Physicians and Machine-Learning Algorithm Performance in Predicting Left-Ventricular Systolic Dysfunction from a Standard 12-Lead-Electrocardiogram.

Early detection of left ventricular systolic dysfunction (LVSD) may prompt early care and improve outcomes for asymptomatic patients. Standard 12-lead ECG may be used to predict LVSD. We aimed to compare the performance of Machine Learning Algorithms (MLA) and physicians in predicting LVSD from a standard 12-lead ECG. By utilizing a dataset of 13,820 pairs of ECGs and echocardiography, a deep residual convolutional neural network was trained for predicting LVSD (ejection fraction (EF) < 50%) from ECG. The ECGs of the test set (n = 850) were assessed for LVSD by the MLA and six physicians. The performance was compared using sensitivity, specificity, and C-statistics. The interobserver agreement between the physicians for the prediction of LVSD was moderate (κ = 0.50), with average sensitivity and specificity of 70%. The C-statistic of the MLA was 0.85. Repeating this analysis with LVSD defined as EF < 35% resulted in an improvement in physicians' average sensitivity to 84% but their specificity decreased to 57%. The MLA C-statistic was 0.88 with this threshold. We conclude that although MLA outperformed physicians in predicting LVSD from standard ECG, prior to robust implementation of MLA in ECG machines, physicians should be encouraged to use this approach as a simple and readily available aid for LVSD screening.

Disparities in the characteristics and outcomes of patients hospitalized with acute decompensated heart failure admitted to internal medicine and cardiology departments: a single-centre, retrospective cohort study.

AIMS: Efforts are constantly made to decrease the rates of readmission after acute decompensated heart failure (ADHF). ADHF admissions to internal medicine departments (IMD) were previously associated with higher risk for readmission compared with those admitted to cardiology departments (CD). It is unknown if the earlier still applies after recent advancement in care over the last decade. This contemporary cohort compares characteristics and outcomes of ADHF patients admitted to IMD with those admitted to CD. METHODS AND RESULTS: The data for this single-centre, retrospective study utilized a cohort of 8332 ADHF patients admitted between 2007 and 2017. We compared patients' baseline characteristics and clinical and laboratory indices of patients admitted to CD and IMD with the outcome defined as 30 day readmission rate. In comparison with those admitted to CD, patients admitted to IMD (89.5% of patients) were older (79 [70-86] vs. 69 [60-78] years; P < 0.001) and had a higher incidence of co-morbidities and a higher ejection fraction. Readmission rates at 30 days were significantly lower in patients admitted to CD (15.9% vs. 19.6%; P = 0.01). Conflicting results of three statistical models failed to associate between the admitting department and 30 day readmission (odds ratio for 30 day readmission in CD: forced and backward stepwise logistic regression 0.8, 95% confidence interval 0.65-0.97, P = 0.02; stabilized inverse probability weights model odds ratio 1.0, confidence interval 0.75-1.37, P = 0.96). CONCLUSIONS: This contemporary analysis of ADHF patient cohort demonstrates significant differences in the characteristics and outcomes of patients admitted to IMD and CD. Thus, focusing strategies for readmission prevention in patients admitted to IMD may be beneficial.

ECG changes after non-cardiac surgery: a prospective observational study in intermediate-high risk patients.

BACKGROUND: Efforts to mitigate the risk for perioperative cardiac events focus on both patient's and operation's risk and often include a preprocedural electrocardiogram (ECG). The merits of postprocedural ECG for detection of occult cardiac events occurring during surgery are unknown. We aim to explore the incidence of pre, and new postprocedural ECG pathologies in an intermediate-high risk population undergoing non-cardiac surgery. METHODS: This single-center, prospective, observational study, included patients older than 18 years with at least two cardiovascular risk factors who were scheduled for non-cardiac surgery. All patients had pre, and postprocedural ECG. The ECG was analyzed and coded according to the Minnesota criteria. A multivariable logistic regression analysis was performed for indices associated with new postoperative ECG pathologies. RESULTS: A total of 217 patients were enrolled. Preoperative pathologic ECG changes were recorded in 62.2% of the patients. Postoperatively, new ECG pathologies were documented in 49.8% of patients, most commonly T-wave changes (36.4% of changes). Pathologic ECG changes at baseline (OR 3.15, 95% CI [1.61-6.17]; P<0.01), diabetes (OR 1.93, 95% CI [1.02-3.64]; P=0.04), history of ischemic heart disease (OR 2.14, 95% CI [1.03-4.47]; P=0.04), higher volumes of fluid replacement (OR 1.70, 95% CI [1.10-2.61]; P=0.01) and higher levels of preoperative hemoglobin (OR 1.24, 95% CI [1.04-1.47]; P=0.01) were all independently associated with postoperative ECG changes. CONCLUSIONS: Pre-, but most importantly, postoperative ECG changes are common in intermediate-high risk surgical patients. Postoperative ECG may be valuable to disclose silent cardiovascular events that occurred during surgery.

Non-invasive hemodynamic profiling of patients undergoing hemodialysis - a multicenter observational cohort study.

BACKGROUND: Intradialytic blood pressure (BP) measurement is currently the main parameter used for monitoring hemodynamics during hemodialysis (HD). Since BP is dependent on cardiac output and total peripheral resistance, knowledge of these parameters throughout the HD treatment would potentially be valuable. METHODS: The use of a novel non-invasive monitoring system for profiling hemodynamic response patterns during HD was explored: a whole-body bio-impedance system was used to assess cardiac index (CI), total peripheral resistance index (TPRI), cardiac power index (CPI) among other parameters in chronic HD patients from 4 medical centers. Measurements were made pre, during and post dialysis. Patients were grouped into 5 hemodynamic profiles based on their main hemodynamic response during dialysis i.e. high TPRI; high CPI; low CPI; low TPRI and those with normal hemodynamics. Comparisons were made between the groups for baseline characteristics and 1-year mortality. RESULTS: In 144 patients with mean age of 67.3 ± 12.1 years pre-dialysis hemodynamic measurements were within normal limits in 35.4% but only 6.9% overall remained hemodynamically stable during dialysis. Intradialytic BP decreased in 65 (45.1%) in whom, low CPI (47 (72.3%)) and low TPRI (18 (27.7%) were recorded. At 1-year follow-up, mortality rates were highest in patients with low CPI (23.4%) and low TPRI (22.2%). CONCLUSIONS: Non-invasive assessment of patients' response to HD provides relevant hemodynamic information that exceeds that provided by currently used BP measurements. Use of these online analyses could potentially improve the safety and performance standards of dialysis by guiding appropriate interventions, particularly in responding to hypertension and hypotension.

Impact of Marital Status on the Outcome of Acute Coronary Syndrome: Results From the Acute Coronary Syndrome Israeli Survey.

Background Marriage is one of the common forms of social support. Conflicting evidence exists about the impact of marital status on the outcomes of patients with acute coronary syndrome ( ACS ). It is further not clear if sex disparity exists in the outcome of married and nonmarried patients with ACS. Methods and Results Data from the ACS Israeli Survey, collected between 2004 and 2016, were used to compare baseline characteristics, clinical indexes, and outcomes of married and nonmarried patients with ACS. Cox regression analysis and propensity score matching were used to explore if marital status was independently associated with long-term outcome. Of 7233 patients included with reported marital status, 5643 (78%) were married. Married patients were younger (62.69±12.07 versus 68.47±14.84 years; P<0.001), more frequently men (83.1% versus 54.8%; P<0.001), and less likely to be hypertensive (61.1% versus 69.3%; P<0.001). All-cause mortality incidence at 30 days and at 1 year was lower in married patients (3.1% versus 7.6% [ P<0.001]; and 7.1% versus 15.3% [ P<0.001], respectively). After adjusting for multiple covariates, the hazard ratio for 5-year all-cause mortality for married patients was 0.74 (95% CI , 0.62-0.88). Similar results were observed after propensity score matching. Kaplan-Meier estimates for all-cause mortality at 5 years demonstrated the best prognosis for married men and the worst for nonmarried women. Conclusions Marriage is independently associated with better short- and long-term outcomes across the spectrum of ACS . Attempts to intensify secondary prevention measures should focus on nonmarried patients and especially nonmarried women.

The association between neck adiposity and long-term outcome.

Anthropometric indices of obesity (e.g. body mass index, waist circumference and neck circumference) are associated with poor long-term cardiovascular outcome. Prior studies have associated neck circumference and central body adiposity. We explored the association between neck fat volume (NFV) and long-term cardiovascular outcome. The study provides a retrospective analysis of all patients undergoing computerized tomography angiography for suspected cerebrovascular accident between January and December 2013. NFV was assessed by three dimensional reconstructions and was adjusted to height to account for differences in body sizes, thus yielding the NFV/height ratio (NHR). Univariate and multivariate analysis were utilized to explore the association between various indices including NHR and all-cause mortality. The analysis included 302 patients. The average age was 61.9±14.3 years, 60.6% of male gender. Diabetes mellitus, hypertension and cardiovascular disease were frequent in 31.5%, 69.9%, and 72.2% of patients, respectively. The median NHR was 492.53cm2 [IQR 393.93-607.82]. Median follow up time was 41.2 months, during which 40 patients (13.2%) died. Multivariate analysis adjusting for age, sex, and diabetes mellitus indicated an independent association between the upper quartile of NHR and all-cause mortality (hazard ratio = 2.279; 95% CI = 1.209-4.299; p = .011). NHR is a readily available anthropometric index which significantly correlated with poor long-term outcome. Following validation in larger scale studies, this index may serve a risk stratifying tool for cardiovascular disease and future outcome.

Usefulness of Serum Procalcitonin as a Markerfor Coexisting Infection in Patients With Acute Myocardial Infarction.

A significant proportion of patients with acute myocardial infarction (AMI) also present with clinical manifestations of inflammatory response, which may be confused with a concomitant infection. This leads to a dilemma regarding the empiric use of antibiotics. We explored if serum procalcitonin (PCT), which is known to be elevated in bacterial infections, may be utilized to rule-out bacterial infection in AMI patients. In this prospective, single center study, PCT was collected within 48 hours from AMI patients. Patients' demographic, clinical, and laboratory data were collected prospectively. Two experienced infectious diseases specialists blinded to the PCT results independently determined the presence of infection in every patient. Sensitivity, specificity, positive predictive value, negative predictive value, and the area under the receiver operating characteristic curve were calculated to determine the accuracy of PCT, fever, white blood cell (WBC) count, and C-reactive protein (CRP) levels for the diagnosis of the infection. The analysis included 230 AMI patients (age 63.0 ± 13.0 years) of whom 36 (15.6%) had coexisting infections. The best cutoff for PCT as a differentiating marker between patients with and without coexisting infection was 0.09 ng/dl (sensitivity 94.4%, specificity 85.1%, area under the curve 0.94). PCT outperformed CRP, WBC, and fever for diagnosing infection. In conclusion, compared with CRP, fever, and WBC, serum PCT had a better performance in differentiating infected from noninfected AMI patients and thus should be considered as an adjunct test when facing the dilemma of initiating empiric antibiotic in AMI patient demonstrating inflammatory signs.