managed carefully with tailored peri-procedural care, cautious use of contrast, and personalized antithrombotic treatment. Future trials focusing on kidney function are needed to guide safer and more effective treatment strategies. Supplementary Material Acknowledgements None. Contributor Information Ahmed Mazen Amin, Faculty of Medicine, Mansoura University, 60 El Gomhouria Street, 35516 Mansoura, Egypt. Ameer Awashra, Department of Medicine, An Najah National University, Rafidia Street, P606 Nablus, Palestine. Ubaid Khan, Division of Cardiology, University of Maryland School of Medicine, 22 S. Greene St, Baltimore, Maryland 21201, USA. Christopher V DeSimone, Department of Cardiovascular Medicine, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA. Supplementary material Supplementary material is available at European Heart Journal Open online. Authors contributions. Ahmed Mazen Amin [Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing (lead)], Ameer Awashra [Methodology, Validation, Writing—original draft, Writing—review & editing (equal)], Ubaid Khan (Data curation, Resources [lead], Investigation, Methodology, Validation [equal], Writing—review & editing [supporting]), and Christopher V. Desimone (Supervision, Writing—review & editing [lead], Writing—original draft [supporting]) Ethics approval and consent to participate. Due to the deidentified nature of the platform's patient data and HCOs in compliance with the Section §164.514 of the HIPAA Privacy Rule, ethical approval was deemed unnecessary. The study complied with the ethical and reporting standards depicted by the TriNetX platform under ‘Publication Guidelines’. Funding We received no funding for this study. References Associated Data Supplementary Materials