Using machine learning to predict bleeding after cardiac surgery.

OBJECTIVES: The primary objective was to predict bleeding after cardiac surgery with machine learning using the data from the Australia New Zealand Society of Cardiac and Thoracic Surgeons Cardiac Surgery Database, cardiopulmonary bypass perfusion database, intensive care unit database and laboratory results. METHODS: We obtained surgical, perfusion, intensive care unit and laboratory data from a single Australian tertiary cardiac surgical hospital from February 2015 to March 2022 and included 2000 patients undergoing cardiac surgery. We trained our models to predict either the Papworth definition or Dyke et al.'s universal definition of perioperative bleeding. Our primary outcome was the performance of our machine learning algorithms using sensitivity, specificity, positive and negative predictive values, accuracy, area under receiver operating characteristics curve (AUROC) and area under precision-recall curve (AUPRC). RESULTS: Of the 2000 patients undergoing cardiac surgery, 13.3% (226/2000) had bleeding using the Papworth definition and 17.2% (343/2000) had moderate to massive bleeding using Dyke et al.'s definition. The best-performing model based on AUPRC was the Ensemble Voting Classifier model for both Papworth (AUPRC 0.310, AUROC 0.738) and Dyke definitions of bleeding (AUPRC 0.452, AUROC 0.797). CONCLUSIONS: Machine learning can incorporate routinely collected data from various datasets to predict bleeding after cardiac surgery.

The association between intraoperative cardiopulmonary bypass power and complications after cardiac surgery.

BACKGROUND: Low cardiac power (product of flow and pressure) has been shown to be associated with mortality in patients with cardiogenic shock after acute myocardial infarction, but has not been studied in cardiac surgical patients. This study's hypothesis was that cardiac power during cardiopulmonary bypass for cardiac surgery would have a greater association with adverse events than either flow or MAP (mean arterial pressure) alone. METHODS: We undertook a retrospective observational study using patient data from February 2015 to March 2022 undergoing cardiac surgery at Fiona Stanley Hospital in Perth Australia. Excluded were patient age less than 18 years old, patients undergoing thoracic transplantation, ventricular assist devices, off pump cardiac surgery and aortic surgery. The primary outcome was a composite outcome of 30-days mortality, stroke or new-onset renal insufficiency. RESULTS: Overall, 1984 cardiac surgeries were included in the analysis. Neither duration nor area below thresholds tested for power, MAP or flow was associated with the primary composite outcome. However, we found that an area below MAP thresholds 35-50 mmHg was associated with new renal insufficiency (adjusted odds ratio 1.17 [95% CI 1.02 to 1.35] for patients spending 10 min at 10 mmHg below 50 mmHg MAP compared to those who did not). CONCLUSIONS: This study suggests that MAP during cardiopulmonary bypass, but not power or flow, was an independent risk factor for adverse renal outcomes for cardiac surgical patients.

COVID-19 Serosurvey of Frontline Healthcare Workers in Western Australia.

We aimed to study COVID-19 infection in healthcare workers (HCWs) during the first wave in a setting of low community incidence prior to HCW vaccination. We performed a cross-sectional study of frontline HCWs in two tertiary hospitals in Western Australia with questionnaire and testing for SARS-CoV-2 IgG antibodies, using a screening assay followed by confirmatory assays for initial reactive results. 799 Frontline HCWs were enrolled in the study, working in the emergency department (n = 194, 24.2%), ICU (n = 176, 22.0%), respiratory ward (n = 20, 2.5%), COVID clinic (n = 37, 4.6%), and theatre (n = 222, 28%). 189 (23.6%) were doctors, 327 (41.0%) nurses, and 283 (35.4%) other. Contact with a known COVID-19-positive patient occurred at work for 337 (42.1%), and outside work for 10 (1.2%). Four were diagnosed with COVID-19 by PCR, acquired overseas in two cases and related to healthcare work in two cases (one acquired from a colleague and one possibly acquired from patient contact in the healthcare setting). Nine HCWs had reactive screening serology, and three had confirmed positive IgG (these three were PCR-positive cases). Infection control procedures in the setting of low community incidence were effective at preventing HCW acquisition of COVID-19 infection.

Endothelial biomarkers of thrombosis in critically ill patients with acquired coagulopathy: a prospective cohort study.

Background: It remains largely unknown whether endothelial biomarkers can be used to predict thromboembolism in critically ill patients with an acquired coagulopathy. Materials & methods: The ability of syndecan-1, P-selectin and microRNA expressions to predict thromboembolism were assessed in 40 critically ill coagulopathic patients. Results: The area under the receiver operating characteristic curves for syndecan-1, P-selectin and the four differentially expressed microRNAs - determined by real-time qPCR analysis - to predict thromboembolism (n = 10, 25%) were 0.71 (95% CI: 0.52-0.89), 0.73 (95% CI: 0.52-0.95) and >0.80, respectively. Syndecan-1 and P-selectin concentrations were also significantly correlated with one another (R = 0.41, p = 0.008). Conclusion: Endothelial function biomarkers, including endothelial microRNA expressions, were associated with subsequent thromboembolism in critically ill patients who had a deranged coagulation profile.

The Influence of Choice of Surgical Procedure on Long-Term Survival After Cardiac Surgery.

BACKGROUND: There is some interest in long-term survival after various cardiac surgical strategies, including off-pump versus on-pump coronary artery surgery (CAG), mitral valve (MV) repair versus replacement, and aortic valve (AV) bioprosthetic versus mechanical replacement. METHODS: We studied patients older than 49 years of age, recording risk factors and surgical details at the time of surgery. We classified procedures as: MV surgery with or without concurrent grafts or valves; AV surgery with or without concurrent CAG; or isolated CAG. Follow-up was through the state death register and state-wide hospital attendance records. Risk-adjusted survival was estimated using Cox proportional hazards. Observed survival was compared to the expected age- and sex- matched population survival. RESULTS: During a median follow-up of 14.8 years 5,807 of 11,718 patients died. The difference between observed and expected survival varied between 3.4 years for AV surgery and 9.6 years for females undergoing MV surgery. The risk-adjusted mortality hazard rate after off-pump CAG was 0.93 (95% CI 0.8-1.0, p=0.84), MV repair 0.67 (95% CI 0.6-0.8, p<0.0001), MV bioprosthesis 0.82 (95% CI 0.81 (0.6-1.0, p=0.11) and bioprosthetic AV replacement 1.02 (95% CI 0.9-1.2, p=0.82). CONCLUSIONS: Compared to the general population, cardiac surgical patients have a shorter than expected life expectancy. We observed a survival benefit of mitral valve repair over replacement. We did not observe significant survival differences between off-pump and on-pump CAG, nor between bioprosthetic and mechanical replacement.

A randomised controlled trial of succinylated gelatin (4%) fluid on urinary acute kidney injury biomarkers in cardiac surgical patients.

BACKGROUND: Fluid resuscitation is frequently required for cardiac surgical patients admitted to the intensive care unit. The ideal fluid of choice in regard to efficacy and safety remains uncertain. Compared with crystalloid fluid, colloid fluid may result in less positive fluid balance. However, some synthetic colloids are associated with increased risk of acute kidney injury (AKI). This study compared the effects of succinylated gelatin (4%) (GEL) with compound sodium lactate (CSL) on urinary AKI biomarkers in patients after cardiac surgery. METHODS: Cardiac surgical patients who required an intravenous fluid bolus of at least 500 mL postoperatively were randomly allocated to receive GEL or CSL as the resuscitation fluid of choice for the subsequent 24 h. Primary outcomes were serial urinary neutrophil gelatinase-associated lipocalin (NGAL) and cystatin C concentrations measured at baseline, 1 h, 5 h and 24 h after enrolment, with higher concentrations indicating greater kidney injury. Secondary biomarker outcomes included urinary clusterin, α1-microglobulin and F2-isoprostanes concentrations. Differences in change of biomarker concentration between the two groups over time were compared with mixed-effects regression models. Statistical significance was set at P < 0.05. RESULTS: Forty cardiac surgical patients (n = 20 per group) with similar baseline characteristics were included. There was no significant difference in the median volume of fluid boluses administered over 24 h between the GEL (1250 mL, Q1-Q3 500-1750) and CSL group (1000 mL, Q1-Q3 500-1375) (P = 0.42). There was a significantly greater increase in urinary cystatin C (P < 0.001), clusterin (P < 0.001), α1-microglobulin (P < 0.001) and F2-isoprostanes (P = 0.020) concentrations over time in the GEL group, compared to the CSL group. Change in urinary NGAL concentration (P = 0.68) over time was not significantly different between the groups. The results were not modified by adjustment for either urinary osmolality or EuroSCORE II predicted risk of mortality. CONCLUSIONS: This preliminary randomised controlled trial showed that use of succinylated gelatin (4%) for fluid resuscitation after cardiac surgery was associated with increased biomarker concentrations of renal tubular injury and dysfunction, compared to crystalloid fluid. These results generate concern that use of intravenous gelatin fluid may contribute to clinically relevant postoperative AKI. Trial registration ANZCTR.org.au, ACTRN12617001461381. Registered on 16th October, 2017, http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=373619&isReview=true .

Preoperative Intra-Aortic Balloon Pumps in Cardiac Surgery: A Propensity Score Analysis.

INTRODUCTION: The role of intra-aortic balloon pumps (IABP) in high-risk patients undergoing coronary artery bypass graft (CABG) surgery remains controversial. We report the 5-year experience from a new Australian centre. METHODS: We retrospectively analysed 690 patients undergoing urgent isolated CABG surgery at a Western Australian tertiary centre from February 2015 to May 2020. De-identified data was obtained from the Australia & New Zealand Society of Cardiothoracic Surgeons database. Patients were stratified according to preoperative IABP use. A propensity score was created for the probability of IABP use and a propensity adjusted analysis was performed using logistic regression. The primary outcome was 30-day mortality. Secondary outcomes were postoperative inhospital outcomes. RESULTS: Preoperative IABP was used in 78 patients (11.3%). After propensity score adjustment, in a subgroup of patients with reduced ejection fraction or left main disease, 30-day mortality (7.0% vs 2.0%, OR 6.03, 95% CI 1.89-19.28, p=0.002) was significantly higher in the IABP group. Red blood cell transfusions (19.7% vs 12.6%, OR 1.86, 95% CI 1.02-3.35, p=0.039), prolonged inotrope use (78.9% vs 50.9%, OR 6.11, 95% CI 2.77-13.48, p<0.001), prolonged invasive ventilation (28.2% vs 3.4%, OR 20.2, 95% CI 8.24-49.74, p<0.001), mesenteric ischaemia (2.8% vs 0%, OR 4.52, 95% CI 1.15-17.77, p=0.031) and multisystem organ failure (1.3% vs 0.7%, OR 25.68, 95% CI 2.55-258.34, p=0.006) were significantly higher in the IABP group. CONCLUSION: In patients undergoing isolated CABG surgery, preoperative IABP use was associated with increased 30-day mortality and adverse outcomes. Large randomised controlled trials are required to confirm our findings.

COVID-19 safety: aerosol-generating procedures and cardiothoracic surgery and anaesthesia - Australian and New Zealand consensus statement.

INTRODUCTION: Coronavirus disease 2019 (COVID-19) is a contagious disease that is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Health care workers are at risk of infection from aerosolisation of respiratory secretions, droplet and contact spread. There are a number of procedures that represent a high risk of aerosol generation during cardiothoracic surgery. It is important that adequate training, equipment and procedures are in place to reduce that risk. RECOMMENDATIONS: We provide a number of key recommendations, which reduce the risk of aerosol generation during cardiothoracic surgery and help protect patients and staff. These include general measures such as patient risk stratification, appropriate use of personal protective equipment, consideration to delay surgery in positive patients, and careful attention to theatre planning and preparation. There are also recommended procedural interventions during airway management, transoesophageal echocardiography, cardiopulmonary bypass, chest drain management and specific cardiothoracic surgical procedures. Controversies exist regarding the management of low risk patients undergoing procedures at high risk of aerosol generation, and recommendations for these patients will change depending on the regional prevalence, risk of community transmission and the potential for asymptomatic patients attending for these procedures. CHANGES IN MANAGEMENT AS A RESULT OF THIS STATEMENT: This statement reflects changes in management based on expert opinion, national guidelines and available evidence. Our knowledge with regard to COVID-19 continues to evolve and with this, guidance may change and develop. Our colleagues are urged to follow national guidelines and institutional recommendations regarding best practices to protect their patients and themselves. ENDORSED BY: Australian and New Zealand Society of Cardiac and Thoracic Surgeons and the Anaesthetic Continuing Education Cardiac Thoracic Vascular and Perfusion Special Interest Group.

The practicalities of establishing a porcine isolated heart model.

BACKGROUND: The isolated heart apparatus is over 100 years old, but remains a useful research tool today. While designs of many large animal systems have been described in the literature, trouble-shooting and refining such a model to yield a stable, workable system has not been previously described. This paper outlines the issues, in tabular form, that our group encountered in developing our own porcine isolated heart rig with the aim of assisting other workers in the field planning similar work. The paper also highlights some of the modern applications of the isolated heart apparatus. Methods Landrace pigs (50-80 kg) were used in a pilot project to develop the model. The model was then used in a study examining the effects of various cardioplegic solutions on function after reanimation of porcine hearts. During the two projects, non-protocol issues were documented as well as their solutions. These were aggregated in this paper. RESULTS: Issues faced by the group without explicit literature solutions included pig size selection, animal acclimatisation, porcine transoesophageal echocardiography, cannulation and phlebotomy for cross-clamping, cardioplegia delivery, heart suspension and rig tuning. CONCLUSION: Prior recognition of issues and possible solutions faced by workers establishing a porcine isolated heart system will speed progress towards a useable system for research. The isolated heart apparatus remains applicable in transplant, ischaemia reperfusion, heart failure and organ preservation research.

Does Furosemide Increase Oxidative Stress in Acute Kidney Injury?

Furosemide, a loop diuretic, is used to increase urine output in patients with acute kidney injury (AKI). It remains uncertain whether the benefits of furosemide in AKI outweigh its potential harms. We investigated if furosemide influenced oxidative stress in 30 critically ill patients with AKI by measuring changes in F2-isoprostanes (F2-IsoPs), markers of in vivo oxidative stress, in plasma and urine following intravenous furosemide. Urine F2-IsoPs were higher in sepsis (p = 0.001) and increased in proportion to urine furosemide (p = 0.001). The furosemide-induced increase in urine F2-IsoPs differed depending on AKI severity (p < 0.001) and was greatest in those with the most severe AKI. Furosemide had no effect on plasma F2-IsoPs. We demonstrate for the first time that furosemide increases renal oxidative stress in AKI and find that patients with the most severe AKI-to whom the largest doses are likely to be administered-showed the greatest increase in oxidative stress. These findings lead to the hypothesis that the common practice of administering high-dose furosemide to convert oliguric to nonoliguric AKI may induce harmful oxidative stress in the kidneys, and an adequately powered, randomized controlled trial is required to determine if clinical benefits of this dosing strategy justify its potential harms. Antioxid. Redox Signal. 26, 221-226.

Determinants of Urinary Output Response to IV Furosemide in Acute Kidney Injury: A Pharmacokinetic/Pharmacodynamic Study.

OBJECTIVES: This study assessed the determinants of urinary output response to furosemide in acute kidney injury; specifically, whether the response is related to altered pharmacokinetics or pharmacodynamics. DESIGN: Prospective cohort. SETTING: Tertiary ICU. PATIENTS: Thirty critically ill patients with acute kidney injury without preexisting renal impairment or recent diuretic exposure. INTERVENTION: A single dose of IV furosemide. MEASUREMENTS AND MAIN RESULTS: Baseline markers of intravascular volume status were obtained prior to administering furosemide. Six-hour creatinine clearance, hourly plasma/urinary furosemide concentrations, and hourly urinary output were used to assess furosemide pharmacokinetics/pharmacodynamics parameters. Of 30 patients enrolled, 11 had stage-1 (37%), nine had stage-2 (30%), and 10 had stage-3 (33%) Acute Kidney Injury Network acute kidney injury. Seventy-three percent were septic, 47% required norepinephrine, and 53% were mechanically ventilated. Urinary output doubled in 20 patients (67%) following IV furosemide. Measured creatinine clearance was strongly associated with the amount of urinary furosemide excreted and was the only reliable predictor of the urinary output after furosemide (area under the receiver-operating-characteristic curve, 0.75; 95% CI, 0.57-0.93). In addition to an altered pharmacokinetics (p < 0.01), a reduced pharmacodynamics response to furosemide also became important when creatinine clearance was reduced to less than 40 mL/min/1.73 m (p = 0.01). Acute kidney injury staging and markers of intravascular volume, including central venous pressure, brain-natriuretic-peptide concentration, and fractional urinary sodium excretion were not predictive of urinary output response to furosemide. CONCLUSIONS: The severity of acute kidney injury, as reflected by the measured creatinine clearance, alters both pharmacokinetics and pharmacodynamics of furosemide in acute kidney injury, and was the only reliable predictor of the urinary output response to furosemide in acute kidney injury.

Incidence of Venous Thromboembolism and Benefits and Risks of Thromboprophylaxis After Cardiac Surgery: A Systematic Review and Meta-Analysis.

BACKGROUND: Optimal thromboprophylaxis after cardiac surgery is uncertain. This systematic review aimed to define the incidence and risk factors for deep vein thrombosis (DVT), fatal and nonfatal pulmonary embolism (PE), and assess whether venous thromboembolism (VTE) prophylaxis was effective in reducing VTE without complications after cardiac surgery. METHODS AND RESULTS: Two reviewers independently searched and assessed the quality and outcomes of randomized, controlled trials (RCTs) and observational studies on VTE after cardiac surgery in the MEDLINE, EMBASE, and Cochrane controlled trial register (1966 to December 2014). Sixty-eight studies provided data on VTE outcomes or complications related to thromboprophylaxis after cardiac surgery. The majority of the studies were observational studies (n=49), 16 studies were RCTs, and 3 were meta-analyses. VTE prophylaxis was associated with a reduced risk of PE (relative risk [RR], 0.45; 95% confidence interval [CI], 0.28-0.72; P=0.0008) or symptomatic VTE (RR, 0.44; 95% CI, 0.28-0.71; P=0.0006) compared to the control without significant heterogeneity. Median incidence (interquartile range) of symptomatic DVT, PE, and fatal PE were 3.2% (0.6-8.1), 0.6% (0.3-2.9), and 0.3% (0.08-1.7), respectively. Previous history of VTE, obesity, left or right ventricular failure, and prolonged bed rest, mechanical ventilation, or use of a central venous catheter were common risk factors for VTE. Bleeding or cardiac tamponade requiring reoperation owing to pharmacological VTE prophylaxis alone, without systemic anticoagulation, was not observed. CONCLUSIONS: Unless proven otherwise by adequately powered RCTs, initiating pharmacological VTE prophylaxis as soon as possible after cardiac surgery for patients who have no active bleeding is highly recommended.

Successful Heart Transplant after Ten Hours Out-of-body Time using the TransMedics Organ Care System.

OBJECTIVE: We report the successful transplantation of a heart following an out-of-body time of 611 minutes into a recipient with dilated cardiomyopathy and left ventricular assist device implant. PATIENTS: Our patient was urgently waiting for a cardiac transplant whilst receiving LVAD support. Recurrent VF and repeated AICD shocks necessitated this action. RESULTS: Although requiring ECMO and inotropic support in the first 17 hours post-transplant, the patient was discharged from hospital on day 15 post-transplant with normal cardiac function. CONCLUSION: We report some of the salient points of the process and discuss the utility of this technology to an Australian transplant unit.

Increased hospital costs associated with red blood cell transfusion.

BACKGROUND: Red blood cell (RBC) transfusion is independently associated in a dose-dependent manner with increased intensive care unit stay, total hospital length of stay, and hospital-acquired complications. Since little is known of the cost of these transfusion-associated adverse outcomes our aim was to determine the total hospital cost associated with RBC transfusion and to assess any dose-dependent relationship. STUDY DESIGN AND METHODS: A retrospective cohort study of all multiday acute care inpatients discharged from a five hospital health service in Western Australia between July 2011 and June 2012 was conducted. Main outcome measures were incidence of RBC transfusion and mean inpatient hospital costs. RESULTS: Of 89,996 multiday, acute care inpatient discharges, 4805 (5.3%) were transfused at least 1 unit of RBCs. After potential confounders were adjusted for, the mean inpatient cost was 1.83 times higher in the transfused group compared with the nontransfused group (95% confidence interval, 1.78-1.89; p < 0.001). The estimated total hospital-associated cost of RBC transfusion in this study was AUD $77 million (US $72 million), representing 7.8% of total hospital expenditure on acute care inpatients. There was a significant dose-dependent association between the number of RBC units transfused and increased costs after adjusting for confounders. CONCLUSION: RBC transfusions were independently associated with significantly higher hospital costs. The financial implication to hospital budgets will assist in prioritizing areas to reduce the rate of RBC transfusions and in implementing patient blood management programs.