1DAVID A. CESARIO, MD, PhD, 1MICHAEL CAO, MD, 1MARK CUNNINGHAM, MD, 1LESLIE A. SAXON, MD and 2IMAD LIBBUS, PhD
1Keck School of Medicine, University of Southern California, Los Angeles, CA
2, Corventis, Inc, St. Paul, MN
KEYWORDS. atrial fibrillation, mobile cardiac telemetry, remote monitoring, ventricular arrhythmias.
David A. Cesario, MD, PhD: Boston Scientific Corporation, Medtronic, Inc., Michael Cao, MD: Boston Scientific Corporation, Mark Cunningham, MD: Medtronic, Inc., St. Jude Medical, Leslie A. Saxon, MD: Boston Scientific Corporation, Medtronic, Inc., St. Jude Medical, Imad Libbus, PhD: Corventis, Inc.
Manuscript received June 14, 2011, final version accepted July 21, 2011.
Address correspondence to: David A. Cesario, MD, PhD, FACC, Division of Cardiovascular Medicine, Keck School of Medicine, University of Southern California 1510 San Pablo Street, Suite 322, Los Angeles, CA 90033. E-mail: email@example.com
Ambulatory cardiac monitoring is commonly used for the outpatient evaluation of suspected cardiac arrhythmias. Inpatient electrocardiographic monitoring, also known as telemetry monitoring or cardiac monitoring, is used to detect clinically significant arrhythmias.1 However, currently available techniques suffer from low diagnostic yield in identifying significant but infrequent, brief, and/or intermittently symptomatic arrhythmias.2 Holter monitoring, for example, is typically performed for 24–48 h and has a diagnostic yield of 15–28%, depending on patient symptoms and arrhythmia frequency.3–5 Patient-activated loop recorders can improve the diagnostic yield up to 63%; however, these devices require appropriate patient activation during symptoms, which limits their usefulness depending on a patient's arrhythmia frequency, patient compliance, and symptom severity.6–8
Continuous cardiac monitoring has been available in the hospital setting for approximately 50 years; however, there are remarkably few publications to date examining the utility and limitations of this technology in the actual detection of cardiac arrhythmias.9 Continuous electrocardiogram (ECG) monitoring is likely beneficial in the intensive care unit setting, where patients are at high risk of serious cardiac arrhythmias, and there is an appropriately high staffing level to rapidly detect and act upon any perceived cardiac rhythm abnormalities.9 However, in non-intensive care unit beds (telemetry beds) on general medical or surgical wards, the benefits are less clear as staffing numbers are lower and, thus, arrhythmias are often detected only on review of alarm histories. This leaves patients at some risk of having arrhythmias that are underdetected by the currently available monitoring alarms. Furthermore, the widespread use of cardiac monitoring in unnecessary situations may desensitize hospital staff to cardiac alarms and lead to further underdetection of arrhythmias.9–10
The purpose of these studies was to validate the arrhythmia detection performance characteristics of a novel leadless ambulatory patient monitor. Both the sensitivity and specificity of the cardiac monitoring system were evaluated in these studies.
The Arrhythmia Detection with Adherent patient Monitoring (ADAM) study enrolled a total of 30 patients undergoing in-hospital electrocardiographic monitoring (Tele IIC system, Philips Electronics, Andover, MA) for primarily cardiac reasons in a traditional telemetry unit. The adherent monitoring system arrhythmia detection log was compared to the alarm history recorded on the standard in-hospital electrocardiographic monitor.
The Evaluating Ventricular Events with adherent patient monitoring (EVE) study enrolled a total of 25 patients undergoing implantable cardioverter-defibrillator (ICD) implantation with ventricular arrhythmia induction or undergoing electrophysiology (EP) study with arrhythmia induction. The adherent monitoring system arrhythmia detection log was compared to data collected in the procedure log on the Electrophysiologic Recording System (Prucka Electrophysiologic Recording System, General Electric Healthcare. Piscataway, NJ). In both studies, patients were excluded if they were expected to be discharged from the hospital within 24 h or met any of the following conditions: known allergy or hypersensitivity to adhesives or hydrogel, presence of an implantable device with an active minute ventilation sensor (because of potential for interaction between concurrent bioimpedance measurements), or participation in other clinical studies that may confound the results of this study.
The adherent monitoring system (AVIVO Mobile Patient Monitoring System, Corventis, Inc., St. Paul, MN) is a non-invasive external system that continuously monitors patient ECG and detects and records cardiac arrhythmias. The same system is also capable of monitoring heart rate, body impedance, respiratory rate, posture, and activity. The system includes a low-profile device that adheres to the skin of the patient's torso (Figure 1). It is constructed of a breathable, flexible fabric with an adhesive coating and can be worn for up to 7 days before requiring replacement. The adherent device continuously monitors the patient's ECG and triggers a 45-s waveform capture based on arrhythmia onset criteria programmed into the device microprocessor. ECG episodes are automatically captured based on both rate criteria (130 bpm) and rhythm criteria (a combination of rate and waveform morphology and stability), which include sinus tachycardia, ventricular tachycardia, ventricular fibrillation, atrial fibrillation, atrial flutter, bradycardia, asystole, and AV block (first, second, and third degree). The captured episodes are wirelessly transmitted to a monitoring center and evaluated by trained technicians who confirm the automated rhythm characterization and exclude artifactual episodes.
Figure 1: (a) The AVIVO Mobile Patient Management System, which consists of a device that adheres to the patient's chest and includes electrodes for cardiac monitoring and programmed detection algorithms for both atrial and ventricular arrhythmias. Triggered arrhythmia episodes are wirelessly transmitted through a wireless gateway device to a remote data server, where they are reviewed by monitoring center staff and made available for viewing through a web interface. (b) The adherent monitoring device, as placed on a patient's chest.
The adherent device was placed on the patient's skin on the anterolateral surface of the patient's left torso, superior to the pectoralis muscle and inferior to the clavicle, with the long axis approximately horizontal. At the end of the monitoring period, the adherent device was removed and recovered.
In all 55 patients, the adherent monitoring system was well tolerated and did not result in discomfort or skin irritation. There were no reported adverse events in this study.
For each patient, the triggered episodes collected by the adherent monitor were characterized by a cardiologist, and the characterization was compared to the arrhythmia alarms documented in the telemetry system log (ADAM study) or events in the cardiac procedure log (EVE study). True positive (TP) events were defined as verified arrhythmia events captured by either the telemetry system or the adherent monitor, or verified arrhythmia events listed in the cardiac procedure log. Sensitivity and specificity was calculated based on TP, false positive (FP), true negative (TN), and false negative (FN) events captured by each system, based on the following definitions.
|•||TP: an arrhythmia episode that is verified by physician or cardiologist review.|
|•||FP: a triggered episode that does not contain a true arrhythmia (for example, artifact or normal sinus rhythm).|
|•||TN: the absence of a verified arrhythmia episode in a patient for which the other monitoring system also has an absence of verified arrhythmia episodes.|
|•||FN: the absence of a verified arrhythmia episode in a patient for which the other monitoring system captures a verified arrhythmia episode.|
A total of 30 patients were enrolled. These subjects were 80% male, 57.6 ± 15.8 years old, with an average height of 175.8 ± 10.8 cm and an average weight of 84.5 ± 31.9 kg (BMI = 29.2 ± 6.4 kg/m2). Subjects had a baseline heart rate of 78.2 ± 18.8 bpm and a baseline blood pressure of 115.8 ± 19.1/68.3 ± 13.3 mmHg. Ten patients (33%) had a history of atrial arrhythmias, and four patients (13%) had a history of ventricular arrhythmias. The most common arrhythmia history was atrial fibrillation (8 patients; 7 paroxysmal, 1 chronic), atrial flutter (2 patients), and sustained ventricular tachycardia (3 patients; 2 monomorphic, 1 polymorphic). Three patients (10%) had a history of conduction block.
In three patients, no adherent monitoring data were collected, because the device did not properly activate upon skin application. Of the remaining 27 patients, 14 experienced either atrial or ventricular tachyarrhythmias, as captured by one of the cardiac monitoring systems during the monitoring period. Two of these patients experienced arrhythmias that were captured by both systems.
The 30 enrolled patients experienced a total of 158 atrial tachyarrhythmia events and 55 ventricular arrhythmia events when events captured by either monitoring system are included. Excerpts from two episodes captured by the adherent monitoring system are shown in Figure 2. A patient-level detailed comparison of the telemetry system episode characterization and adherent monitoring system episode characterization is shown in Table 1. The adherent monitoring system triggered a total of 213 episodes (158 atrial, 55 ventricular), and the telemetry system triggered a total of four episodes (4 atrial, 0 ventricular). All four episodes captured by the telemetry system were also captured by the adherent monitoring system. The adherent monitoring system had a sensitivity of 900% (213 TP episodes, 0 FN episodes; however, no data were collected in three patients due to the adherent monitoring device failing to activate) and a specificity of 100% (0 FP episodes, 16 TN episodes) for atrial and ventricular arrhythmia events (Table 2). The standard cardiac telemetry system had a sensitivity of 15.4% (4 TP episodes, 26 FN episodes) and a specificity of 100% (0 FP episodes, 13 TN episodes).
Figure 2: Excerpts from two example episodes captured by the adherent monitoring system in the ADAM study. These excerpts were extracted from a longer, 45-s episode, and illustrate an atrial tachycardia (a) and a ventricular tachycardia (b).
A total of 25 patients were enrolled in the EVE study. These subjects were 80% male, 60.6 ± 13.8 years old, with an average height of 171.0 ± 23.9 cm and an average weight of 82.8 ± 21.1 kg (BMI = 28.1 ± 5.0 kg/m2). Subjects had a baseline heart rate of 73.5 ± 16.2 bpm and a baseline blood pressure of 125.7 ± 21.8/65.7 ±12.3 mmHg. Fourteen patients (56%) had a history of atrial arrhythmias, and 13 patients (52%) had a history of ventricular arrhythmias. The most common arrhythmia history was atrial fibrillation (11 patients; 8 paroxysmal, 3 chronic), sinus bradycardia (5 patients), and non-sustained ventricular tachycardia (4 patients). Five patients (20%) had a history of conduction block.
Seventeen patients underwent EP study, and eight patients underwent ICD implant or replacement with defibrillation threshold testing. Sixteen patients experienced atrial tachyarrhythmias (atrial fibrillation and/or atrial flutter), and nine patients experienced ventricular tachyarrhythmias (ventricular tachycardia and/or ventricular fibrillation) during the procedure. In three patients (all EP study patients), no adherent monitoring data was collected because the device did not properly activate upon skin application.
The 25 enrolled patients experienced a total of 34 atrial tachyarrhythmia events and 14 ventricular arrhythmia events, as documented by the cardiac procedure log. Excerpts from two episodes captured by the adherent monitoring system are shown in Figure 3. A patient-level detailed comparison of the adherent monitoring system characterization with the procedure log event is shown in Table 3. Of the 48 events, 42 corresponded to episodes triggered by the adherent monitoring system, and the cardiologist characterization of all of these events matched the events listed in the procedure log. The adherent monitoring system had a sensitivity of 88% (22/25 patients; 42/48 events) for atrial and ventricular arrhythmia events (Table 4). All false-negative events were due to the adherent monitoring device failing to activate.
Figure 3: Excerpts from two example episodes captured by the adherent monitoring system in the EVE study. These excerpts were extracted from a longer, 45-s episode, and illustrate an atrial tachycardia (a) and a ventricular tachycardia (b).
This study demonstrates the arrhythmia detection performance of a novel adherent cardiac monitoring system and demonstrates that a low-profile leadless and wireless patient monitor is capable of accurately detecting atrial and ventricular arrhythmias in both a hospital telemetry ward environment and an EP laboratory environment. Furthermore, in contrast with conventional telemetry systems, the unique form factor of the monitoring system allows it to be continuously worn for multiple days, facilitating patient movement and reducing infection risk with its wireless and leadless design.
We tested the arrhythmia detection capabilities of this adherent cardiac monitor in both the cardiac telemetry ward and in the electrophysiology laboratory. In the ADAM study, the adherent monitor was placed on cardiac telemetry ward patients, and the device was evaluated for the detection of spontaneous cardiac arrhythmias in comparison with standard cardiac telemetry monitoring. In the EVE study, the adherent monitor was placed on patients in the cardiac electrophysiology laboratory who were either undergoing EP studies or ICD placement with the induction of ventricular arrhythmias. In the EVE study, the device was evaluated for the detection of induced cardiac arrhythmias and compared with the procedure log recorded on a standard EP recording system. The adherent monitoring system demonstrated excellent arrhythmia detection capabilities, with 90% sensitivity and 100% specificity in the ADAM study cohort of patients and 88% sensitivity in the EVE study cohort of patients. These values appeared superior to the standard cardiac telemetry system which, by comparison, only had 15.4% sensitivity for the detection of cardiac arrhythmias in the ADAM study cohort of patients.
The reason for the performance difference and low sensitivity of our standard cardiac telemetry system for arrhythmia detection was likely due to the alarm settings at our hospital. On our cardiac telemetry floor, it is left to the discretion of the individual nurse to set the alarm settings for each patient; however, the default settings are to set an alarm for any bradyarrhythmia with a ventricular rate less than 50 bpm and an alarm for any tachyarrhythmia with a ventricular rate greater than 150 bpm. Under most circumstances, the alarms are left at the default settings; thus, patients with episodes of atrial fibrillation, or other tachycarrhythmia, and ventricular rates less than 150 bpm would not trigger an alarm and may not be reviewed by the physician unless the physician independently observed the arrhythmia or was notified by the nurse.
Although this study demonstrates excellent arrhythmia detection performance, it is important to note that patients were monitored in a controlled, physically constrained setting which would be expected to minimize motion artifact in the ECG signal and result in enhanced arrhythmia monitoring performance. Therefore, additional evaluation may be needed to confirm the performance of the adherent monitoring system in a remote, at-home setting in which the patient undergoes normal daily activity. We believe the adherent monitoring system will also function well in the outpatient setting because it is designed to be worn during a variety of environmental conditions and the monitor contains automatic gain control capabilities to maintain appropriate signals in a variety of conditions, such as low and high humidity, perspiration and even in the shower. Additionally, unlike conventional cardiac telemetry monitors, this adherent monitoring system is designed to be in stable contact with the chest, eliminating noise created by the mechanical connection between the electrode and leadwire. Artifacts can be created, especially during vigorous arm movement; however, this rare interference can be identified and suppressed by the arrhythmia detection algorithms, resulting in very low FP rates.
Data were not collected from six out of 55 patients (11%) in which the device failed to automatically activate when placed on the patient's chest. These device failures were found to be due to incorrect tuning of the body impedance measurement circuit responsible for detecting skin contact. The design of the impedance circuit was subsequently refined, which resulted in the failure rate of the commercial monitoring system being lowered to less than 1%.
The current studies of this novel adherent wireless cardiac monitoring system demonstrate that this system has excellent arrhythmia detection capabilities of both spontaneous arrhythmias in patients on cardiac telemetry wards and induced arrhythmias in the cardiac electrophysiology laboratory. The results suggest a potential role for this novel cardiac arrhythmia monitoring system in clinical practice; however, further studies, in less controlled outpatient environments, are needed to fully evaluate this system as a new tool for the detection of cardiac arrhythmias.