The population of this single-center, retrospective, observational study included consecutive patients receiving CRT-D devices in the Medical Centre of the Hungarian Defence Forces (Budapest, Hungary) from January 2011 to June 2016. Indication for implantation was established according to current guideline recommendations of the European Society of Cardiology [16,17]. CRT-ICDs from various manufacturers were used (Biotronik, Germany; ELA/Sorin, Italy; Guidant/Boston Scientific, Marlborough, MA; Medtronic, Minneapolis, MN; and St Jude Medical/Abbott, St Paul, MN). Choice of device type was left to the implanting physician’s discretion.

The possibility of remote monitoring was offered to every patient implanted with a wireless telemetry-capable CRT-D device. Patients who consented to remote monitoring formed the remote monitoring group. During the inclusion period, remote monitoring systems of two manufacturers (Medtronic CareLink Network, Medtronic Inc, Minneapolis, MN; and Home Monitoring Service Center, Biotronik GmbH & Co KG, Germany) were available in our center. The control group consisted of CRT-D recipients, who were followed up in our outpatient device clinic without remote monitoring.

The CareLink network operates with scheduled transmissions defined by the physician, and unscheduled transmissions, which can be triggered both by the patient (manual transmission) and by the device itself (alert event). Scheduled remote transmissions were set up every 3 months [18]. It was also recommended to patients to send manual transmissions in case of palpitation, syncope, or worsening of heart failure symptoms. Alert programming was set up according to previously published Medtronic-sponsored trials as follows: OptiVol alert (nominal fluid index ≥60 Ω-day), daily atrial fibrillation burden greater than 6 hours per day, ventricular rate during atrial fibrillation greater than 100 bpm for 6 hours, two or more shocks delivered, all therapies exhausted, lead or device integrity alert, lead impedance out of range, recommended replacement time, and end of service [7,18,19].

Home Monitoring uses a mobile phone network to transmit device data automatically on a daily basis, as well as instantly on the occurrence of a potentially clinically relevant event [20]. These parameters include device and battery status, pacing impedances, bradycardia, tachycardia, and CRT statistics, mean heart rates, patient activity, heart rate variability, and current programming of the device [21].

In-office visits were recommended to patients on remote monitoring without symptoms at least once a year [22]. Remote transmissions were evaluated every day by a team consisting of cardiology trainees and consultant electrophysiologists. In the case of suspected heart failure progression, a heart failure specialist was involved additionally. Transmissions were labeled as clinically nonsignificant, clinically relevant (yellow alert), or highly urgent (red alert). If a clinically relevant event was perceived, patients were contacted via phone calls and were invited to the clinic for a personal visit in a week. Definitions of a clinically relevant event for the two remote monitoring systems are summarized in Multimedia Appendix 1. Transmissions with highly urgent content defined as ventricular arrhythmias treated with more than one ICD shock or system integrity alert were handled within 24 hours. Patients with missed transmissions longer than 4 weeks were also contacted. Follow-up of patients in the control group was performed at intervals of 3 to 6 months.

Patient demographics, comorbidities, pharmacotherapy, electrocardiogram characteristics, echocardiography, and laboratory data were collected at enrollment and during scheduled remote checks or in-office visits.

All patients on remote monitoring signed a related informed consent form. The study complied with the Declaration of Helsinki, and the study protocol was approved by the Institutional Ethics Committee of Hungarian Defence Forces Medical Centre, Budapest, Hungary.

The primary objective of this study was to compare the mortality of remote-monitored patients with patients on conventional follow-up. Survival was assessed as the time from CRT-D implantation to all-cause mortality. Mortality data were retrieved using the Hungarian National Health Fund Death Registry. The unique health insurance number of a patient is deactivated immediately after death. The secondary endpoint was the response to resynchronization therapy at the visit at 6 to 12 months, defined as 5% absolute increase in left ventricular ejection fraction. The number of all ambulant visits, device clinic visits, and heart failure outpatient clinic visits were also analyzed and compared between the two patient groups.

Statistical analysis was performed using PASW Statistics software, version 18.0.0 (WinWrap Basic, Polar Engineering and Consulting). The Kolmogorov-Smirnov test was used to evaluate the normal distribution of continuous data. The chi-square test was applied to test for categorical variables; the two-sample t test or the Mann-Whitney U test was used for continuous variables among patient groups.

To assess the effects of remote monitoring on survival, the Cox proportional hazards regression model was used. Univariate analysis was performed for the following variables: age; gender; heart failure management; upgrade procedure; secondary prevention; ischemic etiology; atrial fibrillation; hypertension; hyperlipidemia; diabetes mellitus; stroke; peripheral artery disease; chronic obstructive pulmonary disease; New York Heart Association (NYHA) functional class; left ventricular ejection fraction; QRS duration; left bundle branch block; estimated glomerular filtration rate; hemoglobin; and therapy with platelet aggregation inhibitors, beta blockers, angiotensin-converting-enzyme inhibitors or angiotensin-receptor blockers, mineralocorticoid antagonists, diuretics, digitalis, amiodarone, and statin. All variables with P ≤.10 on univariate analysis were included in the multivariate Cox models. Two-sided P values <.05 were considered statistically significant. Survival curves were constructed according to the Kaplan-Meier method and compared with the Cox proportional hazard model. To check for interaction between survival and the specific remote monitoring system, all-cause mortality was also compared between the subgroups of patients on CareLink and on Home Monitoring systems.

In this single-center, retrospective, observational study of 231 CRT-D recipients, use of remote monitoring was associated with better survival compared with patients undergoing conventional follow-up. The total number of follow-up visits was not reduced with this technique in our cohort.

Remote monitoring systems proved to be feasible, reliable, accessible, and were supported by the current guideline recommendations [16,23]. They are still underutilized despite the clear advantages, such as early identification of device malfunction or arrhythmic events. Regarding survival benefit, the available clinical data are controversial [10,13,24-27].

There are several proposed mechanisms contributing to the improved clinical outcome: early detection of clinically relevant arrhythmias (ie, atrial fibrillation or ventricular tachycardia) and early recognition of device malfunctions or suboptimal programming, which avoids unnecessary shocks and achieves proper percentage of biventricular pacing, respectively.

Moreover, the number of OptiVol alerts and the related visits at the heart failure outpatient clinic suggest that the observed mortality benefit associated with remote monitoring was also driven by early response to cardiac decompensation in our study. OptiVol is a detection algorithm developed for early recognition of cardiac decompensation using changes of intrathoracic impedance as a marker of lung fluid status [28]. This method, used for the detection of cardiac decompensation, is considered to be a very sensitive but less specific tool, which leads to a high number of false positive alerts. We have previously described a refined device diagnostic algorithm based on parameters of low activity level, high nocturnal heart rate, and suboptimal biventricular pacing, which could significantly improve clinical reliability of OptiVol alerts [9]. However, a recent OptiLink heart failure study analyzing this technology failed to demonstrate any difference in the composite endpoint of all-cause mortality and cardiovascular hospitalization [29]. Notably, only 30.3% of intrathoracic fluid index threshold crossing led to medical action, and it led to altered medication in only 26.0% of patients.

Our analysis shows a survival benefit for patients on remote monitoring; therefore, we are convinced that remote monitoring should be part of the follow-up of patients with CRT-D devices. One possible reason for the conflicting results of previous studies is the wide variety of actions on remote monitoring findings. A recently published meta-analysis by Stockburger et al [30] showed that device-based remote monitoring strategies specifying close-meshed comprehensive data acquisition and defined treatment interventions are able to significantly reduce total mortality and cardiovascular mortality, whereas remote data acquisition alone without specified treatment interventions appears to be ineffective on hard endpoints. The significantly increased number of visits in the heart failure clinic in our study supports this hypothesis, as the demonstrated survival benefit could have been achieved by more frequent follow-up and focused treatment on the high-risk patients.

Additional clinical factors can also modify the clinical benefit of remote monitoring, such as the time from implantation to initiation of remote monitoring [31], the adherence to this technique [11], or the frequency of data transmission. Two recently published meta-analyses demonstrated that significant mortality benefit was only seen in the subset of trials using a daily transmission verification (ie, Biotronik Home Monitoring System) [26,27]. A possible explanation for this difference is that the rate of events recognized within 24 hours is the highest among manufacturers with the Biotronik Home Monitoring system [32]. However, the type of remote monitoring system did not influence the survival benefit in our patient cohort.

This research is a single-center, retrospective study with all the consequential limitations. Despite the adjustment for the most relevant baseline cofounders, a residual selection bias can not be completely excluded. Moreover, the choice of device was not randomized but was left to the implanting physician, which may have modified our results. The patient’s decision to consent to remote monitoring may also have modified the results. The potential sources of bias should also be addressed. The most important one is the patient’s decision to consent with a remote monitoring program. Patients with better adherence and motivation are more likely to participate in such programs, which might have improved the outcomes of patients in the remote monitoring group.

In this single-center, retrospective study of optimally treated heart failure patients undergoing CRT-D implantation, the use of remote monitoring systems was associated with significantly better survival. However, a higher number of follow-up visits in the heart failure outpatient clinic was needed, which suggests that this survival benefit could be achieved by an increased and focused effort to follow and treat high-risk patients. Our results call for further randomized studies with a standardized action plan after certain telemonitoring observations to define the optimal role of this technology in the follow-up of heart failure patients.