When the machine is wrong. Characteristics of true and false predictions of Out-of-Hospital Cardiac arrests in emergency calls using a machine-learning model.

BACKGROUND: A machine-learning model trained to recognize emergency calls regarding Out-of-Hospital Cardiac Arrest (OHCA) was tested in clinical practice at Copenhagen Emergency Medical Services (EMS) from September 2018 to December 2019. We aimed to investigate emergency call characteristics where the machine-learning model failed to recognize OHCA or misinterpreted a call as being OHCA. METHODS: All emergency calls were linked to the dispatch database and verified OHCAs were identified by linkage to the Danish Cardiac Arrest Registry. Calls with either false negative or false positive predictions of OHCA were evaluated by trained auditors. Descriptive analyses were performed with absolute numbers and percentages reported. RESULTS: The machine-learning model processed 169,236 calls to Copenhagen EMS and suspected 5,811 (3.4%) of the calls as OHCA, resulting in 84.5% sensitivity and 97.1% specificity. Among OHCAs not recognised by machine-learning model, a condition completely different from OHCA was presented by caller in 31% of the cases. In 28% of unrecognised calls, patient was reported breathing normally, and language barriers were identified in 23% of the cases. Among falsely suspected OHCA, the patient was reported unconscious in 28% of the cases, and in 13% of the false positive cases the machine-learning model interpreted calls regarding dead patients with irreversible signs of death as OHCA. CONCLUSION: Continuous optimization of the language model is needed to improve the prediction of OHCA and thereby improve sensitivity and specificity of the machine-learning model on recognising OHCA in emergency telephone calls.

Temporal trends in major cardiovascular events following first-time myocardial infarction in the reperfusion era - a Danish nationwide cohort study from 2000 to 2017.

AIM: We investigated temporal trends in major cardiovascular events following first-time myocardial infarction (MI) and trends in revascularization and pharmacotherapy from 2000 to 2017. METHODS AND RESULTS: Using nationwide registries, we identified 120 833 Danish patients with a first-time MI between 2000 and 2017. We investigated 30-day and 1-year mortality and the 1-year risk of first-time admission for heart failure (HF) and recurrent MI. Patients were younger with a higher prevalence of hypertension and diabetes in 2015-2017 compared with 2000-2002. The patients were predominantly male (65.6%), and the median age declined by 3 years through the periods. Percutaneous coronary interventions within 7 days after first-time MI increased significantly (2000: 11.4% vs. 2017: 68.6%; Ptrend < 0.001). Cardiovascular medication after first-time MI changed significantly in the same period. Absolute risks and adjusted rates of outcomes were significantly lower in 2015-2017 compared with 2000-2002: 30-day mortality: 6.5% vs. 14.1% [hazard ratio (HR) 0.52, 95% confidence interval (CI): 0.48-0.55); 1-year mortality 10.7% vs. 21.8% (HR 0.52, 95% CI: 0.50-0.55); recurrent MI: 4.0% vs. 7.8% (HR 0.56, 95% CI: 0.51-0.62); and first-time admission for HF: 2.9% vs. 3.7% (HR 0.82, 95% CI: 0.73-0.92). The rates of 30-day/1-year mortality and recurrent MI showed significantly decreasing trends (Ptrend < 0.001). The rates of first-time admission for HF were borderline significant (Ptrend = 0.045). CONCLUSION: From 2000 to 2017, we observed a decreasing risk of recurrent MI, first-time admission for HF, and all-cause mortality in patients with a first-time MI. In the same period, we observed a high rate of guideline-recommended pharmacological treatment after first-time MI as well as increasing rate of early revascularization in Denmark. TRANSLATIONAL PERSPECTIVES: The results from the current study portrait the risk of all-cause mortality, recurrent MI, and first-time admission for HF in a real-life setting with a very high utilization of early revascularization and guideline-recommended pharmacological therapy. We observed a temporal trend of improved survival, reduced risk of recurrent MI, as well as reduced risk of first-time admission for HF after first-time MI from 2000 through 2017. We observed an increase in the overall use of revascularization, as well as early revascularization and use of guideline-recommended pharmacotherapy. Our study reveals important results from real-life, nationwide data, showing a reduced risk of cardiovascular outcomes after first-time MI during the past 20 years. Current guidelines are based on results from clinical trials. Our real-life results add additionally important knowledge on patients' prognosis after first-time MI and underline the importance of treating MI according to guideline recommendations.

Effect of Machine Learning on Dispatcher Recognition of Out-of-Hospital Cardiac Arrest During Calls to Emergency Medical Services: A Randomized Clinical Trial.

Importance: Emergency medical dispatchers fail to identify approximately 25% of cases of out-of-hospital cardiac arrest (OHCA), resulting in lost opportunities to save lives by initiating cardiopulmonary resuscitation. Objective: To examine how a machine learning model trained to identify OHCA and alert dispatchers during emergency calls affected OHCA recognition and response. Design, Setting, and Participants: This double-masked, 2-group, randomized clinical trial analyzed all calls to emergency number 112 (equivalent to 911) in Denmark. Calls were processed by a machine learning model using speech recognition software. The machine learning model assessed ongoing calls, and calls in which the model identified OHCA were randomized. The trial was performed at Copenhagen Emergency Medical Services, Denmark, between September 1, 2018, and December 31, 2019. Intervention: Dispatchers in the intervention group were alerted when the machine learning model identified out-of-hospital cardiac arrest, and those in the control group followed normal protocols without alert. Main Outcomes and Measures: The primary end point was the rate of dispatcher recognition of subsequently confirmed OHCA. Results: A total of 169 049 emergency calls were examined, of which the machine learning model identified 5242 as suspected OHCA. Calls were randomized to control (2661 [50.8%]) or intervention (2581 [49.2%]) groups. Of these, 336 (12.6%) and 318 (12.3%), respectively, had confirmed OHCA. The mean (SD) age among of these 654 patients was 70 (16.1) years, and 419 of 627 patients (67.8%) with known gender were men. Dispatchers in the intervention group recognized 296 confirmed OHCA cases (93.1%) with machine learning assistance compared with 304 confirmed OHCA cases (90.5%) using standard protocols without machine learning assistance (P = .15). Machine learning alerts alone had a significantly higher sensitivity than dispatchers without alerts for confirmed OHCA (85.0% vs 77.5%; P < .001) but lower specificity (97.4% vs 99.6%; P < .001) and positive predictive value (17.8% vs 55.8%; P < .001). Conclusions and Relevance: This randomized clinical trial did not find any significant improvement in dispatchers' ability to recognize cardiac arrest when supported by machine learning even though artificial intelligence did surpass human recognition. Trial Registration: ClinicalTrials.gov Identifier: NCT04219306.

The DANish randomized, double-blind, placebo controlled trial in patients with chronic HEART failure (DANHEART): A 2 × 2 factorial trial of hydralazine-isosorbide dinitrate in patients with chronic heart failure (H-HeFT) and metformin in patients with chronic heart failure and diabetes or prediabetes (Met-HeFT).

OBJECTIVES: The DANHEART trial is a multicenter, randomized (1:1), parallel-group, double-blind, placebo-controlled study in chronic heart failure patients with reduced ejection fraction (HFrEF). This investigator driven study will include 1500 HFrEF patients and test in a 2 × 2 factorial design: 1) if hydralazine-isosorbide dinitrate reduces the incidence of death and hospitalization with worsening heart failure vs. placebo (H-HeFT) and 2) if metformin reduces the incidence of death, worsening heart failure, acute myocardial infarction, and stroke vs. placebo in patients with diabetes or prediabetes (Met-HeFT). METHODS: Symptomatic, optimally treated HFrEF patients with LVEF ≤40% are randomized to active vs. placebo treatment. Patients can be randomized in either both H-HeFT and Met-HeFT or to only one of these study arms. In this event-driven study, it is anticipated that 1300 patients should be included in H-HeFT and 1100 in Met-HeFT and followed for an average of 4 years. RESULTS: As of May 2020, 296 patients have been randomized at 20 centers in Denmark. CONCLUSION: The H-HeFT and Met-HeFT studies will yield new knowledge about the potential benefit and safety of 2 commonly prescribed drugs with limited randomized data in patients with HFrEF.

The inflammatory biomarker YKL-40 as a new prognostic marker for all-cause mortality in patients with heart failure.

BACKGROUND: Despite progress in management of patients with heart failure (HF) these patients still have a poor prognosis. We tested the hypothesis whether the inflammatory biomarker YKL-40 alone or in combination with high-sensitivity C-reactive protein (hs-CRP) and/or N-terminal-pro-B natriuretic peptide (NT-proBNP) could be a new prognostic biomarker for all-cause mortality in patients with HF. METHODS AND RESULTS: A total of 717 of the 1000 patients with severe left ventricular systolic dysfunction included in the EchoCardiography and Heart Outcome Study were included in Denmark and had blood sample available for serum YKL-40 determination. Mean age of patients was 70 years, and 73% were male. During the 7 years follow-up period 458 patients died. Patients were categorised according to serum YKL-40 at entry into four quartiles: quartile I with median serum YKL-40=60 µg/L (5-95% Confidence interval (CI): 30-82), quartile II: YKL-40=107 µg/L (CI: 86-132), quartile III: YKL-40=169 µg/L (CI: 142-221), and quartile IV: YKL-40=286 µg/L (CI: 230-770). Hazard ratios for all-cause mortality were with quartile I as reference 1.33 (CI: 0.99-1.80), 1.35 (CI: 0.99-1.82), and 1.54 (CI: 1.14-2.08) for serum YKL-40 II to IV quartiles, respectively following multivariable adjustment for cardiovascular risk factors (age, left ventricular ejection fraction, gender, history of heart failure, ischemic heart disease, chronic pulmonary disease, diabetes mellitus, stroke, hypertension, NT-proBNP, hs-CRP, and renal function). CONCLUSION: Serum YKL-40 is significantly associated with all-cause mortality in patients with HF and could potentially be a new prognostic biomarker in these patients.

Risk prediction is improved by adding markers of subclinical organ damage to SCORE.

AIMS: It is unclear whether subclinical vascular damage adds significantly to Systemic Coronary Risk Evaluation (SCORE) risk stratification in healthy subjects. METHODS AND RESULTS: In a population-based sample of 1968 subjects without cardiovascular disease or diabetes not receiving any cardiovascular, anti-diabetic, or lipid-lowering treatment, aged 41, 51, 61, or 71 years, we measured traditional cardiovascular risk factors, left ventricular (LV) mass index, atherosclerotic plaques in the carotid arteries, carotid/femoral pulse wave velocity (PWV), and urine albumin/creatinine ratio (UACR) and followed them for a median of 12.8 years. Eighty-one subjects died because of cardiovascular causes. Risk of cardiovascular death was independently of SCORE associated with LV hypertrophy [hazard ratio (HR) 2.2 (95% CI 1.2-4.0)], plaques [HR 2.5 (1.6-4.0)], UACR > or = 90th percentile [HR 3.3 (1.8-5.9)], PWV > 12 m/s [HR 1.9 (1.1-3.3) for SCORE > or = 5% and 7.3 (3.2-16.1) for SCORE < 5%]. Restricting primary prevention to subjects with SCORE > or = 5% as well as subclinical organ damage, increased specificity of risk prediction from 75 to 81% (P < 0.002), but reduced sensitivity from 72 to 65% (P = 0.4). Broaden primary prevention from subjects with SCORE > or = 5% to include subjects with 1% < or = SCORE < 5% together with subclinical organ damage increased sensitivity from 72 to 89% (P = 0.006), but reduced specificity from 75 to 57% (P < 0.002) and positive predictive value from 11 to 8% (P = 0.07). CONCLUSION: Subclinical organ damage predicted cardiovascular death independently of SCORE and the combination may improve risk prediction.

Rationale and design of ATHENA: A placebo-controlled, double-blind, parallel arm Trial to assess the efficacy of dronedarone 400 mg bid for the prevention of cardiovascular Hospitalization or death from any cause in patiENts with Atrial fibrillation/atrial flutter.

BACKGROUND: Atrial fibrillation (AF) is the most commonly encountered clinical arrhythmia, predominantly affecting elderly patients. There is a continued need for new antiarrhythmic drugs to treat the ever-increasing number of patients with this arrhythmia. Dronedarone is a new antiarrhythmic compound currently being developed for treatment of AF. METHODS: The ATHENA trial (A placebo-controlled, double-blind, parallel arm Trial to assess the efficacy of dronedarone 400 mg bid for the prevention of cardiovascular Hospitalization or death from any cause in patiENts with Atrial fibrillation/atrial flutter) is the largest single antiarrhythmic drug trial ever conducted. More than 4,600 patients with a history of AF or atrial flutter (AFL) have been randomized to receive dronedarone 400 mg bid or matching placebo. The primary study endpoint is time to first cardiovascular hospitalization or death from any cause. The study has completed patient enrollment in December 2006 and is expected to end follow-up 1 year later. CONCLUSION: ATHENA will be the largest efficacy and safety trial of dronedarone, a multichannel blocker compound with properties from class I, II, III, and IV antiarrhythmic drugs developed to treat patients with AF.

Hyperinsulinaemia is associated with increased long-term mortality following acute myocardial infarction in non-diabetic patients.

AIMS: To study the impact of disturbances in glucose metabolism on total mortality in non-diabetic patients with acute myocardial infarction. METHODS AND RESULTS: Four hundred and ninety four patients with a verified myocardial infarction and no history of diabetes were studied. The study population comprised a subgroup of patients screened for participation in the Trandolapril Cardiac Evaluation (TRACE) study. At baseline, fasting insulin, fasting glucose, glycosylated haemoglobin (HbA1c), and urinary albumin excretion were measured. Survival status was determined after 6-8 years. Patients with hyperinsulinaemia were more obese and more frequently suffered from hypertension, previous myocardial infarction and congestive heart failure. In a univariate regression analysis, values in the upper quartile of insulin, glucose, HbA1c, and urinary albumin were associated with an excess mortality risk (RR=1.8 (1.2-2.7), p=0.002; RR=1.6 (1.2-2.1), p=0.001; RR= 1.9 (1.3-2.9), p=0.001; RR=1.6 (1.2-2.1), p=0.02 respectively). However, only a high insulin level remained significant in a multivariable analysis (RR=1.54 (1.03-2.31), p=0.04) including baseline variables, left ventricular systolic function and in-hospital complications. CONCLUSIONS: High fasting plasma insulin is an independent risk factor of all-cause mortality in non-diabetic patients with acute myocardial infarction. This justifies future intervention studies aiming at reducing insulin resistance and using fasting insulin as the target variable.

A benefit-risk assessment of class III antiarrhythmic agents.

The prevalence of arrhythmia in the population is increasing as more people survive for longer with cardiovascular disease. It was once thought that antiarrhythmic therapy could save life, however, it is now evident that antiarrhythmic therapy should be administrated with the purpose of symptomatic relief. Since many patients experience a decrease in physical performance as well as a diminished quality of life during arrhythmia there is still a need for antiarrhythmic drug therapy. The development of new antiarrhythmic agents has changed the focus from class I to class III agents since it became evident that with class I drug therapy the prevalence of mortality is considerably higher. This review focuses on the benefits and risks of known and newer class III antiarrhythmic agents. The benefits discussed include the ability to maintain sinus rhythm in persistent atrial fibrillation patients, and reducing the need for implantable cardioverter defibrillator shock/antitachycardia therapy, since no class III antiarrhythmic agents have proven survival benefit. The risks discussed mainly focus on pro-arrhythmia as torsade de pointes ventricular tachycardia.

Dofetilide: a new drug to control cardiac arrhythmia.

Atrial fibrillation (AF) is the most common cardiac arrhythmia. Mortality, and especially morbidity caused by AF, are major and growing health problems in the western world. AF is strongly associated with arterial hypertension, congestive heart failure, valvular heart disease, ischaemic heart disease, and with prevalence increasing with age. A variety of drugs have been used to terminate or prevent AF but, as many antiarrhythmic agents have the potential life-threatening pro-arrhythmia, safety problems remain. Dofetilide (Tikosyn, Pfizer), a new Vaughan Williams class III antiarrhythmic agent, has been developed and approved for the treatment of AF. In contrast to most antiarrhythmic agents, the development programme included two safety studies in high-risk patients. Dofetilide is effective and safe when an elaborate procedure for dosing is implemented. Along with amiodarone and betablockers, dofetilide is the only antiarrhythmic drug, which is recommended by guidelines for the treatment of AF in a wide range of patients.

QTc interval in the assessment of cardiac risk.

In the United States alone 300,000-400,000 people die of sudden cardiac death every year. Much of this mortality is assumed to be caused by ventricular tachyarrhythmias. Prolonged QTc reflect cardiac repolarization prolongation and/or increased repolarization inhomogenity known to be associated with increased risk of arrhythmias. The paper gives a review of the possibilities to assess the risk of ventricular arrhythmia and/or cardiac death from QTc. Prolonged QTc may hold independent prognostic importance for mortality in common diseases as ischemic heart disease and diabetes mellitus where as the prognostic importance in heart failure and arterial hypertension is more uncertain. In more rare diseases as the inborn long QT syndrome the QT interval gives not only important hint to the diagnosis but the magnitude also provides information on prognosis. QTc has probably no independent prognostic importance in hypertrophic cardiomyopathy or in the arrhythmogenic right ventricular disease. The degree of QTc prolonging during treatment with QTc prolonging drugs is prognostic for the risk of ventricular arrhythmia in form of torsade de pointes and QTc prolonging drugs should probably not be prescribed for patients with a QTc greater than 460 ms and withdrawn if QTc exceeds 500 ms during treatment. Data from the DIAMOND study suggest that QTc can be used to point out those heart failure patients who will benefit from antiarrhythmic therapy.