The multiple arrhythmia dataset evaluation database (M.A.D.A.E.).

The convergence between wearable and medical device technologies is a natural progression. Miniaturization has allowed the design of small, compact monitoring systems that can record physiological signals over longer periods of time. Thus, the potential for these devices to expand the understanding of disease progression and patients' clinical status is very high. The accuracy of these devices, however, is dependent upon the computer algorithms utilized in the analysis of the large volume of physiological data monitored and/or recorded by the devices. Automated interpretation of the data by these new technologies, therefore, necessitates closer examination by regulatory organizations. The current requirements for the validation of novel Ambulatory ECG (A-ECG) annotation algorithms are based on the AAMI/ANSI-EC57 and IEC60601-2-47 Standard. These standards are being updated, but they rely on a very limited set of digitized ECG recordings from a couple of ECG databases built in the first half of the 70's. These reference signals are obsolete. We are developing a validation tool for computerized methods designed to detect and monitor cardiac activities based on body-surface ECGs. We will rely on a set of existing digital high-resolution 12­lead A-ECG recordings acquired in cardiac patients and healthy individuals. These ECG signals include a large and unique set of electrocardiographic events. This tool is being qualified by the Center for Devices and Radiological Health of the United States Food and Drug Administration (FDA) as a Medical Device Development Tool (MDDT). This document provides insights into the design of the M.A.D.A.E. database, its functionalities, and its ultimate role in enabling the next generations of automatic interpretation of ECG signals.

The chronology and suddenness of cardiac death after myocardial infarction.

A prospective postinfarction study of 978 patients less tha 66 years of age followed from 1 to 5 years was utilized to evaluate the chronology (interval from hospital discharge to demise) and suddenness (elapsed time from the onset of terminal symptoms to demise) of cardiac death. Clinical information including the patient's history and CCU, 6-hour Holter electrocardiographic, medication, and mortality event data was available on 112 cardiac deaths, with 56% of those with witness deaths dying suddenly (less than or equal to 1 hour). During the first postinfarction year 50% of the nonsudden deaths occurred within the first month after hospital discharge, whereas 84% of the sudden deaths occurred in the 2 to 12 month period after infarction (Chi Square = 6.25, p less than 0.02). There were no clinical variables including Holter-recorded ventricular premature beats that distinguished between early and late or sudden and nonsudden cardiac death. These findings indicate that the chronology and suddenness of the terminal cardiac event are more difficult to predict than had previously been appreciated. The therapeutic implications of these observations are discussed.

Digitalis-associated cardiac mortality after myocardial infarction.

The effect of digitalis therapy on 4-month posthospital cardiac mortality was investigated in 812 patients who survived the hospital phase of acute myocardial infraction. A stepwise multiple logistic regression analysis was used to identify variables associated with increased mortality and to adjust for differences in confounding variables between digitalis and nondigitalis patients. The major 4-month mortality (10 of 26 patients [38.5%]) occurred in digitalis-treated patients with congestive heart failure in the coronary care unit and complex ventricular premature depolarizations (VPDs) on the predischarge Holter recording. Logistic analyses that controlled for confounding variables indicated that digitalis use contributed to the increased mortality rate in this high-risk subset. The predicted mortality difference due to digitalis in patients with congestive heart failure and complex VPDs, adjusted for relevant nondigitalis risk factor variables, was 30% (90% confidence interval 18-42%). This retrospective study suggests that digitalis use increases the early posthospital mortality of myocardial infarction patients with combined electrical and mechanical dysfunction.

Frequency and complexity of ventricular ectopic beats in the posthospital phase of myocardial infarction.

This study is a descriptive report of the occurrence of ventricular arrhythmias in a large, prospectively designed follow-up study of postcoronary patients. Six-hour Holter recordings were obtained on 954 postinfarction patients prior to hospital discharge and at 4 (n = 837), 8 (n = 762), 12 (n = 713), and 24 (n = 487) months post-discharge. Ventricular ectopic beats (VEB) were quantitated by frequency and patterns at each recording, and antiarrhythmic medication usage was recorded. The proportion of patients wtih each VEB frequency (any, less than 9.9/hour, 10 - 19.9/hour, greater than or equal to 20/hour) increased progressively during the 24-month follow-up, with the most significant increase in the interval between baseline and the four-month follow-up recording. Significant (p less than 0.01) associations were found between VEB frequency greater than or equal to 20/hour and almost all VEB patterns for each follow-up; no significant associations were found between VEB prematurity (RR'/QT less than or equal to 1.00) and any of the VEB patterns. Among the survivors, the utilization of diuretics and beta blockers increased (p less than 0.01) and antiarrhythmic agents declined (p less than 0.01) between predischarge and all subsequent follow-up visits. The significance of these findings is discussed.

Ventricular ectopic beats and their relation to sudden and nonsudden cardiac death after myocardial infarction.

The role of ventricular ectopic beats (VEBs) in identifying patients who die of cardiac cause in the posthospital phase of myocardial infarction was evaluated in 940 patients who survived an acute coronary event. Six-hour Holter ECG recordings were obtained before hospital discharge, and VEBs were classified as complex (bigeminal, multiform, repetitive or R on T), simple (one or more VEBs that did not have complex patterns), or not present. Patients were followed 1-60 months (average 36 months) and cardiac mortality was categorized as sudden (less than or equal to 1 hour) or nonsudden (greater than 1 hour) among 98 witnessed cardiac deaths. Complex VEBs were associated with a significantly increased cardiac death rate, but did not discriminate between sudden and nonsudden death. Simple VEBs were associated with a 3-year cardiac mortality rate intermediate between those with complex and those with no VEBs. The relationship between complex VEBs and cardiac mortality was independent of 10 relevant clinical variables.

Cardiac Death in the first 6 months after myocardial infarction: potential for mortality reduction in the early posthospital period.

In a prospective postmyocardial infarction study of 759 patients aged less than 66 years, 42 posthospital cardiac deaths (42 of 759; 6 percent) occurred during a 6 month follow-up period. The average age of those who died was 53.5 +/- 8.8 (+/- standard error) years, and postmortem examination was obtainedon 36 percent. Almost 60 percent of the 6 month posthospital mortality occurred within 2 months after hospital discharge. Fifty-five percent of the cardiac deaths occurred either outside th ehospital or within hospital emergency departments, and 62 percent of the deaths were sudden (within 12 hours) or unwitnessed. The suspected mechanism of cardiac death was a primary arrhythmia in 62 percent, and a definite or probable myocardial infarction was diagnosed in only 41 percent. Use of digitalis and diuretic and antiarrhythmic agents was significantly (P is less than 0.025) greater in this group during the week before death than in a comparison survivor group; no difference in use of propranolol or tranquilizers was noted between the two groups. Fifty percent of the group that died had two or more of the following factors: death outside the hospital, sudden death, primary arrhythmic death. These findings indicate that a considerable potential exists for reducing cardiac death in the early posthospital phase of myocardial infarction.

Clinical significance of ventricular ectopic beats in the early posthospital phase of myocardial infarction.

The clinical significance of ventricular ectopic beats in the posthospital phase of myocardial infarction was studied in 272 patients aged 65 years or less who were followed up for 1 year after the infarction. Ventricular ectopic beats, identified in 6 hour electrocardiographic tape recordings, obtained before hospital discharge (study 1) and 5 months after discharge (study 2) increased in frequency and complexity in the 5 month interval. Ventricular ectopic beats at a rate of 20 or more per hour recorded before discharge were associated with complex ventricular ectopic patterns in the same 6 hour recording and with frequent (20 or more per hour), early cycle and bigeminal patterns in recordings mad 5 months later. Analysis with log-linear modeling indicated that the occurrence of complex ventricular ectopic beats at follow-up examination was associated with the concomitant use of antiarrhythmic agents,but not with use of digitalis, propranolol or tranquilizers. A ventricular ectopic beat frequency of 20 or more per hour at discharge was associated with increased (P less than 0.05) cardiac mortality in the initial 0 to 4 months after discharge but not in the subsequent 8 months; ectopic beats recorded in the 5 month follow-up study were not associated with increased cardiac mortality in the subsequent 5 to 12 months. The prognostic significance of ventricular ectopic beats is discussed in the light of these findings.

The early posthospital phase of myocardial infarction. Prognostic stratification.

Prognostic stratification was carried out on 518 patients less than or equal to 65 years of age who were discharged from the hospital following a definite or probable acute myocardial infarction and followed for four months. The total population was made up of 272 patients hospitalzed in 1973 and 246 patients hospitalized in 1974; one hundred and forty-two variables were collected on each patient. The clinical characteristics of the 1973 and 1974 populations were remarkably similar, and both groups had a four-month posthospital cardiac mortality rate of 4%. Two prognostic stratification schemes were developed on the 1973 population which identified low and high risk groups with meaningfully different four-month cardiac death rates. Both stratification schemes were tested on the 1974 population, and one of the two schemes was validated as identifying a significantly increased cardiac mortality rate in the high as opposed to the low risk group. The four-month posthospital cardiac mortality rate was 3% in the low and 14% in the high risk group (Z = 2.70, P less than 0.003). The high risk group was characterized by two or more of the following characteristics: 1) history of angina at ordinary levels of activity or at rest; 2) CCU hypotension and/or congestive heart failure; 3) ventricular premature beat frequency greater than or equal to 20/hr on a six-hour electrocardiographic tape recording. The low risk group had none or only one of the above characteristcis. The prognostic power of this stratification scheme is such that sixteen percent of the posthospital population can be identified as high risk, and this subgroup contains forty-six percent of the patients who die of cardiac cause in the four-month posthospital interval.

Prognostic grading and significance of ventricular premature beats after recovery from myocardial infarction.

A 6-hour taped electrocardiogram recording was obtained prior to hospital discharge on 193 patients less than 66 years of age with a definite or probable myocardial infarction. All ventricular premature beats (VPBs) were identified on the 6-hour recording, and the various VPB characteristics were related to a complicated course (CC) defined as cardiac death or myocardial reinfarction within 4 months after hospital discharge. Only three univariate VPB characteristics (multiform pattern, bigeminal rhythm, and frequency greater than or equal to 20/hr) were significantly associated with a 4-month posthospital CC. By use of bivariate and multivariate combinations of these three VPB characteristics, a two-grade prognostic stratification system was derived: Grade I VPBs (CC = 8%) included patients with less than 20 VPB/hr but with neither multiform nor bigeminal beats: Grade II VPBs (CC = 31%, P less than 0.001) contained patients with greater than or equal to 20 VPBs/hr or with multiform or bigeminal beats or both. Patients with Grade II VPBs had more severe cardiac disease, but when patients were stratified by the clinical severity of their cardiac disease, Grade II VPBs identified patients with higher CC rates. The clinical significance of these findings together with a critical analysis of the prognostic stratification method are discussed.