Association Between Off-site Central Monitoring Using Standardized Cardiac Telemetry and Clinical Outcomes Among Non–Critically Ill Patients

Importance  Telemetry alarms involving traditional on-site monitoring rarely alter management and often miss serious events, sometimes resulting in death. Poor patient selection contributes to a high alarm volume with low clinical yield.

Objective  To evaluate outcomes associated with an off-site central monitoring unit (CMU) applying standardized cardiac telemetry indications using electronic order entry.

Design, Setting, and Participants  All non–intensive care unit (ICU) patients at Cleveland Clinic and 3 regional hospitals over 13 months between March 4, 2014, and April 4, 2015.

Exposures  An off-site CMU applied standardized cardiac telemetry when ordered for standard indications, such as for known or suspected tachyarrhythmias or bradyarrhythmias.

Main Outcomes and Measures  CMU detection and notification of rhythm/rate alarms occurring 1 hour or less prior to emergency response team (ERT) activation, direct CMU-to-ERT notification outcomes, total telemetry census, and cardiopulmonary arrests in comparison with the previous 13 months.

Results  The CMU received electronic telemetry orders for 99 048 patients (main campus, 72 199 [73%]) and provided 410 534 notifications (48% arrhythmia/hemodynamic) among 61 nursing units. ERT activation occurred among 3243 patients, including 979 patients (30%) with rhythm/rate changes occurring 1 hour or less prior to the ERT activation. The CMU detected and provided accurate notification for 772 (79%) of those events. In addition, the CMU provided discretionary direct ERT notification for 105 patients (ventricular tachycardia, n = 44; pause/asystole, n = 36; polymorphic ventricular tachycardia/ventricular fibrillation, n = 14; other, n = 11), including advance warning of 27 cardiopulmonary arrest events (26%) for which return of circulation was achieved in 25 patients (93%). Telemetry standardization was associated with a mean 15.5% weekly census reduction in the number of non-ICU monitored patients per week when compared with the prior 13-month period (580 vs 670 patients; mean difference, −90 patients [95% CI, −82 to −99]; P < .001). The number of cardiopulmonary arrests was 126 in the 13 months preintervention and 122 postintervention.

Conclusions and Relevance  Among non–critically ill patients, use of standardized cardiac telemetry with an off-site central monitoring unit was associated with detection and notification of cardiac rhythm and rate changes within 1 hour prior to the majority of ERT activations, and also with a reduction in the census of monitored patients, without an increase in cardiopulmonary arrest events.

Quiz Ref IDIn studies involving traditional on-site monitoring for non–intensive care unit patients, more than 90% of alarms were without immediate clinical relevance and contributed to clinical desensitization referred to as alarm fatigue.^1-4 In the ALARMED study, on-site cardiac telemetry alarms altered management in only 0.2% of patients while generating an average 4.7 alarms per hour.⁴ Alarm fatigue has been linked to serious adverse events including death and was likely underreported.^5-9 In a 5-year observational study of nearly 9000 patients, only 56% of cardiac arrests among telemetry-monitored patients were detected appropriately.¹ True events are missed amid systematic desensitization created by false alarms.³ Alarm management failures have garnered national attention, leading to a 2014 National Patient Safety Goal from The Joint Commission requiring effective alarm management policies by 2016.¹⁰ Low yield and poor patient selection led the Society of Hospital Medicine to designate non–intensive care unit cardiac telemetry monitoring as a practice requiring protocol-driven governance to avoid overutilization.¹¹

Cardiac telemetry standardization to the 2004 American Heart Association recommendations¹² has been associated with a 70% reduction in telemetry utilization without increasing rapid response activations, codes, or deaths and a $13 199 per day cost reduction.¹³ Dedicating a nurse to continuous rhythm monitoring was previously associated with improved fidelity of true rhythm recognition events, including sustained ventricular tachycardia.^14,15 However, it is unknown if dedicated monitoring personnel at an off-site, central monitoring unit (CMU) can provide effective detection and notification to clinical nursing personnel and also integrate with the dedicated emergency response teams in a large multihospital system. Quiz Ref IDOff-site monitoring can minimize noise distraction from hospital activity, centralize staffing, and allow standardized practices. The purpose of this study is to evaluate the clinical outcomes associated with an off-site CMU applying standardized cardiac telemetry.

Box Section Ref ID

Quiz Ref IDA dedicated off-site facility provided continuous cardiac rhythm monitoring for the Cleveland Clinic main campus and 3 regional hospitals and was in full clinical deployment prior to telemetry standardization. In this model, 1 monitoring technician provides continuous cardiac monitoring for up to 48 patients and also provides blood pressure, pulse oximetry, and respiratory rate notifications on request. Lead technicians provide on-site oversight and supervision for real-time rhythm interpretation and detection management requiring clinical escalation to charge nursing personnel. Each workstation includes the electronic medical record with a monitored patient census populated by a clinician’s order, and a telephone system with regularly updated nursing assignments. Patient monitoring is regarded as a shared responsibility between the CMU and nursing, with both sides accountable in the process. Protocol-driven CMU communication with nursing staff occurs via direct mobile phone or use of a crisis phone for emergencies to prevent delays. Updated nursing assignments are provided to the CMU at shift change and disseminated appropriately.

Estimated costs were provided by clinical engineering and the technology departments based on institutional contracts negotiated with specific vendors. Central monitoring capital and operational costs are determined by vendor-specific hardware, routers, software platforms, licensing agreements, secure data lines with information technology support, and monitoring personnel costs, using a ratio of 5.2 full-time employees per 48 monitored patients.

CMU notifications to each clinical nursing unit and hospital were categorically tracked as arrhythmia or hemodynamic notifications, and nonarrhythmia or hemodynamic notifications in a call log database. The latter category includes calls for disruption of monitoring, including lead failures, and other maintenance calls such as low battery alarms. CMU policy authorizes direct communication with an emergency response team (ERT) at the main campus in parallel with bedside nursing when rapid patient deterioration is suspected. The ERT includes dedicated physician, nursing, and respiratory therapy personnel. The ERT is authorized to bypass the primary clinical service to assume temporary emergency management and achieve compliance with Advanced Cardiac Life Support protocols.

Effective March 4, 2014, the Cleveland Clinic health system applied guideline-based standardized telemetry criteria requiring electronic order entry with selection of an approved indication or an “other” category with free text designation of need falling outside approved indications. These standardized criteria were developed by a multidisciplinary physician and nursing task force using available evidence, the 2004 American Heart Association guidelines, and perceived health system needs. The electronic order alerts the patient’s nurse and triggers enrollment in the CMU census. The nurse acknowledges the order electronically, applies the electrodes, and then provides electronic verification of task completion. The assigned monitor technician is then prompted to initiate monitoring according to the clinician specified or to defaulted parameters. An advisory is automatically generated at 72 hours, by the electronic health record, as a reminder of the need to reassess ongoing telemetry and reorder if appropriate. Clinicians were asked to select the most representative indication if more than 1 is applied.

The protocol received Cleveland Clinic institutional review board approval with participant consent waiver for exempt status. Variables of interest were collected over the 13-month study period (March 4, 2014, to April 4, 2015) timing from the onset of telemetry standardization. The total weekly monitored telemetry census and overall hospital census were tracked for CMU hospitals and compared with the previous 13 months. The principal CMU outcome measure of accurate detection and notification for discernible rhythm and rate alarms occurring within 1 hour of ERT activation were characterized over the study period. In addition, all outcomes resulting from discretionary direct CMU-to-ERT communications were tracked and reported including subsequent ERT team deployments, cardiopulmonary arrest events, and return of spontaneous circulation survival outcomes. The total number cardiopulmonary arrest events, which included monitored and unmonitored patients, were also reported over the entire study period and for the previous 13 months.

All ERT activations were adjudicated and coded into the study database (REDCap version 5.8.2 [2015], Vanderbilt University) by a lead technician, along with digitally attached completely scanned rhythm strips inclusive of the event and 1 hour prior (ie, “full disclosure” rhythm strips). The database was regularly audited by an independent TeleHealth Outcomes Coordinator. The coordinator provided regular outcomes reporting to the medical director throughout the study period for quality assurance purposes and attended monthly cardiopulmonary arrest reviews with representation from nursing, quality, risk management, and the ERT. The outcomes coordinator was authorized to investigate all source documentation and modify the database to adjudicate any discovered errors or discrepancies, with input from the CMU medical director and nursing leadership.

Categorical variables were expressed as numbers and percentages and 2-sided statistical comparisons made by χ² testing. Continuous variables were expressed as means, and comparisons made by analysis of variance or t test reporting mean difference, with 95% confidence intervals and 2-tailed significance testing.

P < .05 was considered statistically significant. Additional exported REDCap data analysis was performed using SPSS software (SPSS Inc).

Electronic telemetry orders were placed with indications selected for 99 048 patients, including 72 199 at the main campus (73%) and the remaining 26 849 (27%) at the 3 regional hospitals. Utilization frequencies for each indication category are listed in Table 1. The most common indication for cardiac telemetry was for known or suspected atrial or ventricular tachyarrhythmias (n = 16 243 [16%]). The least common indication involved palliative care in which arrhythmias were known to cause patient discomfort (n = 44 [<1%]). There was significant clinician utilization of indications absent from the 2004 guidelines, including metabolic derangement (n = 5774 [6%]), respiratory disorders (n = 3159 [3%]), seizure monitoring (n = 2834 [3%]), evaluation of stroke/transient ischemic attack (n = 2820 [3%]), deep vein thrombosis/pulmonary embolism (n = 969 [1%]), and drug exposures (n = 577 [<1%]). The free-texted “other” category was frequently used (n = 15 375 [16%]), of which the 2 most common responses were grouped as “hypotensive disease states” that included gastrointestinal bleeding, sepsis or bacteremia, and pancreatitis (n = 1206 [1%]). All remaining free-texted responses accounted for less than 1% after nonunique responses were grouped; these remaining responses were reassigned when appropriate to the other appropriate categories.

Representative capital and operational costs are provided in the eTable in the Supplement. Estimated capital costs for CMU implementation range between $2.3 million to $4.7 million, which include both capital and operational expenditures, and vary according to the monitored census and negotiated vendor pricing agreements.

Over 13 months, standardized cardiac telemetry was associated with a mean 15.5% weekly census reduction in the number of non-ICU monitored patients per week when compared with the prior 13-month period (580 vs 670 patients; mean difference, −90 patients [95% CI, −82 to −99]; P < .001) (Figure). The reduction in telemetry census occurred immediately (first-week reduction, 666 vs 595 patients [−11%]) and was sustained over the entire study period. The measured difference was not attributable to changes in total hospital occupancy census when compared with the previous 13 months (1004 vs 997 patients; mean difference, +7 patients [95% CI, −8 to +12]; P = .38). The number of cardiopulmonary arrests was 126 in the 13 months preintervention and 122 postintervention. The CMU provided 410 534 notifications among 61 nursing units, including 312 363 (76%) at the main campus and 98 171 (24%) at the regional hospitals. The distribution of CMU notifications were divided into those for arrhythmia or hemodynamic reasons vs nonarrhythmia or hemodynamic reasons (Table 2). The dominant nonarrhythmia or hemodynamic notification accounting for 80% of this category was for lead failure with total telemetry disruption (n = 170 604).

There were a total of 5656 ERT events during the study period, of which 3243 (57%) occurred among monitored patients. Of those 3243 ERT activations, there were 979 patients (30%) with a detectable rhythm or rate change within 1 hour prior to ERT activation in review of adjudicated full disclosure data. The CMU detected and provided accurate notification in 772 of these 979 patients (79%). The remaining 207 patients (21%) went without notification and were divided into missed events (n = 176/207 [85%]), simultaneous alarm events with ERT activation (n = 16/207 [8%]), and process failures (n = 15/207 [7%]).

A partial explanation was identified in 126 of the 176 missed events (61%), including 39 cases in which ERT activation was triggered within 5 minutes of the alarm event, which is the allowable time to review an event and contact the nurse. There were also 53 cases in which failure to notify was partially explained because of minor heart rate changes (10/min to 20/min from baseline), 30 nonsustained atrial arrhythmia events, and 4 rhythm changes to rate-controlled atrial fibrillation in patients with prior paroxysmal atrial fibrillation. Although of limited clinical significance, these explanations are not regarded as desirable.

In the remaining 50 of 176 cases (39%), there was no explanation identified. This included 4 missed events for lead failure with telemetry disruption, 13 bradyarrhythmia events, 20 instances of sinus tachycardia, 9 instances of supraventricular tachycardia, 2 premature ventricular contractions, and 2 nonsustained ventricular tachycardia alarms. There were no adverse events in any of these 50 patients. However, there were a total of 7 cardiopulmonary arrest events in the patients without CMU notification. This included 1 bradycardic arrest with successful resuscitation in the simultaneous event group and 3 pulseless electrical activity arrests with successful resuscitation in the process failure group. In each case, the CMU was unable to reach the bedside nurse and was in the process of escalating the call to a nursing supervisor when the event occurred. Two cardiopulmonary arrest events occurred within 5 minutes of the alarm, including 1 patient with a bradycardic arrest who was resuscitated and another patient with a pulseless electrical activity arrest who died. The remaining code occurred in a patient who experienced minor heart rate alarms (10/min to 20/min from baseline) in the preceding hour who later developed a pulseless electrical activity arrest (with resuscitation) without a clear association.

At the main campus, there were 105 events in which the CMU exercised its discretion to provide direct communication to the ERT in parallel with the nursing unit for serious events occurring in real time, including monomorphic ventricular tachycardia (n = 44), prolonged pause events or asystole (n = 36), polymorphic ventricular tachycardia or fibrillation (n = 14), and other (n = 11). All of those provided notifications resulted in formal activation and mobilization of the ERT and subsequent clinical intervention, including 27 cardiopulmonary arrest events (26%). In those cardiopulmonary arrest cases, return of spontaneous circulation was achieved in 25 of 27 patients (93%). This includes shock-appropriate rhythms in 12 of 27 patients (41%), with appropriate defibrillation within 3 minutes or less in 10 of 10 patients (100%). Two patients with shock-appropriate rhythms spontaneously converted during cardiopulmonary resuscitation prior to shock administration. Two patients died despite use of appropriate Advanced Cardiac Life Support protocol, 1 patient with pulseless electrical activity arrest and 1 patient with ventricular fibrillation arrest despite timely defibrillation.

Quiz Ref IDThe use of a centralized monitoring unit with standardized order entry at the Cleveland Clinic main campus and 3 regional hospitals accurately detected and reported 79% of adjudicated rhythm and rate changes within 1 hour of an ERT activation. In addition, the CMU used its judgment to directly notify the ERT of the impending deterioration of the condition of 105 patients to elicit urgent clinical intervention, including 27 patients who subsequently coded and an impressive 25 of whom survived (93%). Meanwhile, telemetry standardization was associated with a 15.5% census reduction, likely owing to the removal of low-risk patients as demonstrated by no overall increase in codes (122 vs 126 in the previous 13 months) and not accounted for by overall hospital census, which was slightly greater during the study period. In outcome studies involving nonstandardized telemetry applying traditional on-site monitoring, only 56% patients with cardiac arrest were appropriately detected.^1-4 Normal hospital activities occurring at the nursing station might potentially distract on-site personnel from continuous vigilant patient monitoring, in addition to the possibility of vigilance being divided by other on-site duties. Off-site monitoring allows dedicated personnel to provide patient monitoring removed from the hospital wards with centralized staffing and standardized practices. A CMU also allows oversight and supervision by lead technicians (ie, somebody to watch those who are watching) to try to ensure continuous monitoring and mitigate lapses.

In the present analysis, telemetry standardization was associated with a modest 15.5% census reduction, compared with the 70% reduction reported by a 2014 telemetry standardization study by Dressler et al.¹³ Both studies demonstrated comparable clinical outcomes ostensibly related to the removal of low-risk patients unlikely to benefit from telemetry but likely to generate nuisance alarms, with resulting alarm fatigue. However, it is unclear why the 2 studies had a large discordance in census reduction. One possibility is that the study by Dressler et al did not include indications absent from the 2004 guidelines and also did not include an “other” category. It is also possible that our institution exhibited less telemetry overutilization at baseline. Nonetheless, patient selection is only one aspect of inefficient telemetry utilization. More than half of the 410 534 calls made by the CMU were for nonarrhythmia or hemodynamic reasons, including 80% for connectivity failures. This suggests further efforts are needed to mitigate alarm fatigue. Future studies can use our descriptive data as a benchmark, given limited scientific data to date, and should evaluate new technologies with better skin adherence, such as patch-based monitors.

Despite a large call volume, CMU technicians were able to apply good discretionary judgment in providing direct ERT notification. These data demonstrate that integrating a CMU and an ERT team is feasible, which is particularly important for hospital systems with dedicated emergency response teams in which operational and capital costs permit scalability. Once the required resources are in place, the CMU can extend its operability to hospitals that are widely geographically separated while interfacing directly with site-specific ERTs. Future work and technological innovation are needed to further improve efficiency and reduce costs.

Quiz Ref IDLimitations of the present analysis include lack of a randomized or crossover design model. There were 2 simultaneous changes as this program was implemented. These included a centralized monitoring system and education on the use of referrals. These findings come from a single large major tertiary medical system, and generalization to other institutions will require further evaluation. The reported data are largely descriptive and dependent on historical and literature-based comparisons. However, these findings may inform future studies.

Among non–critically ill patients, use of standardized cardiac telemetry with an off-site central monitoring unit was associated with detection and notification of cardiac rhythm and rate changes within 1 hour prior to the majority of ERT activations, and also with a reduction in the census of monitored patients, without an increase in cardiopulmonary arrest events.

Corresponding Author: Daniel J. Cantillon, MD, Heart and Vascular Institute, Cleveland Clinic, 9500 Euclid Ave, Desk J2-2, Cleveland, OH 44195 (cantild@ccf.org).

Author Contributions: Dr Cantillon had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Cantillon, Loy, Pengel, Brosovich, Hamilton.

Acquisition, analysis, or interpretation of data: Cantillon, Loy, Burkle, Hamilton, Khot, Lindsay.

Drafting of the manuscript: Cantillon, Loy, Burkle.

Critical revision of the manuscript for important intellectual content: Cantillon, Loy, Pengel, Brosovich, Hamilton, Khot, Lindsay.

Statistical analysis: Cantillon, Burkle.

Obtaining funding: Cantillon, Pengel, Brosovich.

Administrative, technical, or material support: Cantillon, Loy, Pengel, Brosovich, Hamilton, Khot, Lindsay.

Study supervision: Cantillon, Pengel, Brosovich, Hamilton.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Additional Contributions: We would like to recognize the clinical contributions of all current and former Cleveland Clinic lead and monitor technicians, in addition to Bryan Dodrill, central monitoring unit (CMU) supervisor. The CMU workforce are Cleveland Clinic employees who receive paid compensation for their clinical work. We also recognize the Cleveland Clinic Telemetry Standardization Task Force, who received paid compensation from the Cleveland Clinic for clinical work, administrative work, or both but not specifically for this committee service: Michael Amalfitano, DO (Cardiology, Cleveland Clinic Medina Hospital), Ravi Bolla, MD (Cardiology, Cleveland Clinic Lutheran Hospital), Daniel Cantillon, MD (Cardiology/EP, Cleveland Clinic Main Campus), Jason Confino, MD (Cardiology, Cleveland Clinic Euclid Hospital), Mary Curran, MBA (Executive Administration, Cleveland Clinic Main Campus), Meghana Halkar, MD (Hospital Medicine, Cleveland Clinic Hillcrest Hospital), Steven Hata, MD (Anesthesia/Critical Care, Cleveland Clinic Main Campus), M. Shazam Hussain, MD (Neurology, Cleveland Clinic Main Campus), Michael Kalus, MD (Cardiology, Cleveland Clinic South Pointe Hospital), Terry Kilroy, MD (Critical Care, Cleveland Clinic Lakewood Hospital), Wael Khoury, MD (Cardiology, Cleveland Clinic Marymount Hospital), Molly Loy, RN, MSN (Central Monitoring Unit, Cleveland Clinic Administrative Campus), Praful Maroo, MD (Cardiology, Cleveland Clinic Fairview Hospital), Linda McHugh, MBA (Executive Administration, Cleveland Clinic Main Campus), Shannon Pengel, RN, MSN (Nursing, Cleveland Clinic Main Campus), Marc Petre, PhD (Clinical Engineering, Cleveland Clinic Medical Operations), Gwen Print, RN (Nursing, Cleveland Clinic Fairview Hospital), Kathleen Rorapaugh, RN (Nursing, Cleveland Clinic Lakewood Hospital), Anthony Vlastaris, MD (Cardiology, Cleveland Clinic Fairview Hospital), Martin Wiseman, MD (Cardiology/EP, Cleveland Clinic Hillcrest Hospital).