Understanding late nonfatal events after mechanical aortic valve replacement: Importance of definitions and thoroughness of follow-up Authors: Harrison R Herrera, Hartzell V Schaff, Joseph A Dearani, Gabor Bagameri, Kevin L Greason, Austin Todd, Juan A Crestanello Published: 2026 ABSTRACT AbstractObjectiveWe evaluated how event definitions, data-collection methods, and follow-up completeness influence reporting of long-term nonfatal complications after mechanical aortic valve replacement in local versus referral patients.MethodsWe analyzed 558 patients undergoing primary mechanical aortic valve replacement from 2000 to 2023. Patients were stratified by residential distance (<50 miles, local vs >50 miles, referral) to assess completeness of follow-up. Outcomes included major bleeding, neurological thromboembolic events, and survival. Major bleeding was defined by International Society on Thrombosis and Hemostasis criteria or by restricted International Classification of Diseases codes. Events were captured by manual review and by International Classification of Diseases coding. Propensity score–matched local (n = 186) and referral groups (n = 372) were analyzed using Fine-Gray competing risk models for nonfatal outcomes and Kaplan–Meier analysis for survival.ResultsBaseline and perioperative characteristics of propensity score–matched local and referral patients were comparable. Median clinical follow-up was longer in the less than 50-mile group (10.2 vs 4.1 years, P < .001). Local patients had higher crude event rates for major bleeding (40.9% vs 24.2%, P < .001) and neurological events (13.4% vs 4.0%, P < .001). These differences were mitigated but still statistically significant in time-dependent competing risk analysis. Rates varied with ascertainment method: International Classification of Diseases coding amplified differences compared with manual adjudication. Long-term survival did not differ significantly between groups (P = .44).ConclusionsReporting late nonfatal events after aortic valve replacement varies with event definitions, data-collection methods, and follow-up completeness. Although mortality estimates remain robust, bleeding and stroke may be underestimated in referral patients. Competing risk analysis enables more accurate interpretation of late outcomes.Key Words: aortic valve replacement, ascertainment bias, major bleeding, nonfatal complications, stroke Nonfatal event incidence varies by definition, adjudication, and follow-up completeness. Central Message Late nonfatal event reporting after mechanical AVR is influenced by factors such as event definitions, data-collection methods, and patient residence, requiring cautious interpretation. Perspective Missing data in reporting late nonfatal events after mechanical AVR may mislead interpretation and affect clinical accuracy. These data gaps arise from factors such as loss to follow-up due to patient residence, event definitions, data-collection methods, and statistical approaches, potentially biasing assessment of procedural, management strategies, or institutional effects. Analysis of outcomes after aortic valve replacement (AVR) includes assessment of both mortality and important nonfatal complications, particularly stroke and bleeding events.1,2 Databases such as the National Death Index provide reasonably up-to-date information on patient mortality; however, accurate reporting of nonfatal complications depends on several factors, such as event definitions and methods of data collection, as well as statistical approach and completeness of follow-up, which may be limited by patient residence. Single-institution and national database studies contain outcome information from variable patient encounters at irregular intervals. Because of this variability, some may question the accuracy of reporting of late event rates. To address the impact of these factors on the accuracy of late event reporting, we analyzed late outcomes in 558 patients undergoing AVR with mechanical valves, comparing local/regional patients within our care network with referral patients who largely obtained medical care at their home institutions. Outcomes were analyzed using alternate event definitions, methods of data collection, and statistical approaches. We hypothesized that capture of nonfatal late events would be under-reported in referral patients who did not receive care at our clinic and that the observed event rates would vary depending on the definition used and statistical approach applied. Materials and Methods Patients and Study Design The Mayo Clinic Institutional Review Board in Rochester, Minnesota, approved the study (IRB 25-007336; approved August 1, 2025). Consent was waived. All patients authorized the use of their clinical data for research. Between 2000 and 2023, 3596 patients underwent primary mechanical AVR at our institution. Patients who underwent concomitant procedures (aortic surgery, coronary artery bypass grafting, mitral or tricuspid valve surgery, septal myectomy, or others), nonelective surgery, or prior sternotomy were excluded (n = 2574) (Figure 1). The remaining 1022 patients constituted the study cohort before stratification and matching, as described in the completeness of follow-up section. Flow diagram of the study population. AVR, Aortic valve replacement. Data Collection Clinical and operative data were obtained from a prospectively maintained surgical database and a retrospective review of electronic health records (EHRs). The primary study outcomes were reported major bleeding events, neurological thromboembolic complications, and survival from the surgery date through the last available follow-up visit or electronic documentation in the EHR, up to April 2025. Only the first was included in the analysis for patients with more than 1 event. Vital status was determined through LexisNexis Accurint, which consolidates information from multiple sources, including the Social Security Death Master File and state death records (New York City, NY). Outcome events were collected by 2 methods. We used end points identified by the International Classification of Diseases (ICD) codes (ICD 9th and 10th Revisions) and through manual review of EHR documentation, including records from referring providers and external healthcare organizations, and direct patient communication. External records from referral patients were often available only as scanned or electronic documents without ICD coding. Access to external records was facilitated through the Health Information Exchange initiative, which enables cross-institutional record sharing within the Epic system when patients provide explicit consent. Additional clinical information was obtained from follow-up surveys.3 Variation in Definitions of Outcomes Major bleeding was defined in 2 ways for comparative analysis. A restricted definition, used in many studies and easily obtained from ICD codes, includes gastrointestinal bleeding, intracranial bleeding, and hemoptysis. A broader definition follows the International Society on Thrombosis and Hemostasis (ISTH) criteria:4 (1) fatal bleeding; (2) symptomatic bleeding in a critical area or organ (intracranial, intraspinal, intraocular, retroperitoneal, intra-articular, pericardial, or intramuscular with compartment syndrome); or (3) bleeding resulting in a hemoglobin decrease 2 g/dL or more or requiring transfusion of 2 units or more of whole blood or red cells. Under the ISTH criteria, we included any bleeding type (gastrointestinal, hematuria, menorrhagia, epistaxis, post-traumatic, or postprocedural hemorrhage while on anticoagulation) that met the transfusion or hemoglobin threshold, regardless of whether the event was acute or subacute. We compared rates of late bleeding defined by ICD codes versus bleeding event rates calculated from manual record review using ISTH definitions. Neurological thromboembolic complications included ischemic stroke and transient ischemic attack (TIA), defined according to the American Heart Association/American Stroke Association.5,6 Stroke was defined as an episode of neurological dysfunction caused by focal cerebral infarction, evidenced by pathology or imaging. TIA was defined as a transient episode of neurological dysfunction caused by focal brain ischemia without acute infarction. Variation in Completeness of Follow-Up To assess the effect of follow-up completeness on identification of late nonfatal events, patients were stratified by residential distance from our institution into 2 groups: those local patients residing within 50 miles (<50-mile group), who largely obtained follow-up care at Mayo Clinic, and those living beyond 50 miles (>50-mile group), referral patients, who generally obtained follow-up elsewhere. The 50-mile cutoff was selected as an approximate midpoint between Mayo Clinic and other major healthcare systems in our region. In this analysis of the influence of completeness of follow-up defined by geographic radius, we used propensity score matching to mitigate any potential bias due to baseline characteristics. The initial study cohort included 1022 patients, 213 in the less than 50 miles radius group and 809 in the more than 50 miles radius group. The variables used in the propensity score and model details are presented in Tables E1 and E2, and Figure E1. After 1:2 matching, the final study cohort included 186 patients in the less than 50 miles group and 372 patients in the more than 50 miles radius group. Dot plot of standardized mean difference before and after propensity score matching, demonstrating adequate covariate balance after matching (with SMD <0.2 in absolute values). NYHA, New York Heart Association. Statistical Analysis Categorical variables were reported as numbers (percentage), and continuous variables were reported as medians (25th through 75th interquartile range [IQR]) or means (SD). To account for baseline differences, our more than 50-mile radius patients were 2:1 propensity score matched to our less than 50-mile radius patients. The propensity scores were derived from a multivariable logistic regression for predicting distance groups using potential confounders. Nearest neighbor matching was used with a caliper of 0.2 of the SD of the propensity scores. Unmatched comparisons between groups were made using Wilcoxon rank-sum or Pearson chi-square tests as appropriate. For the primary outcome, group comparisons of unadjusted cumulative incidence rates for major bleeding and neurological thromboembolic events were estimated using the Fine-Gray sub-distribution hazard regression model with all-cause mortality as the competing risk. The Kaplan–Meier method was used for survival analysis, and differences between groups were analyzed using the 2-tailed log-rank test. Statistical significance was defined at a P < .05. All statistical analyses were performed using R software (version 4.2.2; R Foundation for Statistical Computing). Results Study Population and Baseline Patient Characteristics As seen in Table 1, most baseline characteristics of matched patients were not significantly different compared with those living within 50 miles of the surgical center (N = 186) and those residing farther than 50 miles (N = 372). The mean age was 58 years in both groups, with a similar gender male predominance and White racial composition. Comorbidities associated with cardiovascular risk were comparable between the groups. Aortic valve stenosis was present in more than 84% of patients in both groups, with a mean aortic valve gradient of 47.5 ± 17.4 in local patients versus 51.1 ± 14.6 mm Hg in those residing more than 50 miles from our center. The most common indications for mechanical AVR were bicuspid valve and degenerative fibrocalcific disease. More than 80% of patients in both groups had mild to moderate functional impairment based on the New York Heart Association classification (II and III), with comparable preoperative left ventricular ejection fractions in the 2 groups. Table 1Baseline characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD, median (interquartile range) or N (%)VariableMissing<50 miles(n = 186)>50 miles(n = 372)P valueAge (y)58 ± 10.658 ± 10.2.59Male sex128 (68.8%)249 (66.9%).65Follow-up time (y)10.2 (4.9-17.0)4.1 (0.03-10.6)<.001White12176 (95.7%)347 (95.9%).91Body mass index (kg/m2)31.4 ± 6.931.5 ± 7.8.92STS predicted risk of mortality351.53 ± 1.91.48 ± 1.5.77NYHA class.74 I24 (12.9%)43 (11.6%) II82 (44.1%)169 (45.4%) III74 (39.8%)153 (41.1%) IV6 (3.2%)7 (1.9%)Dyslipidemia139 (74.7%)273 (73.4%).73Hypertension122 (65.6%)247 (66.4%).85Diabetes mellitus37 (19.9%)74 (19.9%)1.0Cerebrovascular disease14 (7.5%)28 (7.5%)1.0History of atrial fibrillation/flutter16 (8.6%)47 (12.7%).32Current smoking21 (11.4%)47 (12.6%).66Dialysis6 (3.2%)8 (2.2%).44Previous myocardial infarction12 (6.5%)13 (3.5%).11Previous cerebrovascular accident5 (2.7%)13 (3.5%).61Indication for AVR.27 Bicuspid valve113 (60.8%)218 (58.6%) Degenerative fibrocalcific stenosis52 (28.0%)96 (25.8%) Regurgitation13 (7.0%)27 (7.3%) Rheumatic disease3 (1.6%)17 (4.6%) Active endocarditis2 (1.1%)1 (0.3) % Other3 (1.3%)13 (3.5%)Preoperative LVEF (%)60.4 ± 10.060.2 ± 9.6.84AV gradient10647.5 ± 17.451.1 ± 14.6.01AV stenosis158 (84.3%)313 (84.1%).80AV regurgitation12.48 None21 (11.4%)51 (14.1%) Trivial41 (22.3%)70 (19.3%) Mild39 (21.2%)88 (24.3%) Moderate28 (15.2%)64 (17.7%) Severe55 (29.9%)89 (24.6%)NYHA, New York Heart Association; STS, Society of Thoracic Surgeons; AVR, aortic valve replacement; LVEF, left ventricular ejection fraction; AV, aortic valve; N, number of patients. Baseline characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD, median (interquartile range) or N (%) NYHA, New York Heart Association; STS, Society of Thoracic Surgeons; AVR, aortic valve replacement; LVEF, left ventricular ejection fraction; AV, aortic valve; N, number of patients. The follow-up time (clinical follow up) for nonfatal events differed significantly (P < .001) between groups, with a median of 10.2 years (IQR, 4.9-17.0) in the less than 50-mile group and 4.1 years (IQR, 0.03-10.6) in the more than 50-mile group, reflecting longer follow-up and more exposure to nonfatal events among local patients. Preoperative and Early Postoperative Characteristics Perioperative variables were similar in the matched groups, as detailed in Table 2. Rates of postoperative complications did not differ significantly. However, patients living closer to the surgical center experienced a higher rate of pneumonia (3.2% vs 0%, P < .001) and permanent pacemaker implantation (6.5% vs 3.0%, P = .05). The length of intensive care unit and total in-hospital stay were similar between groups. Risks of reoperation for bleeding and mortality were low and similar in the 2 groups. Notably, the report of readmission within 30 days was significantly more common among patients living closer to the institution (13.5% vs 8.2%, P = .04). Table 2Perioperative characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD or N (%)VariableMissing<50 miles(n = 186)>50 miles(n = 372)P valueCardiopulmonary bypass time (min)78.7 ± 31.478.2 ± 32.0.87Aortic crossclamp time (min)60.0 ± 24.960.8 ± 25.0.69Intra-aortic ballon pump1 (0.5%)2 (0.5%)1.0Length of ICU stay (h)6127.2 ± 17.730.7 ± 22.8.85Postoperative blood transfusion38 (20.4%)102 (27.4%).07Atrial fibrillation42 (22.6%)93 (25.0%).52Stroke2 (1.1%)2 (0.5%).47Acute kidney injury1 (0.5%)7 (1.9%).20Dialysis required0 (0.0%)4 (1.1%).15Pneumonia6 (3.2%)0 (0.0%)<.001Permanent pacemaker12 (6.5%)11 (3.0%).05Reexploration for bleeding3 (1.6%)14 (3.8%).16Postoperative tamponade4 (2.2%)13 (3.5%).38Prolonged mechanical ventilation (>24 h)1 (0.5%)7 (1.9%).20Length of in-hospital stay (d)6.1 ± 2.26.2 ± 2.7.61Readmission within 30 d of dismissal25 (13.5%)30 (8.2%).04Operative mortality (30 d)1 (0.5%)4 (1.1%).52ICU, Intensive care unit; N, number of patients. Perioperative characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD or N (%) ICU, Intensive care unit; N, number of patients. Late Major Bleeding Events As seen in Table 3, the risk of major bleeding events identified by manual record review and expressed as a simple percentage was significantly greater among patients living less than 50 miles from our center compared with that of patients living a greater distance (40.9% vs 24.2%, P < .001). There was no important difference in bleeding site between the 2 patient groups. When bleeding was analyzed as time-to-event using the Fine-Gray sub-distribution hazard regression model with all-cause mortality as the competing risk, the cumulative incidence of major bleeding increased steadily over follow-up in both distance groups (Figure 2). Table 3Reported postoperative late nonfatal events after primary mechanical aortic valve replacement stratified by distance from the surgical centerComplicationsAll(n = 558)<50 miles(n = 186)>50 miles(n = 372)P valueMajor bleeding ICD codes16191 (48.9%)70 (18.8%)<.001Major bleeding ISTH/Manual16676 (40.9%)90 (24.2%)<.001Type of bleeding Gastrointestinal23 (30.2%)33 (36.6%) Intramuscular/hemarthrosis/post-traumatic16 (21.0%)14 (15.5%) Intracranial12 (15.7%)15 (16.6%) Bleeding postprocedure11 (14.4%)10 (11.1%) Hematuria/menorrhagia/epistaxis6 (7.8%)6 (6.6%) Intraocular5 (6.5%)7 (7.7%) Other3 (3.9%)5 (5.5%)Stroke - ICD codes5334 (18.2%)19 (5.1%)<.001Stroke - Manual4025 (13.4%)15 (4.0%)<.001ICD, International Classification of Diseases; ISTH, International Society on Thrombosis and Hemostasis; TIA, transient ischemic attack. Reported postoperative late nonfatal events after primary mechanical aortic valve replacement stratified by distance from the surgical center ICD, International Classification of Diseases; ISTH, International Society on Thrombosis and Hemostasis; TIA, transient ischemic attack. Cumulative incidence of major bleeding among propensity score–matched patients who underwent primary mechanical AVR, stratified by residential distance from the surgical center, with death as a competing risk (95% CI). A, Events defined by the ISTH criteria using manual data collection. B, Events defined by a restricted ICD code–based definition. The ISTH analysis showed similar cumulative incidence between groups, whereas the ICD-based analysis suggested lower rates in the more than 50-mile group. Using the ISTH definition of bleeding events (Figure 2, A), the estimated 20-year cumulative incidence was greater among patients residing within 50 miles compared with those living farther away (P = .013). At 10 years, the cumulative incidence was approximately 33% in the less than 50-mile group versus 27% in the more than 50-mile group. Although statistically significant, the difference in event rates between the groups was less than that observed when bleeding events were defined by ICD codes as reported below. When late bleeding events were identified by ICD-9 and ICD-10 codes (Table E3) for gastrointestinal, hemoptysis, and intracranial bleeding (Figure 2, B), the cumulative incidence curves demonstrated clear differences between local and more distant patients (P < .001). At 10 years, the cumulative incidence of bleeding events was 37% in the less than 50-mile group and 25% in the more than 50-mile group. Late Neurological Events Expressed as a simple percentage, the occurrence of neurologic thromboembolic events (stroke determined by manual record review) was significantly higher in the less than 50-mile group compared with the more than 50-mile group (13.4% vs 4.0%, P < .001) (Table 3). This difference was also reflected in the cumulative incidence analysis (Figure 3, A). At 10 years, the estimated cumulative incidence of stroke was 14% in the less than 50-mile group versus 6% in the more than 50-mile group, increasing to approximately 18% and 12% by 20 years (P = .003). Crude cumulative incidence of reported stroke among propensity score–matched patients who underwent primary mechanical AVR, stratified by residential distance from the surgical center, with death considered a competing risk (95% CI). A, Events identified by manual data collection. B, Events identified by ICD code–based collection. Both approaches showed similar cumulative incidence trends between groups, although the results and statistical significance varied by definition. As shown in Figure 3, B, when late neurologic events were captured by ICD-9 and ICD-10 codes (Table E3), event rates were higher than with manual record review, but still significantly greater rates in local patients compared with those residing more than 50 miles from our center (P < .001). At 10 years postoperatively, the estimated incidence of stroke defined by ICD coding was approximately 15% in the less than 50-mile group compared with 8% in the more than 50-mile group, and by 20 years risks reached approximately 25% versus 13%. When stroke and TIA were analyzed together based on manual record review (Figure E2, A), the cumulative incidence curves diverged progressively, with significantly higher event rates in the less than 50-mile group (P = .004). At 10 years, the estimated probability of stroke/TIA was 13% in the less than 50-mile group compared with 7% in the more than 50-mile group, and by 20 years the incidences reached approximately 22% and 15%, respectively. Likewise, when stroke/TIA was captured using ICD-9 and ICD-10 codes (Figure E2, B), both event rates and separation between groups were greater. At 10 years, the estimated cumulative incidence of stroke/TIA was 17% in the less than 50-mile group versus 11% in the more than 50-mile group, and by 20 years it reached approximately 30% compared with 20%, respectively (P < .001). Crude cumulative incidence of reported stroke/TIA among propensity score–matched patients who underwent primary mechanical AVR, stratified by residential distance from the surgical center, with death considered a competing risk (95% CI). A, Events identified by manual data collection. B, Events identified by ICD code–based collection. Both approaches showed similar cumulative incidence trends between groups, although the results and statistical significance varied by definition. TIA, Transient ischemic attack. Late Survival Kaplan–Meier survival analysis demonstrated no statistically significant difference in overall survival between patients living within 50 miles and those living farther than 50 miles from the surgical center (log-rank P = .44) (Figure 4). Survival probabilities remained high in both groups during the first decade and declined gradually over time; in 10 years, estimated survival was approximately 85% in both groups. By 20 years, survival was slightly higher in the less than 50-mile group (∼65%) compared with the more than 50-mile group (∼55%). Unadjusted survival of propensity score–matched patients who underwent primary mechanical AVR, stratified by residential distance from the surgical center. Survival was similar for local and referral patients (95% CI). Discussion In this study, we examined the impact of event definitions, data-collection methods, statistical approach, and completeness of follow-up—stratified by patient residency—on reporting nonfatal complications after mechanical AVR. We found considerable variability in results depending on these factors, ranging from differences between crude event rates and time-to-event analyses to shifts in trends and statistical significance across stratified groups (>50 miles vs <50 miles). In contrast, mortality data, collected through standardized and highly reliable sources, showed no statistically significant difference between groups, with comparable long-term trends. This detailed analysis provides an important reference against which to interpret nonfatal event reporting. The absence of major differences in baseline characteristics and perioperative outcomes between groups suggests that referral distance was not associated with demographic, clinical, or surgical risk factors. Our findings highlight the importance of thoroughness of follow-up when interpreting long-term nonfatal outcomes. We selected a 50-mile radius as the cutoff because it represents an approximate midpoint between our center and the nearest major metropolitan area and healthcare systems, beyond which patients were more likely to seek follow-up locally. As hypothesized, patients living closer to our surgical center had significantly longer median clinical follow-up, which led to a higher likelihood of event identification and reporting, and thus higher crude event rates. The proportion of major bleeding and neurological thromboembolic events was approximately 2-fold to 3-fold higher in the less than 50-mile group. However, after time-to-event analysis with adjustment for censoring and competing risks, these differences were attenuated. This suggests that the excess events observed among local patients are partially explained by longer follow-up rather than greater burden of complications. These observations reinforce a key statistical principle. Analyses based on raw event counts or percentages assume equal follow-up across groups, an assumption appropriate for fixed short-term outcomes such as 30-day complications but inappropriate for long-term outcomes spanning decades. In such settings, unequal follow-up introduces bias and may exaggerate differences. Time-to-event methodology, particularly competing risks analysis for nonfatal outcomes, provides a more accurate assessment of incidence over time and mitigates underreporting bias inherent to referral populations.7,8 An additional insight from this study was the magnitude of variability in event rates depending on the definition and method of data collection. When bleeding was identified only through ICD codes and restricted to gastrointestinal, hemoptysis, and intracranial events, the less than 50-mile group showed a significantly higher incidence compared with that of patients living farther from our center. This difference, although still statistically significant, was mitigated when major bleeding was adjudicated manually using the ISTH definition, highlighting how definitions and ascertainment methods shape outcome reporting. Manual review captures a broader spectrum of clinically relevant events, including those documented in scanned or unstructured records shared via the Health Information Exchange, whereas ICD coding may underestimate events in referral patients due to inconsistent coding. At the same time, ICD codes may overestimate bleeding by classifying minor events as major, because they do not consider hemoglobin decrease or transfusion requirements as specified in the ISTH criteria. These differences likely explain why the less than 50-mile group showed similar trends across both definition analyses, with the broader capture of ISTH offset by the less specific classification of ICD coding. Our findings also contribute to the ongoing discussion regarding the reliability of nonfatal late complication reporting from referral centers. Critics argue that referral institutions may underestimate complication rates because patients often return to their home communities for long-term care, where complications are managed and recorded elsewhere. Indeed, our findings support this notion but also highlight the important influence of event definitions and data-collection methods on the quantification of late nonfatal events after mechanical AVR. Limitations A limitation of the study is that it contains data from a single center and is retrospective. EHR documentation may be incomplete. Although our approach incorporated death as a competing event, cumulative incidence methods may still overestimate risk in the presence of unmeasured factors. We used an arbitrary but practical 50-mile cutoff to identify local and regional patients; using distance from our center as a continuous variable would be expected to add little to the analysis. Conclusions Clinicians should carefully examine definitions of events when interpreting late nonfatal complications after mechanical AVR. Future efforts to enhance data exchange, such as broader implementation of the Health Information Exchange initiative, may reduce gaps in accurate documentation of late complications of referral patients. Until then, our findings emphasize that application of time-to-event analysis and recognition of differences in event ascertainment are essential for accurate interpretation of long-term nonfatal events. Declaration of Generative AI and AI-Assisted Technologies in the Writing Process During the preparation of this manuscript, the author(s) used ChatGPT to assist in improving clarity and language during the writing process. No artificial intelligence tool was used for data analysis, interpretation, or the generation of scientific content. After using this tool, the author(s) carefully reviewed and edited the text and take full responsibility for the content of this publication. Conflict of Interest Statement The authors reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. Footnotes This study was approved by the Institutional Review Board at Mayo Clinic (Rochester, Minnesota), with a waiver for informed consent (25-007336, approved on August 1, 2025). Audio Audio 1. This is an audio recording of the presentation of the associated abstract and discussion. Appendix E1 Table E1Baseline characteristics of patients stratified by residential distance from the surgical center before propensity score matching, mean ± SD or N (%)Unmatched dataVariable<50 miles(n = 213)>50 miles(n = 809)P valueSMDAge (y)58.5 ± 10.656.8 ± 11.7.060.149Gender, female68 (31.9%)268 (33.1%).800.026Diabetes43 (20.2%)126 (15.6%).130.121Hypertension139 (65.3%)475 (58.7%).090.135Renal failure (dialysis or creatinine >2 mg/dL)8 (3.8%)21 (2.6%).490.067History of cerebrovascular accident7 (3.3%)25 (3.1%)10.011NYHA class.150.189 I25 (13.3%)136 (19.3%) II82 (43.6%)308 (43.8%) III75 (39.9%)247 (35.1%) IV6 (3.2%)13 (1.8%)Preoperative LVEF (%)60.76 ± 9.660.50 ± 9.71.720.027AFib/aflutter.710.1 No193 (90.6%)719 (88.9%) Yes17 (8.0%)68 (8.4%) Yes, but timing unknown3 (1.4%)21 (2.6%)Smoking status.390.132 Never86 (40.4%)375 (46.4%) Former81 (38.0%)275 (34.0%) Current25 (11.7%)77 (9.5%) Unknown21 (9.9%)82 (10.1%)SMD, Standardized mean difference; NYHA, International Society on Thrombosis and Hemostasis; LVEF, left ventricular ejection fraction. Baseline characteristics of patients stratified by residential distance from the surgical center before propensity score matching, mean ± SD or N (%) SMD, Standardized mean difference; NYHA, International Society on Thrombosis and Hemostasis; LVEF, left ventricular ejection fraction. Table E2Baseline characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD or N (%)Matched dataVariable<50 miles(n = 186)>50 miles(n = 372)P valueSMDAge (y)57.9 ± 10.6058.4 ± 10.28.590.047Gender, female58 (31.2%)123 (33.1%).720.04Diabetes37 (19.9%)74 (19.9%)1<0.001Hypertension122 (65.6%)247 (66.4%).920.017Renal failure (dialysis or creatinine >2 mg/dL)8 (4.3%)14 (3.8%).930.027History of cerebrovascular accident5 (2.7%)13 (3.5%).790.047NYHA class.740.097 I24 (12.9%)43 (11.6%) II82 (44.1%)169 (45.4%) III74 (39.8%)153 (41.1%) IV6 (3.2%)7 (1.9%)Preoperative LVEF (%)60.3 ± 10.060.2 ± 9.6.840.017AFib/aflutter.320.183 No170 (91.4%)324 (87.1%) Yes15 (8.1%)39 (10.5%) Yes, but timing unknown1 (0.5%)8 (2.2%)Smoking status.950.053 Never73 (39.2%)144 (38.7%) Former71 (38.2%)136 (36.6%) Current21 (11.3%)47 (12.6%) Unknown21 (11.3%)45 (12.1%)Year of surgery2011.0 ± 7.12010.7 ± 6.9.640.042LVEF, Left ventricular ejection fraction; SMD, standardized mean difference. Baseline characteristics of patients stratified by residential distance from the surgical center after propensity score matching, mean ± SD or N (%) LVEF, Left ventricular ejection fraction; SMD, standardized mean difference. Table E3Major bleeding, stroke, and transient ischemic attack International Classification of Diseases codesCodeDiagnosisICD-9Major bleeding events430Subarachnoid hemorrhage431Intracerebral hemorrhage432Other and unspecified intracranial hemorrhage432Nontraumatic extradural hemorrhage432.1Subdural hemorrhage432.9Unspecified intracranial hemorrhage456Esophageal varices with bleeding456.2Esophageal varices in diseases classified elsewhere, with bleeding459Hemorrhage, unspecified530.21Ulcer of esophagus with bleeding530.82Esophagitis with hemorrhage531Acute gastric ulcer with hemorrhage, unspecified as to obstruction531.01Acute gastric ulcer with hemorrhage, with obstruction531.2Acute gastric ulcer with hemorrhage and perforation, unspecified as to obstruction531.21Acute gastric ulcer with hemorrhage and perforation, with obstruction531.4Chronic gastric ulcer with hemorrhage531.4Chronic or unspecified gastric ulcer with hemorrhage, unspecified as to obstruction531.41Chronic or unspecified gastric ulcer with hemorrhage, with obstruction531.6Chronic or unspecified gastric ulcer with hemorrhage and perforation, unspecified as to obstruction531.61Chronic or unspecified gastric ulcer with hemorrhage and perforation, with obstruction532Acute duodenal ulcer with hemorrhage, unspecified as to obstruction532.01Acute duodenal ulcer with hemorrhage, with obstruction532.2Acute duodenal ulcer with hemorrhage and perforation, unspecified as to obstruction532.21Acute duodenal ulcer with hemorrhage and perforation, with obstruction532.4Chronic or unspecified duodenal ulcer with hemorrhage, unspecified as to obstruction532.41Chronic or unspecified duodenal ulcer with hemorrhage, with obstruction532.6Chronic or unspecified duodenal ulcer with hemorrhage and perforation, unspecified as to obstruction532.61Chronic or unspecified duodenal ulcer with hemorrhage and perforation, with obstruction533Acute peptic ulcer, site unspecified, with hemorrhage, unspecified as to obstruction533.01Acute peptic ulcer, site unspecified, with hemorrhage, with obstruction533.2Acute peptic ulcer, site unspecified, with hemorrhage and perforation, unspecified as to obstruction533.21Acute peptic ulcer, site unspecified, with hemorrhage and perforation, with obstruction533.4Chronic or unspecified peptic ulcer, site unspecified, with hemorrhage, unspecified as to obstruction533.41Chronic or unspecified peptic ulcer, site unspecified, with hemorrhage, with obstruction533.6Chronic or unspecified peptic ulcer, site unspecified, with hemorrhage and perforation, unspecified as to obstruction533.61Chronic or unspecified peptic ulcer, site unspecified, with hemorrhage and perforation, with obstruction534Acute gastrojejunal ulcer with hemorrhage, unspecified as to obstruction534.01Acute gastrojejunal ulcer with hemorrhage, with obstruction534.2Acute gastrojejunal ulcer with hemorrhage and perforation, unspecified as to obstruction534.21Acute gastrojejunal ulcer with hemorrhage and perforation, with obstruction534.4Chronic or unspecified gastrojejunal ulcer with hemorrhage, unspecified as to obstruction534.41Chronic or unspecified gastrojejunal ulcer with hemorrhage, with obstruction534.6Chronic or unspecified gastrojejunal ulcer with hemorrhage and perforation, unspecified as to obstruction534.61Chronic or unspecified gastrojejunal ulcer with hemorrhage and perforation, with obstruction535.01Acute gastritis with hemorrhage535.11Atrophic gastritis with hemorrhage535.21Gastric mucosal hypertrophy with hemorrhage535.31Alcoholic gastritis with hemorrhage535.41Other specified gastritis with hemorrhage535.51Unspecified gastritis with hemorrhage535.61Duodenitis with hemorrhage569.3Hemorrhage of rectum and anus578Hematemesis578.1Blood in stool (melena)578.9Gastrointestinal hemorrhage, unspecified537.84Dieulafoy lesion of stomach and duodenum562.02Diverticulosis of small intestine with hemorrhage562.03Diverticulitis of small intestine with hemorrhage562.12Diverticulosis of colon with hemorrhage562.13Diverticulitis of colon with hemorrhage784.8Hemorrhage from throat568.81Hemoperitoneum (nontraumatic)784.7Epistaxis786.3Hemoptysis, unspecified786.39Other hemoptysisICD-10Major bleeding eventsI60.9Nontraumatic subarachnoid hemorrhage, unspecifiedI61.9Intracerebral hemorrhage, unspecifiedI62.1Nontraumatic intracerebral hemorrhage, multiple localizedI62.00Nontraumatic intracerebral hemorrhage, unspecified hemisphereI62.9Nontraumatic intracranial hemorrhage, unspecifiedI85.01Esophageal varices with bleedingI85.11Secondary esophageal varices with bleedingR58Hemorrhage, not elsewhere classifiedK22.11Ulcer of esophagus with bleedingK22.8Other specified disease of esophagus with bleedingK25.0Acute gastric ulcer with hemorrhageK25.2Acute gastric ulcer with hemorrhage and perforationK25.4Chronic or unspecified gastric ulcer with hemorrhageK25.6Chronic or unspecified gastric ulcer with hemorrhage and perforationK26.0Acute duodenal ulcer with hemorrhageK26.2Acute duodenal ulcer with hemorrhage and perforationK26.4Chronic or unspecified duodenal ulcer with hemorrhageK26.6Chronic or unspecified duodenal ulcer with hemorrhage and perforationK27.0Acute peptic ulcer, site unspecified, with hemorrhageK27.2Acute peptic ulcer, site unspecified, with hemorrhage and perforationK27.4Chronic or unspecified peptic ulcer, site unspecified, with hemorrhageK27.6Chronic or unspecified peptic ulcer, site unspecified, with hemorrhage and perforationK28.0Acute gastrojejunal ulcer with hemorrhageK28.2Acute gastrojejunal ulcer with hemorrhage and perforationK28.4Chronic or unspecified gastrojejunal ulcer with hemorrhageK28.6Chronic or unspecified gastrojejunal ulcer with hemorrhage and perforationK29.01Acute gastritis with hemorrhageK29.41Atrophic gastritis with hemorrhageK29.51Alcoholic gastritis with hemorrhageK29.61Other specified gastritis with hemorrhageK29.21Chronic superficial gastritis with hemorrhageK29.71Gastritis, unspecified, with hemorrhageK29.91Other gastritis, unspecified with hemorrhageK29.81Duodenitis with hemorrhageK62.5Hemorrhage of anus and rectumK92.0HematemesisK92.1MelenaK31.82Dieulafoy lesion of stomach and duodenumK92.2Gastrointestinal hemorrhage, unspecifiedK57.11Diverticulosis of small intestine with hemorrhageK57.13Diverticulitis of small intestine with hemorrhageK57.31Diverticulosis of large intestine with hemorrhageK57.33Diverticulitis of large intestine with hemorrhageR04.1Hemorrhage from throatK66.1HemoperitoneumR04.0EpistaxisR04.2HemoptysisR04.9Hemorrhage from respiratory passages, unspecifiedR04.89Other specified hemorrhage from respiratory passagesICD-9Ischemic stroke/TIA433.01Occlusion and stenosis of basilar artery with infarction433.11Occlusion and stenosis of carotid artery with infarction433.21Occlusion and stenosis of vertebral artery with infarction433.31Occlusion and stenosis of multiple and bilateral precerebral arteries with infarction433.81Occlusion and stenosis of other specified precerebral artery with infarction433.91Occlusion and stenosis of unspecified precerebral artery with infarction434.01Cerebral thrombosis with infarction434.11Cerebral embolism with infarction434.91Cerebral artery occlusion, unspecified, with infarction435.0Basilar artery syndrome (TIA)435.1Vertebral artery syndrome (TIA)435.3Vertebro-basilar artery syndrome (TIA)435.8Other specified transient cerebral ischemia435.9Unspecified transient cerebral ischemiaICD-10Ischemic stroke/TIAI63.0Cerebral infarction due to thrombosis of precerebral arteriesI63.1Cerebral infarction due to embolism of precerebral arteriesI63.2Cerebral infarction due to unspecified occlusion or stenosis of precerebral arteriesI63.3Cerebral infarction due to thrombosis of cerebral arteriesI63.4Cerebral infarction due to embolism of cerebral arteriesI63.5Cerebral infarction due to unspecified occlusion or stenosis of cerebral arteriesI63.6Cerebral infarction due to cerebral venous thrombosis, nonpyogenicI63.8Other cerebral infarctionI63.9Cerebral infarction, unspecifiedG45.0Vertebro-basilar artery syndrome (TIA)G45.1Carotid artery syndrome (hemispheric TIA)G45.2Multiple and bilateral precerebral artery syndromes (TIA)G45.8Other specified transient cerebral ischemic attacks and related syndromesG45.9Transient cerebral ischemic attack, unspecifiedICD, International Classification of Diseases; TIA, transient ischemic attack. Major bleeding, stroke, and transient ischemic attack International Classification of Diseases codes ICD, International Classification of Diseases; TIA, transient ischemic attack. References