Remote monitoring of cardiac implantable electronic devices and disease management

Abstract

This reviews the transition of remote monitoring of patients with cardiac electronic implantable devices from curiosity to standard of care. This has been delivered by technology evolution from patient-activated remote interrogations at appointed intervals to continuous monitoring that automatically flags clinically actionable information to the clinic for review. This model has facilitated follow-up and received professional society recommendations. Additionally, continuous monitoring has provided a new level of granularity of diagnostic data enabling extension of patient management from device to disease management. This ushers in an era of digital medicine with wider applications in cardiovascular medicine.

Introduction

The year 2023 marks the 25th year of Europace Journal and also the 10th anniversary of a prescient Europace special issue recognizing the technological revolution of automatic remote monitoring (RM) of patients with cardiac implantable electronic devices (CIEDs).¹ The pioneering unit was a mobile cellular phone that relayed CIED diagnostic data gathered daily to a service centre that then forwarded clinically actionable information to the clinic for review. The system provided a new level of granularity of details regarding device function and arrhythmia burden and flagged critical issues for immediate attention. The issue anticipated how this model would facilitate CIED follow-up, set the standard of care, and usher in an era of digital medicine with wider applications in cardiovascular medicine.^2,3 Here, we review the progress in the last decade.^4–6

Technical advances

The last two decades have witnessed evolution of remote technologies from intermittent patient activated to automatic systems that test every day, changing the paradigm of remote patient management from remote follow-up to continuous RM.

The concept of remote management of CIEDs was born in the early 1970s.⁷ Remote pacemaker follow-up was initially done through trans-telephonic transmissions (TTMs).^4,8 It enabled a restricted interrogation providing rate, rhythm, and device integrity (mainly battery status). Its main efficiency was for CIEDs approaching battery elective replacement interval. It could not detect intermittent problems nor retrieve diagnostic data or rule out device and lead malfunction and therefore could not replace in-office visits.

In the 1990s, a more advanced form of TTM was tested in implantable cardioverter-defibrillators (ICDs) and was effectively the precursor to modern remote follow-up.⁹ The TTM transmitters contained metal bracelets and interrogation heads placed over the ICDs and, when connected to the telephone headset, allowed automatic transmission to the device clinic with a wealth of information including pacing parameters, tachycardia detection criteria and treatment, battery status, last charge time, episode counter, and therapy logs for ventricular tachycardia (VT)/(VF) ventricular fibrillation.

An advance in CIED-based inductive technology allowed for remote interrogation (RI) through wand-based radiofrequency data transfer, in a similar fashion to in-office visits.^8,10 When used, data were transferred from the patient’s CIED via a home transceiver through phone lines or cellular/wireless connections to a remote storage server that could be accessed remotely by the caregivers. Further technological advance allowed for the introduction of an automated RM system that transmits daily data via a transceiver located next to the patient.^11,12 More recently, low-energy Bluetooth technology allows coupling the CIED with the patient’s smartphone, further enhancing connectivity.

Contemporary remote management of patients with CIED involves remote follow-up in which a scheduled device interrogation is performed and RM that automatically transmits alert events and patient-initiated interrogation, which are usually initiated due to concern of device malfunction or suspected arrhythmia or another clinical event.^5,13

Several factors must be taken into account when integrating wireless technology to CIEDs. The first is limiting the power consumption of the system so as not to affect battery longevity significantly (one proprietary technology is associated with minimal drain¹⁴). Communication range of the device is a compromise as longer ranges require more power. Most CIED wireless systems work well within a 3 m range. Similarly, the rate of data transfer must be considered as faster rates require more power and more sophisticated and costly designs. Size is important when dealing with compact implantable devices such as CIEDs, and the antenna can have an important effect on the overall device dimensions. Finally, another important consideration when dealing with wireless technology is the surrounding environment and potential interference by other radiofrequency sources. Most CIEDs utilize the medical implant communication service (MICS) band at approximately 400 MHz frequency. The MICS is an unlicensed, ultralow-power (ULP) mobile radio service for transmitting data in implanted medical devices, thus avoiding interference with other users of the radio spectrum. Each manufacturer has its own implementation of this standard in their proprietary hardware and software, and it is therefore not compatible between vendors. Low-energy Bluetooth is a paradigm shift as connectivity is no longer solely controlled by the device manufacturer but also by the patient's mobile phone and operating system manufacturer (see section below).

Remote device follow-up

Traditionally, post-implant follow-up was conducted in person according to calendar-based intervals. The frequency of 3 monthly ICD interrogations was based on the need to manually reform the capacitor and check battery and lead integrity from first-generation devices. This has been unnecessary for decades since CIEDs are reliable and component failure is infrequent.¹⁵ At the same time, the workload associated with routine follow-up has become challenging in view of increasing volume of implants and data transfer.

Remote patient management is an appealing solution. Remote interrogations promised patients easier access (e.g. from remote locations) and hospitals reduced in-clinic load. However, coordination of 3 monthly checks with patient consumed time.¹⁶ Moreover, this model continued calendar-based appointments that do not detect potentially serious problems occurring between interrogations. Automatic RM changed the paradigm, providing a mechanism for continuous surveillance of an ambulatory population with CIEDs and virtually immediate alert-based notification of changes in device (or patient) condition, even if asymptomatic, e.g. failure of an ICD to properly charge its capacitors and deliver appropriate therapy,¹⁷ inappropriate VT detection caused by supraventricular tachycardia, or development of atrial fibrillation (AF).¹⁸

The large randomized TRUST trial demonstrated that continuous RM reduced the load of in-clinic evaluations by almost 50% (principally by reducing non-actionable routine device evaluations) while maintaining patient safety.¹⁹ This resulted in more efficient deployment of clinic resources to those patients who needed to be seen.²⁰ Counterintuitively, patient retention to follow-up was improved.⁶ Additionally, TRUST confirmed superiority of automatic RM to conventional care for maintaining adherence to 3 monthly checks and for detecting abnormalities as and when they occurred, mostly the same day.^12,19,21,22 This might be critical, e.g. for battery or lead failure (Figures 1 and 2).²³ In comparison, conventional care under-reported occurrence of ICD dysfunction and discovered these events late (47% of all detected events were asymptomatic) (Figures 3). This led to the recommendation of RM for management of advisories that centre around disintegration of high-voltage circuitry, battery depletion, and lead failure and are captured by event triggers.²⁴ In comparison, intensifying surveillance with conventional detection methods, such as increasing the frequency of office visits, e.g. monthly,²⁵ is impractical, onerous, and inefficient since problem incidence is low and denominators high.

Figure 1.

(Left) Event notifications suggestive of AF (top) but accompanying wirelessly transmitted intra-cardiac electrograms indicate that the abnormality was caused by lead noise (likely fracture-related) permitting the diagnosis of a false positive mode switch. (Right) Event notification received for an aborted shock (7:16 p.m.) with an automatic wirelessly transmitted electrogram showing non-physiological signals due to electromagnetic interference. The subject was asymptomatic but, in response to the notification, was seen in office within 24 h (compiled with permission from Varma et al.²³).

Figure 2.

Implantable cardioverter-defibrillator generator with automatic wireless remote monitoring coupled to a Fidelis (MDT 6949) lead. Two event notifications that were transmitted immediately on occurrence of lead fracture, occurring silently during sleep at 4:43 a.m., 6 weeks after last clinic follow-up on 14 November. (Left panel) Wirelessly transmitted electrograms demonstrated irregular sensed events (coupling intervals as short as 78 ms) with VF detection (marked). No therapy was delivered. (Right panel) A separate notification for the same event indicating a lead impedance alert. Electrogram definition was modest in first-generation device (compiled with permission from Varma²⁴).

Figure 3.

Event free survival rates in HM compared with conventional care in the TRUST cohort. Protocol required event notifications were system related (end of service, elective replacement indicator, atrial impedance <250 or >1500 Ohm, ventricular impedance <250 or >1500 Ohm, daily shock impedance <30 or >100 Ohm, and shock impedance <25 or >110 Ohm); arrhythmia-related events (atrial burden > 10%, supraventricular tachycardia detected, VT1 detected, VT2 detected, and VF detected), and ineffective ventricular maximum energy shock (notified on first shock of any sequence in a given episode) (compiled with permission from Varma et al.²³).

The roles of RM extend beyond managing clinic workload and device integrity. Effective follow-up should aim to identify ‘actionable’ interrogations and trigger adequate reactions. Reprogramming changes accounted for the majority of ‘actionable’ interrogations in the TRUST trial.¹⁹ Arrhythmia management may be facilitated. In the multi-centre ECOST trial, clinical reactions enabled by early detection resulted in a large reduction in the number of actually delivered shocks (−72%), the number of charged shocks (−76%), the rate of inappropriate shocks (−52%), and at the same time exerting a favourable impact on battery longevity (also noted in TRUST).^14,26 This was confirmed in meta-analysis²⁷ although more recent studies suggested lesser effects.^28–30 This finding may be related to contemporary better ICD programming, although adoption of recommended settings is surprisingly deficient.³¹ Repeated adenosine triphosphate (ATP) delivery may herald a preventable shock^32,33 and, if notified early, presents an opportunity for intervention (but not all RM platforms alert for this³⁴). Inappropriate shocks related to ICD lead fractures may be reduced by RM.³⁵

The ability to collect detailed device-specific data, with component function assessed daily, sets a precedent for longitudinal evaluation of lead and generator performance. Artificial intelligence (AI) may improve prediction of significant events, e.g. future VT storm and ICD shocks,^36,37 and detection of lead failure³⁸ by a combination of two or more parameter deviations. Although TRUST showed that a core set of alerts generated a minimal load,³⁹ AI may be especially important for managing the large load of false positives associated with implantable loop recorder (ILR) management.⁴⁰

The results have led to professional societies strongly recommending RM in preference to conventional in-person evaluation for CIED follow-up over the last decade.^{5,6,26,41–43} Automatic continuous RM represents an early warning system that can provide assurance to both patients and their physicians.

Disease management: atrial fibrillation

Atrial fibrillation detection

Remote monitoring of patients with CIEDs facilitates the early AF detection and quantification and calculates arrhythmia burden.^19,44–46 That is particularly meaningful for asymptomatic (‘silent’) AF and for patients without any prior AF history, 30% of whom develop newly detected AF.^47–49 In cryptogenic stroke patients, AF detection rate by implantable cardiac monitors ranges from 15% to 30% and is a function of monitoring duration.⁵⁰ Detection of AF occurs 1–5 months earlier with RM vs. conventional follow-up. Remote monitoring has a sensitivity of nearly 95% for true AF detection,⁴⁶ and approximately 90% of AF episodes that trigger alerts are asymptomatic. Early AF detection promises earlier treatment, potentially preventing adverse events such as stroke, shock therapy, and heart failure (HF).⁵¹ Remote monitoring can enable tailored and dynamic patient treatment pathways^52–54 and assess success rates of AF therapies.

Atrial fibrillation burden and atrial fibrillation progression

Remote monitoring provides a chronologic plot of all AF events and their individual durations, accompanying intra-cardiac electrograms, AF burden trends, and associated ventricular rates. Insights into temporal patterns may be instructive; e.g. paroxysmal AF comprises heterogeneous entities.⁵⁵ Thus, in self-terminating AF, the AF burden is low and determined by number of co-morbidities. Atrial fibrillation progression occurred in approximately 25% of patients.⁵⁶ Patients with either predominant nocturnal or daytime onset of AF episodes had less associated co-morbidities and less AF progression compared with that of patients with mixed onset of AF.⁵⁷

Atrial fibrillation alert setting should be modified during follow-up for individual patients according to their individual clinical profile in order to identify clinically meaningful events (otherwise daily alerts for single short episodes may increase clinic work burden).⁵⁸ In patients with permanent AF, all alerts should be switched off and only ventricular rate should be monitored.

Risk stratification

Continuous atrial rhythm monitoring may predict AF onset and, in combination with clinical data and risk scores, may improve patient risk stratification for major clinical events. The daily number of premature atrial complexes significantly increased in the 5 days preceding the AF onset.⁵⁹ In a CIED population without previous clinical AF, atrial high rate episode (AHRE) incidence increased with a higher CHA₂DS₂-VASc score. The association was stronger with longer episodes, even if the accuracy of CHA₂DS₂-VASc as a predictor was moderate.⁶⁰ In ICD patients, AHREs were an independent predictor of sustained ventricular arrhythmias at 30 days and that patients with ventricular arrhythmias preceded by AHRE had a worse long-term survival.⁶¹ Remote monitoring enabled reduction in AF-related inappropriate ICD shocks in the ECOST randomized trial.²⁶ Mean ventricular heart rate monitoring⁶² may further improve inappropriate shock prevention.

Stroke prevention

Early detection of AF creates a unique opportunity for pre-emptive intervention aimed at preventing stroke, particularly in asymptomatic patients. However, evidence is scarce. Monte Carlo simulations based on a real population⁶³ suggested that daily monitoring could reduce the 2-year stroke risk by 18%. A sub-analysis of the randomized COMPAS trial⁴² showed a significant reduction of hospitalizations for atrial arrhythmias and related stroke in the RM group. In the HomeGuide Registry⁶⁴ (1650 patients), stroke incidence at 4 years was extremely low, lower than that expected for the estimated thromboembolic risk profile of the enrolled population. However, the randomized IMPACT trial⁶⁵ testing oral anticoagulation therapy for AF guided by RM reported no difference in stroke or all-cause mortality in the RM-guided intervention group. In fact, since the study protocol called for discontinuation of oral anticoagulation if there was no AF for 30 or 90 days, according to individual risk profile, the majority of patients were not anticoagulated at the index event. Furthermore, a proximate temporal relationship of device-detected atrial arrhythmias to the occurrence of strokes was lacking; i.e. the mechanism of stroke is not solely related to the AF episodes. The safety of discontinuation of anticoagulation in any sub-groups of AF patients has to be considered carefully.

Recently, a meta-analysis of 16 randomized trials, including studies enrolling CIED patients and studies on wearable devices, demonstrated that, compared with patients receiving standard follow-up, patients undergoing RM had a significantly higher detection rate of atrial arrhythmia and a lower risk of stroke. The lower risk of stroke was only noted in patients with CIEDs [risk ratio (RR) 0.513; 95% confidence interval (CI) 0.265–0.996].⁶⁶ Future results from the randomized controlled NOAH-AFNET⁶⁷ and ARTESiA⁶⁸ trials may provide guidance of anticoagulation strategy in high-risk CIED patients with AHRE for stroke prevention.

Atrial fibrillation and heart failure

A strong association has been observed between AF and hard clinical endpoints in patients with HF.^69,70 Unrecognized AF may exacerbate HF, increase frequency of hospitalizations, lead to loss of cardiac resynchronization therapy (CRT) and inappropriate ICD shocks, increase sympathetic tone, and induce haemodynamic compromise and thromboembolism. Remote monitoring has the potential of impacting patient outcome by early detection and continuous monitoring of AF. Evidence of these benefits has yet to be confirmed by randomized studies.

Disease management: heart failure

There are several means by which RM of CIEDs may improve HF management.^71,72 First, continuous recording of device-derived variables can enable alerts on imminent fluid decompensation. Since volume overload decompensations are typically preceded by a gradual fluid accumulation over several weeks prior to the acute clinical event, RM during this period may provide early warning to enable timely pre-emptive interventions and avoid harmful clinical events. Second, regular evaluation of multiple diagnostic parameters may add valuable information to the clinical assessment of ambulatory HF patients⁷³ and, thus, facilitate the triage into different risk categories for individualized follow-up.^74,75 Third, by enabling patients to interpret and respond to their own physiological data, RM offers a tool to enhance patient empowerment and self-care.^76,77

Several device-based sensors obtain information potentially relevant to HF management, including intra-thoracic impedance,^78–80 heart rate and heart rate variability,⁸¹ physical activity,⁸² minute ventilation,⁸³ heart sound amplitude⁸⁴ in addition to arrhythmia detection, and statistics on the percentage of biventricular stimulation in CRT patients. Single use of these sensors, however, is associated with sub-optimal sensitivity and specificity to predict HF decompensation, as shown in the example of impedance-based fluid detection.⁸⁵ However, Boehmer et al.⁸⁶ demonstrated that a multi-sensor algorithm combining heart rate, heart sounds, thoracic impedance, respiration, and activity can markedly improve diagnostic accuracy with a sensitivity of 70% to predict HF events, which is balanced against an unexplained alert rate of 1.47 per patient and year.

More recently, D'Onofrio et al.⁸⁷ developed and validated a longitudinal predictor for worsening HF events by combining information from seven different sensors obtained by RM. The Seattle HF model was used to risk-stratify patients at baseline. The algorithm showed promising sensitivity of predicting HF events by 65.5% (CI 45.7–82.1%) at a median alerting time of 42 days [interquartile range (IQR) 21–89], and a false (or unexplained) alert rate of 0.69 (CI 0.64–0.74) or 0.63 (CI 0.58–0.68). The clinical effectiveness of these algorithms remains to be confirmed in prospective outcome trials.

Remote monitoring and heart failure outcomes

Despite early promise from real-world data,^29,88 randomized controlled trials (RCTs) and meta-analyses evaluating the effect of RM on survival have been largely neutral^27,89 (although associated with a marked reduction in planned hospital visits and lower costs⁹⁰). In MORE-CARE, randomizing 865 CRT with defibrillator (CRT-D) patients, the composite of death, cardiovascular, and device-related hospitalization was similar between groups [hazard ratio (HR) 1.02, P = 0.89].⁹¹ Likewise, the OptiLink trial, including 1002 patients with ICD or CRT-D, failed to show an improvement of all-cause death and cardiovascular hospitalization (HR 0.87, P = 0.13).⁹² Remote management of heart failure, including 1650 patients,⁹³ did not find a difference in the composite of death and unplanned cardiovascular hospitalization (HR 1.01, P = 0.87). By confirming the non-inferiority of RM to routine in-clinic visits in terms of clinical outcomes, a recent study provided evidence that routine in-person follow-up can be safely substituted by RM.²⁸ Nevertheless, such a follow-up strategy does not necessarily improve patient-reported health status and ICD acceptance.⁹⁴

Interestingly, the IN-TIME trial, using daily multi-parameter telemonitoring in ICD patients, marks an impressive exception.⁹⁵ The primary endpoint, a composite clinical score combining all-cause death, overnight HF admissions, changes in New York Heart Association (NYHA) class, and in-patient global self-assessment, was improved by telemonitoring (odds ratio 0.63, 95% CI 0.43–0.90). Furthermore, a reduction in the secondary endpoint of all-cause mortality (HR 0.36, 95% CI 0.17–0.74, P = 0.004) and cardiovascular mortality (HR 0.37, 95% CI 0.16–0.83, P = 0.012) was found. These results were largely driven by the prevention of worsening HF events⁹⁶ suggesting that patients with more severe HF may gain a greater clinical benefit of RM.⁹⁷ Similarly, HF management based on home transmission of haemodynamic data from a pulmonary artery pressure sensor significantly decreased hospital admission rates for HF (HR 0.64, 95% CI 0.55–0.75, P < 0.0001) and also in a sub-group of CRT non-responders with severe HF.⁹⁸ Common to these positive studies was a RM strategy employing daily transfer of data. While a larger trial failed to reproduce effects on mortality and HF events,⁹⁹ these results were likely compromised by the ongoing COVID-19 pandemic.¹⁰⁰ In the future, pressure monitoring may possibly be used to fine-tune the programming of other HF devices.¹⁰¹

Workflow, resource consumption, and patient satisfaction

Workflow

Incorporating CIED RM in standard clinical practice is challenging and requires dedicated and structured workflow models and infrastructures. This is necessary to capitalize on the early detection (‘same day') power of continuous monitoring (Figure 4). An integrated working group, comprising allied professionals (nurses and technicians), physicians, and ancillary staff with well-defined roles and responsibilities and dedicated time, space, and equipment, is necessary.¹⁰² Allied professionals are mainly responsible for continuity of care. More specifically, their duties include patient training and education, website data entry, remote data review, data screening, critical case submission to the physician, contacting patients, and checking patient compliance and therapy benefits.¹⁰³ They need to be experts and continuously updated on CIED technology and ideally have the European Heart Rhythm Association (EHRA) or International Board of Heart Rhythm Examiners (IBHRE) certification of competence (or equivalent). The physicians (usually an expert electrophysiologist) are responsible for informed consent, overview and monitoring of the full process, and patient clinical management. The ancillary staff may assist the clinical staff in scheduling appointments, checking patient connectivity, communicating results, billing, and update of patient electronic files. The RM working group should interact with other instances, in particular with HF and emergency care specialists, general practitioners, service providers, and referral hospitals. This model has been found to be highly effective, safe, and efficient in the Italian HomeGuide Registry^46,104 and has been recommended in the 2015 HRS Consensus Document on RM⁸ and in the 2020 position paper of the Italian Association of Arrhythmology and Cardiac Pacing (AIAC).¹⁰⁵

Figure 4.

Days to detection of ICD problems in patients assigned to remote home monitoring. Overall, 22/43 (51%) were detected within 24 h (compiled with permission from Varma et al.²²).

For small centres without dedicated on-site personnel, a model based on a collaborative network between neighbouring structures (hub and spoke model) has been suggested, in which the hub centre first receives and processes all the transmissions. In case of meaningful alerts, the data are sent to the spoke centre for patient recall and clinical management.¹⁰⁶

For hospitals without the necessary staff, third-party resources may help with RM tasks.^107,108 This strategy may have clinical, financial and workflow benefits, but it may induce the patient’s perception of losing the human relationship with the clinical staff with decreased compliance and satisfaction.

In-person clinical evaluation is recommended generally at least once yearly but may be extended to once every 2 years or more for patients without or with mild structural heart disease.^109–111 An alert-based only follow-up strategy could be considered in the future.²⁰

Resource consumption

Pursuing the strategy to offer RM to all patients leads to an exponential increase of transmissions and clinic workload. Unscheduled transmissions may increase the work burden by 30–40%.⁴⁰ Transmission review is usually fast, ranging from 2 to 5 min according to manufacturer website and data content, but complete patient remote management includes technology and connectivity troubleshooting, diagnosis, patient communication, clinical action, electronic charting, scheduling, and billing. Associated time consumption is usually low for pacemakers but increases exponentially for loop recorders.^112,113 According to available data, an estimated 30–40 h a week are required to manage 1000 patients on RM.

In 2016, Maines et al.¹¹⁴ started the routine implementation of the RM service for all patients by applying the workflow model suggested by the AIAC.¹⁰⁵ After 2 years, 93% of patients were remotely monitored. The transmission rates were as follows: 5.3/patient-year for pacemakers, 6.0/patient-year for defibrillators, and 14.1/patient-year for loop recorders. Only 21% of transmissions were submitted by the nurse to the physician for further clinical evaluation, and 3% of transmissions necessitated an unplanned in-hospital visit for further assessment. Clinical events of any type were detected in 39% of transmissions. Overall, the nurse total workload was 1.95 full-time equivalent, which resulted in 1038 patients/nurse. The total workload for physicians was 0.29 full-time equivalent.

The impact on resource consumption of scheduled vs. alert transmissions was further evaluated.¹¹⁵ Scheduled transmissions generated 67% of remote data reviews for pacemakers and ICDs, but their ability to detect clinically relevant events was extremely low. On the contrary, 24% of the alert transmissions required clinical discussion, and 3% necessitated in-hospital visits. The proportion of alerts judged clinically meaningful was 7%.

Patient satisfaction

Patient acceptance of RM is variable. In some studies, no difference was observed in quality of life and patient satisfaction when comparing RM with a conventional follow-up strategy.⁴² Others have shown a high patient satisfaction rate, ease of use, and good compliance with the use of RM systems.^116,117 Patients also reported a positive impact on perceived safety and general health and rarely complained of the lack of interaction with healthcare professionals.¹¹⁴ The organizational model may be critical to improve patient satisfaction. Remote monitoring has been demonstrated to be well accepted even for elderly people and for patient with a low level of scholarity.^118,119 Automaticity and reliability of remote technology are important. In the TRUST trial,²¹ transmission failure occurred in less than 1%. No patient assigned to Home Monitoring (HM) crossed over during the study and 98% elected to retain this follow-up mode on trial conclusion, indicating patient acceptance and confidence in follow-up with this technology. In striking contrast, with conventional management, patient attrition was significantly higher. Introduction of smartphone-based RM using patient applications has been demonstrated to further improve patient compliance and connectivity compared with traditional bedside monitor RM.¹²⁰

Hurdles to remote management adoption

In addition to the challenges with workflow and consumption of resources resulting from managing alerts¹²¹ (covered in the previous section), there may be concerns for data protection and for legal exposure. Finally, there may also be no financial incentive to adopt remote device management due to lack of reimbursement.

The need for guidance on legal aspects of remote device management has been recognized for years.^5,122,123 In 2020, EHRA and the European Society of Cardiology (ESC) Regulatory Affairs Committee published a joint task force report on legal requirements for RM, in light of the European Union General Data Protection Regulation (GDPR), which has been implemented since 2018.¹²⁴ The main objective of the GDPR is to ensure ‘the protection of natural persons with regard to the processing of personal data and on the free movement of such data’. The processing of personal data (which includes health data) should be undertaken only with specific consent, and individuals or organizations collecting and retaining data are accountable.

The ‘data controller’ is defined as the person or body, which determines the purposes and means of processing personal data. As manufacturers determine the ‘means’ for remote device management with the technology used to collect data, they are considered to be data controllers and are responsible for providing access to the personal data collected and should offer the possibility to delete them. The healthcare facility or hospital is considered along with the manufacturer as a joint data controller, since they determine the clinical indications for collecting data remotely, and which data are collected and analysed. Sometimes this may be done together with the physician if the healthcare professional is self-employed. Otherwise, a hospital takes legal responsibility as controller for staff whom it employs. The healthcare facility should make a contractual agreement with the manufacturer regarding the roles and responsibilities of each party, taking into account the specific rules that apply when transferring personal data outside the European Union (where the servers of some of the device manufacturers are located). Patients are granted the right to obtain information from the data controller about the nature and use of the data stored, and they have the right of access to their own data. With certain exceptions, they have the right to be forgotten and to have their data erased. The ‘data processor’ handles personal data on behalf of the controller and in accordance with any limitations established by the controller. Third-party providers (‘middleware’ companies) are data processors. A legal contract should be made also between the data controller and a third-party provider.

Cybersecurity is another concern and manufacturers should ensure adequate encryption of transmitted data.^124,125

Physicians may be concerned about legal liability for failure for timely response to remote alerts to prevent adverse events.^13,126 It is therefore important that signed patient informed consent forms outline the services provided by the healthcare facility (e.g. evaluation of alerts during working hours only), as well as the limitations of the service. An aspect that also needs to be considered is the liability incurred by not providing RM in patients in whom this is indicated according to guidelines (e.g. in patients with devices under recall¹¹¹).

Reimbursement is a major factor that affects uptake of remote device management worldwide, which remains heterogeneous.^127–133 Evidence regarding the economic benefit of RM is growing, which should hopefully convince more payers to cover this service.^90,134–138 Interestingly, the COVID-19 pandemic sparked some payers to introduce reimbursement for remote device management.¹³⁹

Transition to digital health

The growing availability of smartphones and application to health queries has resulted in several smartphone applications focused on CIED patient engagement and interpretation of the data.⁵² Some can already replace conventional RM and CIED interrogations. For this purpose, one study showed that 250 patients achieved a higher rate of completing transmissions vs. historical control subjects who used traditional or wireless bedside monitors.¹⁴⁰ Younger patients may be more receptive.¹⁴¹ The systems are generally safe. Risk of electromagnetic interference between modern smartphones and contemporary CIEDs is low^142–144; however, patients should be instructed to avoid direct contact between their mobile phone and their CIED.

Incoming data from RM databases may be processed by AI. For example, algorithms to optimize scanning of all transmissions to identify CIED patients with AF enhanced AF detection sensitivity by 10% and improved oral anticoagulation optimal treatment by 6%.¹⁴⁵ An AI tool to filter AF alerts resulted in an 84% reduction in notification workload, while preserving patient safety.¹⁴⁶ Interestingly, neural network modelling of remotely acquired device-based data helped to identify patients at risk of malignant ventricular arrhythmias and appropriate device therapies within 30 days.³⁷

Hence, the application of patient smartphone applications to encourage patient compliance with RM by providing information on communicator setup, troubleshooting, and connection status of the communicator^120,147 can be an efficient way to ensure continuity of patient monitoring while limiting workload at the centres. Barriers occur at the time of activating RM and keeping patients monitored, both of which are dependent on patient factors. Despite the general increase in digital literacy, there are still patient sub-groups with limited confidence in digital technology, requiring dedicated education in the use of smartphone apps and other digital devices.³⁷ A structured patient pathway may facilitate engagement of patients with digital health technology.⁵² Personalized and fully automated RM device setup programmes are designed to educate people about technology, while simultaneously helping reduce the ‘digital divide’ and relieving the load on physician and clinic staff as well.

Conclusion

Remote monitoring of CIED function has become the standard of care and professional society recommendations have been updated frequently reflecting the vigorous development in technology and applications and the need for patient education.^4–6,8 The most recent moves to surveillance with continuous RM and clinic evaluation only when indicated by personalized RM notifications, representing a sea change in the standard of care from a decade ago, i.e. structured follow-up based on regular intermittent appointments. Clinically stable patients may be seen in office only every 24 months and perhaps longer (Figure 5). This model of asynchronous care in turn has vast implications for workflow, reimbursement, and data management reflecting entry into the digital age.^20,146,148

Figure 5.

Conventional follow-up based on intermittent calendar-based care is likely to be replaced by digitally driven alert based care (left) to identify subjects needing care promptly and reducing the non-actionable work performed by clinics, thus easing workflow (compiled with permission from Varma et al.²⁰).

Funding

None declared.

Data availability

No primary data were analyzed for this manuscript which reviewed existing literature.