Readmissions After Surgical Aortic Valve Replacement: Influence of Prosthesis Type.

INTRODUCTION: Prosthesis choice during aortic valve replacement (AVR) weighs lifelong anticoagulation with mechanical valves (M-AVR) against structural valve degeneration in bioprosthetic valves (B-AVR). METHODS: The Nationwide Readmissions Database was queried to identify patients who underwent isolated surgical AVR between January 1, 2016 and December 31, 2018, stratifying by prothesis type. Propensity score matching was used to compare risk-adjusted outcomes. Readmission at 1 y was estimated with Kaplan-Meier (KM) analysis. RESULTS: Patients (n = 109,744) who underwent AVR (90,574 B-AVR and 19,170 M-AVR) were included. B-AVR patients were older (median 68 versus 57 y; P < 0.001) and had more comorbidities (mean Elixhauser score: 11.8 versus 10.7; P < 0.001) compared to M-AVR patients. After matching (n = 36,951), there was no difference in age (58 versus 57 y; P = 0.6) and Elixhauser score (11.0 versus 10.8; P = 0.3). B-AVR patients had similar in-hospital mortality (2.3% versus 2.3%; P = 0.9) and cost (mean: $50,958 versus $51,200; P = 0.4) compared with M-AVR patients. However, B-AVR patients had shorter length of stay (8.3 versus 8.7 d; P < 0.001) and fewer readmissions at 30 d (10.3% versus 12.6%; P < 0.001) and 90 d (14.8% versus 17.8%; P < 0.001), and 1 y (P < 0.001, KM analysis). Patients undergoing B-AVR were less likely to be readmitted for bleeding or coagulopathy (5.7% versus 9.9%; P < 0.001) and effusions (9.1% versus 11.9%; P < 0.001). CONCLUSIONS: B-AVR patients had similar early outcomes compared to M-AVR patients, but lower rates of readmission. Bleeding, coagulopathy, and effusions are drivers of excess readmissions in M-AVR patients. Readmission reduction strategies targeting bleeding and improved anticoagulation management are warranted in the first year following AVR.

Considerations for Left Atrial Appendage Occlusion During Cardiac Surgery.

Atrial fibrillation is associated with an increased risk of stroke secondary to thrombus formation in the left atrial appendage. Left atrial appendage occlusion (LAAO) is an effective method of reducing the risk of stroke in patients with atrial fibrillation. Although LAAO does not remove the requirement for anticoagulation, it reduces the risk of stroke when compared to anticoagulation alone. We critically analyze the data on LAAO in cardiac surgery. We also discuss the methods of LAAO, the risks of LAAO, and patient populations that could benefit from LAAO. We discuss high-level evidence that LAAO at the time of cardiac surgery reduces the risk of stroke in patients with a history of atrial fibrillation. In patients without a history of atrial fibrillation undergoing cardiac surgery, we suggest that LAAO should be considered in select patients at high risk of atrial fibrillation and stroke, when technically feasible.

Global Noncoding microRNA Profiling in Mice Infected with Partial Human Mouth Microbes (PAHMM) Using an Ecological Time-Sequential Polybacterial Periodontal Infection (ETSPPI) Model Reveals Sex-Specific Differential microRNA Expression.

Periodontitis (PD) is a polymicrobial dysbiotic immuno-inflammatory disease. It is more prevalent in males and has poorly understood pathogenic molecular mechanisms. Our primary objective was to characterize alterations in sex-specific microRNA (miRNA, miR) after periodontal bacterial infection. Using partial human mouth microbes (PAHMM) (Streptococcus gordonii, Fusobacterium nucleatum, Porphyromonas gingivalis, Treponema denticola, and Tannerella forsythia) in an ecological time-sequential polybacterial periodontal infection (ETSPPI) mouse model, we evaluated differential mandibular miRNA profiles by using high-throughput Nanostring nCounter® miRNA expression panels. All PAHMM mice showed bacterial colonization (100%) in the gingival surface, an increase in alveolar bone resorption (p < 0.0001), and the induction of a specific immunoglobin G antibody immune response (p < 0.001). Sex-specific differences in distal organ bacterial dissemination were observed in the heart (82% male vs. 28% female) and lungs (2% male vs. 68% female). Moreover, sex-specific differential expression (DE) of miRNA was identified in PAHMM mice. Out of 378 differentially expressed miRNAs, we identified seven miRNAs (miR-9, miR-148a, miR-669a, miR-199a-3p, miR-1274a, miR-377, and miR-690) in both sexes that may be implicated in the pathogenesis of periodontitis. A strong relationship was found between male-specific miR-377 upregulation and bacterial dissemination to the heart. This study demonstrates sex-specific differences in bacterial dissemination and in miRNA differential expression. A novel PAHMM mouse and ETSPPI model that replicates human pathobiology can be used to identify miRNA biomarkers in periodontitis.

A Wrist-Worn Sensor-Derived Frailty Index Based on an Upper-Extremity Functional Test in Predicting Functional Mobility in Older Adults.

INTRODUCTION: Preoperative frailty is an independent risk factor for postoperative complications across surgical specialties. Functional mobility such as gait, timed up and go (TUG), and 5 times sit-to-stand (5-STS) are popular preoperative frailty measurements but are not suitable for patients with severe mobility impairment. A wrist-worn sensor-derived frailty index based on an upper-extremity functional test (20-s repetitive elbow flexion-extension task; UEFI) was developed previously; however, its association with functional mobility remained unexplored. We aimed to investigate the predictive power of the UEFI in predicting functional mobility. METHODS: We examined correlation between the UEFI and gait speed, TUG duration, and 5-STS duration in 100 older adults (≥ 65 years) using multivariate regression analysis. The UEFI was calculated using slowness, weakness, exhaustion, and flexibility of the sensor-based 20-s repetitive elbow flexion-extension task. RESULTS: The UEFI was a significant predictor for gait speed and TUG duration and 5-STS duration (all R ≥ 0.60; all p < 0.001) with the variance (adjusted R2) of 35-37% for the dependent variables. The multivariate regression analysis revealed significant associations between the UEFI and gait speed (ß = -0.84; 95% confidence interval [95% CI] = [-1.19, -0.50]; p < 0.001) and TUG duration (ß = 16.2; 95% CI = [9.59, 22.8]; p < 0.001) and 5-STS duration (ß = 33.3; 95% CI = [23.6, 43.2]; p < 0.001), found after accounting for confounding variables (e.g., age and fear of falling scale). CONCLUSIONS: Our findings suggest that the UEFI can be performed with a wrist-worn sensor and has been validated with other established measures of preoperative frailty. The UEFI can be applied in a wide variety of patients, regardless of mobility limitations, in an outpatient setting.

Mid-Term Efficacy of Subxiphoid Versus Transpleural Pericardial Window for Pericardial Effusion.

BACKGROUND: The optimal surgical technique for drainage of pericardial effusions is frequently debated. Transpleural drainage via thoracotomy or thoracoscopy is hypothesized to provide more durable freedom from recurrent pericardial effusion than a subxiphoid pericardial window. We sought to compare operative outcomes and mid-term freedom from recurrent effusion between both approaches in patients with nontraumatic pericardial effusions. METHODS: All patients at our institution who underwent a pericardial window from 2001 to 2018 were identified. After excluding those who underwent recent cardiothoracic surgery or trauma, patients (n = 46) were stratified by surgical approach and presence of malignancy. Primary outcome was freedom from recurrent moderate or greater pericardial effusion. Secondary outcomes included operative mortality and morbidity and mid-term survival. Follow-up was determined by medical record review, with a follow-up of 67 patient-years. Fisher's exact test and Wilcoxon rank-sum test were used to compare groups. Mid-term survival and freedom from effusion recurrence were determined using Kaplan-Meier method. RESULTS: Subxiphoid windows (n = 31; 67%) were more frequently performed than transpleural windows (n = 15; 33%) and baseline characteristics were similar. Effusion etiologies included malignancy (n = 22; 48%), idiopathic (n = 12; 26%), uremia (n = 8; 17%), and collagen vascular disease (n = 4; 9%). Perioperative outcomes were comparable between the two surgical approaches, except for longer drain duration (7 versus 4 d, P = 0.029) in the subxiphoid group. Operative mortality was 19.6% overall and 36.4% in patients with malignancy. Mid-term survival and freedom from moderate or greater pericardial effusion recurrence was 37% (95% confidence interval [CI]: 19%-54%) and 69% (95% CI: 52%-86%) at 5 y, respectively. There was no difference in mid-term survival (P = 0.90) or freedom from pericardial effusion recurrence (P = 0.70) between surgical approaches. Although malignant etiology had worse late survival (P < 0.01), freedom from effusion recurrence was similar to nonmalignant etiology (P = 0.70). CONCLUSIONS: Pericardial window provides effective mid-term relief of pericardial effusion. Subxiphoid and transpleural windows are equivalent in mid-term efficacy and both surgical approaches can be considered. Patients with malignancy have acceptable operative mortality with low incidence of recurrent effusion, supporting palliative indications.

Improved outcomes and value in staged hybrid extent II thoracoabdominal aortic aneurysm repair.

OBJECTIVE: Complex Crawford extent II thoracoabdominal aortic aneurysms (TAAAs) can be treated in a hybrid manner with proximal thoracic endovascular aneurysm repair, followed by staged distal open thoracoabdominal repair. This study evaluated the outcomes and health care-associated value of this new method compared with traditional open repair over 10 years. METHODS: A prospectively collected database was used to identify all patients with an extent II TAAA undergoing repair at a single institution between 2005 and 2015. Patient characteristics, postoperative outcomes, and incidence of major adverse events (MAEs; renal failure, spinal cord ischemia, death) were compared. After adjusting for time since surgery, value was analyzed looking at quality (1/MAE) divided by cost (total health system cost). This was multiplied by a constant to set the value of open TAAA repair to 100. RESULTS: A total of 113 consecutive patients underwent extent II TAAA repairs, of whom 25 (22.1%) had a staged hybrid approach with a median of 129 days between procedures. No baseline differences in demographic or comorbidity variables existed between groups (P > .05). The hybrid group had shorter operative time (255 vs 306 minutes; P = .01), shorter postoperative length of stay (10.1 vs 13.3 days; P = .02), and reduced blood loss (1300 vs 2600 mL; P = .01) at the time of open operation. Despite higher rates of acute kidney injury in the hybrid group (76.0% vs 51.1%; P = .03), there was no difference in renal failure (8.0% vs 4.5%; P = .84). The incidence of MAEs was lower in the staged hybrid group (20.0% vs 48.9%; P = .01), without a difference in hospital mortality (4.0% vs 3.4%; P = .89). Median total cost was higher in the hybrid group ($112,920 vs $72,037; P = .003). Value was improved in the hybrid group by 56% using mean cost and 178% by median cost. CONCLUSIONS: The 20% MAE rate associated with staged hybrid repair of extent II TAAA was significantly decreased compared with open repair, with a relative reduction of >50%. Despite higher total hospital costs, staged hybrid repair had 56% to 178% higher health care-related value compared with standard open repair. In an era of increasing focus on costs and quality, staged hybrid repair of extensive TAAAs is associated with fewer complications than open TAAA repair, resulting in a good value investment from a resource utilization perspective.

Pre-implant left ventricular apex position predicts risk of HeartMate II pump thrombosis.

BACKGROUND: Thrombosis within a left ventricular assist device (LVAD) is a devastating complication that often necessitates device exchange. Few studies have evaluated the relationship between patient anatomy and pump thrombosis. We hypothesize that lateral displacement of the left ventricular (LV) apex increases risk for pump thrombosis. METHODS: All patients who underwent primary implantation of a HeartMate II (HM2) device (Thoratec, Pleasanton, CA) at a single center (2009-2015) were evaluated. Operative mortalities and patients without imaging were excluded. The angle of the LV apex relative to the midline was measured on preoperative computed tomography scans by two independent surgeons. Pump thrombosis was defined as lactic dehydrogenase >700 with clinical symptoms of hemolysis or LVAD malfunction. Univariate and Cox Proportional Hazards analysis was performed to evaluate the impact of LV apex angle on long-term freedom from pump exchange for thrombosis. RESULTS: Of 122 patients who met inclusion criteria for this study, 16 (13.1%) underwent exchange for presumed pump thrombosis. Of these patients, six (37.5%) required more than one exchange. Patients undergoing exchange for thrombosis had greater LV angle (43.8 ± 9.7 vs 49.5 ± 11.2, p = 0.037) with LV apex angle being a significant predictor of LVAD exchange for thrombosis (hazard ratio = 1.047, P = 0.046). Additionally, when surgeon measurements were compared there was good inter-observer reliability (Pearson Correlation = 0.89). CONCLUSION: A laterally displaced left ventricular apex correlates with a higher risk of pump thrombosis in patients undergoing HM2 implantation. LV apex angle is an easily obtained, reproducible measurement that should be considered when selecting a ventricular assist device.

Staged hybrid repair of extensive thoracoabdominal aortic aneurysms secondary to chronic aortic dissection.

OBJECTIVE: Many patients with aortic dissection develop Crawford extent I or II thoracoabdominal aortic aneurysms (TAAA). Because open repair is associated with a high morbidity and mortality, hybrid approaches to TAAA repair are emerging. In this study, we evaluated the midterm outcomes and aortic remodeling of a hybrid technique that combines proximal thoracic endovascular aneurysm repair (TEVAR), followed by staged distal open thoracoabdominal repair for patients with Crawford extent I or II TAAAs secondary to chronic aortic dissection. METHODS: We identified 19 patients with Crawford extent I (n = 1) or extent II (n = 18) TAAAs secondary to chronic aortic dissection who underwent a staged hybrid repair from 2007 to 2014 at our institution. Nine patients had previous open ascending aortic surgery for type I aortic dissection. Stage 1 TEVAR was performed via percutaneous (n = 8), femoral cutdown (n = 8), or iliac exposure (n = 3). The left subclavian artery was covered in nine patients and revascularized in eight patients using carotid-subclavian bypass (n = 7) or laser fenestration (n = 1). Stage 2 open repair was performed a median of 18 weeks later with partial cardiopulmonary bypass via left femoral arterial and venous cannulation for visceral and lower body perfusion. The open thoracoabdominal graft was anastomosed proximally in an end to end fashion with the endograft. We then assessed surgical morbidity and mortality, midterm survival, and freedom from reintervention. Aortic remodeling was measured and change in maximum aortic and false lumen diameter at last follow-up (median, 3 years) from baseline was assessed. RESULTS: There were no deaths, strokes, or chronic renal failure in this cohort. After stage 1 TEVAR, three patients required repeat intervention for endoleak (type Ia, n = 1; type Ib, n = 1; type II, n = 1) before open repair. After stage 2 open repair, there was a single delayed permanent paralysis 2 weeks after discharge. At a median 3-year follow-up (range, 6 months-6.2 years), there were no deaths, neurologic events, endoleaks, or TAAA reinterventions. Complete false lumen thrombosis occurred in 100% of the patients, with maximum false lumen diameter decreasing from 34.3 ± 15.3 mm to 13.2 ± 12.0 mm (P < .01) and total aortic diameter decreasing from 60.2 ± 9.0 mm to 49.4 ± 9.6 mm (P < .01). CONCLUSIONS: Staged hybrid TAAA repair, using a combination of proximal TEVAR with open distal repair, can be performed using established endovascular skills and technology coupled with traditional open aortic surgical techniques, with low surgical morbidity and mortality. In the midterm, staged hybrid TAAA repair was associated favorable survival, aortic remodeling, and freedom from reintervention.

Premature Bioprosthetic Aortic Valve Degeneration Associated with Allergy to Galactose-Alpha-1,3-Galactose.

We present the cases of two patients with bioprosthetic aortic valves who developed an allergy to alpha-gal. Each had premature degeneration of their bioprosthesis and demonstrated rapidly increasing transvalvular gradients after development of their allergy. Each underwent successful replacement with a mechanical aortic valve within 1-2 years of symptom onset. doi: 10.1111/jocs.12764 (J Card Surg 2016;31:446-448).

Machine learning in cardiac surgery: a narrative review.

Background and Objective: Machine learning (ML) is increasingly being utilized to provide data driven solutions to challenges in medicine. Within the field of cardiac surgery, ML methods have been employed as risk stratification tools to predict a variety of operative outcomes. However, the clinical utility of ML in this domain is unclear. The aim of this review is to provide an overview of ML in cardiac surgery, particularly with regards to its utility in predictive analytics and implications for use in clinical decision support. Methods: We performed a narrative review of relevant articles indexed in PubMed since 2000 using the MeSH terms "Machine Learning", "Supervised Machine Learning", "Deep Learning", or "Artificial Intelligence" and "Cardiovascular Surgery" or "Thoracic Surgery". Key Content and Findings: ML methods have been widely used to generate pre-operative risk profiles, consistently resulting in the accurate prediction of clinical outcomes in cardiac surgery. However, improvement in predictive performance over traditional risk metrics has proven modest and current applications in the clinical setting remain limited. Conclusions: Studies utilizing high volume, multidimensional data such as that derived from electronic health record (EHR) data appear to best demonstrate the advantages of ML methods. Models trained on post cardiac surgery intensive care unit data demonstrate excellent predictive performance and may provide greater clinical utility if incorporated as clinical decision support tools. Further development of ML models and their integration into EHR's may result in dynamic clinical decision support strategies capable of informing clinical care and improving outcomes in cardiac surgery.

Outcomes after bioprosthetic versus mechanical mitral valve replacement for infective endocarditis in the United States.

Objective: In patients who underwent mitral valve replacement for infectious endocarditis, we evaluated the association of prosthesis choice with readmission rates and causes (the primary outcomes), as well as with in-hospital mortality, cost, and length of stay (the secondary outcomes). Methods: Patients with infectious endocarditis who underwent isolated mitral valve replacement from January 2016 to December 2018 were identified in the United States Nationwide Readmissions Database and stratified by valve type. Propensity score matching was used to compare adjusted outcomes. Results: A weighted total of 4206 patients with infectious endocarditis underwent bioprosthetic mitral valve replacement (n = 3132) and mechanical mitral valve replacement (n = 1074) during the study period. Patients in the bioprosthetic mitral valve replacement group were older than those in the mechanical mitral valve replacement group (median 57 vs 46 y, P < .001). After propensity matching, the bioprosthetic mitral valve replacement group (n = 1068) had similar in-hospital mortality, length of stay, and costs compared with the mechanical mitral valve replacement group (n = 1056). Overall, 90-day readmission rates were high (28.9%) and comparable for bioprosthetic mitral valve replacement (30.5%) and mechanical mitral valve replacement (27.5%, P = .4). Likewise, there was no difference in readmissions over a calendar year by prosthesis type. Readmissions for infection and bleeding were common for both bioprosthetic mitral valve replacement and mechanical mitral valve replacement groups. Conclusions: Outcomes and readmission rates were similar for mechanical mitral valve replacement and bioprosthetic mitral valve replacement in infectious endocarditis, suggesting that valve choice should not be determined by endocarditis status. Additionally, strategies to mitigate readmission for infection and bleeding are needed for both groups.

Development of a Sensor Technology to Objectively Measure Dexterity for Cardiac Surgical Proficiency.

BACKGROUND: Technical skill is essential for good outcomes in cardiac surgery. However, no objective methods exist to measure dexterity while performing surgery. The purpose of this study was to validate sensor-based hand motion analysis (HMA) of technical dexterity while performing a graft anastomosis within a validated simulator. METHODS: Surgeons at various training levels performed an anastomosis while wearing flexible sensors (BioStamp nPoint, MC10 Inc) with integrated accelerometers and gyroscopes on each hand to quantify HMA kinematics. Groups were stratified as experts (n = 8) or novices (n = 18). The quality of the completed anastomosis was scored using the 10 Point Microsurgical Anastomosis Rating Scale (MARS10). HMA parameters were compared between groups and correlated with quality. Logistic regression was used to develop a predictive model from HMA parameters to distinguish experts from novices. RESULTS: Experts were faster (11 ± 6 minutes vs 21 ± 9 minutes; P = .012) and used fewer movements in both dominant (340 ± 166 moves vs 699 ± 284 moves; P = .003) and nondominant (359 ± 188 moves vs 567 ± 201 moves; P = .02) hands compared with novices. Experts' anastomoses were of higher quality compared with novices (9.0 ± 1.2 MARS10 vs 4.9 ± 3.2 MARS10; P = .002). Higher anastomosis quality correlated with 9 of 10 HMA parameters, including fewer and shorter movements of both hands (dominant, r = -0.65, r = -0.46; nondominant, r = -0.58, r = -0.39, respectively). CONCLUSIONS: Sensor-based HMA can distinguish technical dexterity differences between experts and novices, and correlates with quality. Objective quantification of hand dexterity may be a valuable adjunct to training and education in cardiac surgery training programs.

Safety of Early Discharge After Coronary Artery Bypass Grafting: A Nationwide Readmissions Analysis.

BACKGROUND: We determined the safety of early discharge after coronary artery bypass grafting (CABG) in patients with uncomplicated postoperative courses and compared outcomes to routine discharge in a national cohort. We identified preoperative factors associated with readmission following early discharge after CABG. METHODS: The Nationwide Readmissions Database was queried to identify patients undergoing CABG from 01/2016-12/2018. Patients were stratified based on length of stay (LOS) as early (≤4 days) versus routine (5-10 days) discharge. Patients were excluded with hospital courses indicative of complicated stays (emergent procedures, LOS>10 days, discharge to extended care facility or with home health, index-hospitalization mortality). Propensity-score matching was performed to compare outcomes between cohorts. Multivariable logistic regression models were used to identify factors associated with readmission following early discharge. RESULTS: A total of 91,861 patients underwent CABG with an uncomplicated postoperative course during the study period (≈20% of CABG population). Of these 31% (28,790/91,861) were discharged early and 69% (63,071/91,861) routine. After propensity-score matching, patients discharged early had lower readmission rates at 30-days, 90-days, and up to one year (P<.001, all). Index-hospitalization cost was lower with early discharge ($26,676 versus $32,859; P<.001). Early discharge was associated with a lower incidence of nosocomial infection at index-hospitalization (0.17% versus 0.81%, P<.001) and readmission from infection (14.5% versus 18%, P=.016). CONCLUSIONS: Early discharge after uncomplicated CABG can be considered in a highly selective patient population. Early discharge patients are readmitted less frequently than matched routine discharge patients, with a lower incidence of readmission from infection. Appropriate post-discharge processes to facilitate early discharge after CABG should be further pursued.

Coronary artery bypass grafting after acute ST-elevation myocardial infarction.

OBJECTIVES: The study objectives were to describe the trends and outcomes of isolated coronary artery bypass grafting after ST-elevation myocardial infarction using a nationwide database. METHODS: We queried the 2002-2016 National Inpatient Sample database for hospitalized patients with ST-elevation myocardial infarction who underwent isolated coronary artery bypass grafting. We report temporal trends, predictors, and outcomes of coronary artery bypass grafting in the early (2002-2010) and recent (2011-2016) cohorts. RESULTS: Of 3,347,470 patients hospitalized for ST-elevation myocardial infarction, 7.7% underwent isolated coronary artery bypass grafting. The incidence of isolated coronary artery bypass grafting after ST-elevation myocardial infarction decreased over time (9.2% in 2002 vs 5.5% in 2016, Ptrend < .001), whereas perioperative crude in-hospital mortality did not change (5.1% in 2002 vs 4.2% in 2016, Ptrend = .66), coinciding with an increase in the burden of comorbidities. There was an increase in performing isolated coronary artery bypass grafting on hospitalization day 3 or more, as well as an increase in the use of mechanical support devices and precoronary artery bypass grafting percutaneous coronary intervention. In the early cohort, isolated coronary artery bypass grafting on days 1 and 2 was associated with higher in-hospital mortality. In the recent cohort, coronary artery bypass grafting on day 2 had similar in-hospital mortality compared with day 3 or more and lower rates of acute kidney injury, ischemic stroke, ventricular arrhythmia, and length of hospital stay. CONCLUSIONS: In this nationwide analysis, there has been a decline in the use of isolated coronary artery bypass grafting after ST-elevation myocardial infarction. Isolated coronary artery bypass grafting on day 1 was performed in sicker patients and was associated with higher in-hospital mortality than coronary artery bypass grafting performed on day 3 or more. In the recent cohort, isolated coronary artery bypass grafting on day 2 had similar in-hospital mortality compared with day 3 or more.

Machine learning for dynamic and early prediction of acute kidney injury after cardiac surgery.

OBJECTIVE: Acute kidney injury after cardiac surgery increases morbidity and mortality. Diagnosis relies on oliguria or increased serum creatinine, which develop 48 to 72 hours after injury. We hypothesized machine learning incorporating preoperative, operative, and intensive care unit data could dynamically predict acute kidney injury before conventional identification. METHODS: Cardiac surgery patients at a tertiary hospital (2008-2019) were identified using electronic medical records in the Medical Information Mart for Intensive Care IV database. Preoperative and intraoperative parameters included demographics, Charlson Comorbidity subcategories, and operative details. Intensive care unit data included hemodynamics, medications, fluid intake/output, and laboratory results. Kidney Disease: Improving Global Outcomes creatinine criteria were used for acute kidney injury diagnosis. An ensemble machine learning model was trained for hourly predictions of future acute kidney injury within 48 hours. Performance was evaluated by area under the receiver operating characteristic curve and balanced accuracy. RESULTS: Within the cohort (n = 4267), there were approximately 7 million data points. Median baseline creatinine was 1.0 g/dL (interquartile range, 0.8-1.2), with 17% (735/4267) of patients having chronic kidney disease. Postoperative stage 1 acute kidney injury occurred in 50% (2129/4267), stage 2 occurred in 8% (324/4267), and stage 3 occurred in 4% (183/4267). For hourly prediction of any acute kidney injury over the next 48 hours, area under the receiver operating characteristic curve was 0.82, and balanced accuracy was 75%. For hourly prediction of stage 2 or greater acute kidney injury over the next 48 hours, area under the receiver operating characteristic curve was 0.95 and balanced accuracy was 86%. The model predicted acute kidney injury before clinical detection in 89% of cases. CONCLUSIONS: Ensemble machine learning models using electronic medical records data can dynamically predict acute kidney injury risk after cardiac surgery. Continuous postoperative risk assessment could facilitate interventions to limit or prevent renal injury.

Racial/ethnic differences persist in treatment choice and outcomes in isolated intervention for coronary artery disease.

OBJECTIVE: Studies have noted racial/ethnic disparities in coronary artery disease intervention strategies. We investigated trends and outcomes of coronary artery disease treatment choice (coronary artery bypass grafting or percutaneous coronary intervention) stratified by race/ethnicity. METHODS: We queried the National Inpatient Sample for patients who underwent isolated coronary artery bypass grafting or percutaneous coronary intervention (2002-2017). Outcomes were stratified by race/ethnicity (White, African American, Hispanic, Asian). Multivariable logistic regression evaluated associations between race/ethnicity and receiving coronary artery bypass grafting versus percutaneous coronary intervention, in-hospital mortality, and costs. RESULTS: Over the 15-year period, 2,426,917 isolated coronary artery bypass grafting surgeries and 7,184,515 percutaneous coronary interventions were performed. Compared with White patients, African American patients were younger (62 [interquartile range, 53-70] vs 66 [interquartile range, 57-75] years), were more likely to have Medicaid insurance (12.2% vs 4.4%), and had more comorbidities (Charlson-Deyo index, 1.9 ± 1.6 vs 1.7 ± 1.6) (all P < .01). After adjustment for patient comorbidities, presence of acute myocardial infarction, insurance status, and geography, African Americans were the least likely of all racial/ethnic groups to undergo coronary artery bypass grafting (odds ratio, 0.76; P < .01), a consistent trend throughout the study. African American patients had higher risk-adjusted mortality after coronary artery bypass grafting (odds ratio, 1.09; P < .01). Race/ethnicity was not associated with increased mortality after percutaneous coronary intervention. African American patients had higher hospitalization costs for coronary artery bypass grafting (+$5816; P < .01) and percutaneous coronary intervention (+$856; P < .01) after controlling for confounders. CONCLUSIONS: In this contemporary national analysis, risk-adjusted frequency of coronary artery bypass grafting versus percutaneous coronary intervention for coronary artery disease differed by race/ethnicity. African American patients had lower odds of undergoing coronary artery bypass grafting and worse outcomes. Reasons for these differences merit further investigation to identify opportunities to reduce potential disparities.

Development of a Machine Learning Model to Predict Outcomes and Cost After Cardiac Surgery.

BACKGROUND: Machine learning (ML) algorithms may enhance outcomes prediction and help guide clinical decision making. This study aimed to develop and validate a ML model that predicts postoperative outcomes and costs after cardiac surgery. METHODS: The Society of Thoracic Surgeons registry data from 4874 patients who underwent cardiac surgery (56% coronary artery bypass grafting, 42% valve surgery, 19% aortic surgery) at our institution were divided into training (80%) and testing (20%) datasets. The Extreme Gradient Boosting decision-tree ML algorithms were trained to predict three outcomes: operative mortality, major morbidity or mortality, and Medicare outlier high hospitalization cost. Algorithm performance was determined using accuracy, F1 score, and area under the precision-recall curve (AUC-PR). The ML algorithms were validated in index surgery cases with The Society of Thoracic Surgeons risk scores for mortality and major morbidities and with logistic regression and were then applied to nonindex cases. RESULTS: The ML algorithms with 25 input parameters predicted operative mortality (accuracy 95%; F1 0.31; AUC-PR 0.21), major morbidity or mortality (accuracy 71%, F1 0.47; AUC-PR 0.47), and high cost (accuracy 84%; F1 0.62; AUC-PR 0.65). Preoperative creatinine, complete blood count, patient height and weight, ventricular function, and liver dysfunction were important predictors for all outcomes. For patients undergoing nonindex cardiac operations, the ML model achieved an AUC-PR of 0.15 (95% CI, 0.05-0.32) for mortality and 0.59 (95% CI, 0.51-0.68) for major morbidity or mortality. CONCLUSIONS: The extreme gradient boosting ML algorithms can predict mortality, major morbidity, and high cost after cardiac surgery, including operations without established risk models. These ML algorithms may refine risk prediction after cardiac surgery for a wide range of procedures.

Coronary artery bypass grafting at safety-net versus non-safety-net hospitals.

Objectives: Safety-net hospitals (SNHs) provide essential services to predominantly underserved patients regardless of their ability to pay. We hypothesized that patients who underwent coronary artery bypass grafting (CABG) would have inferior observed outcomes at SNHs compared with non-SNHs but that matched cohorts would have comparable outcomes. Methods: We queried the Nationwide Readmissions Database for patients who underwent isolated CABG from 2016 to 2018. We ranked hospitals by the percentage of all admissions in which the patient was uninsured or insured with Medicaid; hospitals in the top quartile were designated as SNHs. We used propensity-score matching to mitigate the effect of confounding factors and compare outcomes between SNHs and non-SNHs. Results: A total of 525,179 patients underwent CABG, including 96,133 (18.3%) at SNHs, who had a greater burden of baseline comorbidities (median Elixhauser score 8 vs 7; P = .04) and more frequently required urgent surgery (57.1% vs 52.8%; P < .001). Observed in-hospital mortality (2.1% vs 1.8%; P = .004) and major morbidity, length of stay (9 vs 8 days; P < .001), cost ($46,999 vs $38,417; P < .001), and readmission rate at 30 (12.4% vs 11.3%) and 90 days (19.0% vs 17.7%) were greater at SNHs (both P < .001). After matching, none of these differences persisted except length of stay (9 vs 8 days) and cost ($46,977 vs $39,343) (both P < .001). Conclusions: After matching, early outcomes after CABG were comparable at SNHs and non-SNHs. Improved discharge resources could reduce length of stay and curtail cost, improving the value of CABG at SNHs.

Socioeconomic disparities in procedural choice and outcomes after aortic valve replacement.

Objective: To identify potential socioeconomic disparities in the procedural choice of patients undergoing surgical aortic valve replacement (SAVR) versus transcatheter aortic valve replacement (TAVR) and in readmission outcomes after SAVR or TAVR. Methods: The Nationwide Readmissions Database was queried to identify a total of 243,691 patients who underwent isolated SAVR and TAVR between January 2016 and December 2018. Patients were stratified according to a tiered socioeconomic status (SES) metric comprising patient factors including education, literacy, housing, employment, insurance status, and neighborhood median income. Multivariable analyses were used to assess the effect of SES on procedural choice and risk-adjusted readmission outcomes. Results: SAVR (41.4%; 100,833 of 243,619) was performed less frequently than TAVR (58.6%; 142,786 of 243,619). Lower SES was more frequent among patients undergoing SAVR (20.2% [20,379 of 100,833] vs 19.4% [27,791 of 142,786]; P < .001). Along with such variables as small hospital size, drug abuse, arrhythmia, and obesity, lower SES was independently associated with SAVR relative to TAVR (adjusted odds ratio [aOR], 1.17; 95% confidence interval [CI], 1.11 to 1.24). After SAVR, but not after TAVR, lower SES was independently associated with increased readmission at 30 days (aOR, 1.19; 95% CI, 1.07-1.32), 90 days (aOR, 1.27; 95% CI, 1.15-1.41), and 1 year (adjusted hazard ratio, 1.19; 95% CI, 1.11 to 1.28; P < .05 for all). Conclusions: Our study findings indicate that socioeconomic disparities exist in the procedural choice for patients undergoing AVR. Patients with lower SES had increased odds of undergoing SAVR, as well as increased odds of readmission after SAVR, but not after TAVR, supporting that health inequities exist in the surgical care of socioeconomically disadvantaged patients.