Remote monitoring for heart failure management during COVID-19 pandemic

Graphical abstract


Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has broken out, resulting into a global and rapid diffusion of the coronavirus disease 2019 (COVID- 19), leading to excess mortality [1]. To limit viral spreading, governments issued restrictive measures, including forced homestay (i.e., lockdown), largely affecting people's lifestyle and behaviors [2,3]. Preliminary studies identified higher-risk subgroups, with patients suffering from heart conditions (e.g., heart failure [HF]) displaying a tendency towards worse outcomes, including severe or critical disease and mortality [4,5].
Physical inactivity is acknowledged as a key modifiable risk factor for cardiovascular diseases and in particular for HF exacerbation [6]. In addition, beyond its detrimental effects on exercise, lockdown may increase anxiety and/or depression levels, which in turn are known to reduce physical activity in a chronic HF population [7]. An overview on physical activity reduction during the lockdown can be derived from patients implanted with a cardiac implantable electronic device (CIED). CIEDs are recommended for HF patients with severe systolic dysfunction to prevent sudden cardiac death (implantable cardioverter defibrillator, ICD) and/or to improve cardiac contractile function by resynchronizing left and right ventricles (cardiac resynchronization therapy, CRT) [8]. Advances in health technologies allow for an accurate remote monitoring of HF patients' vital signs and activity indexes by collecting and forwarding data to the reference medical center [9]. This strategy is useful to early identify worsening HF patients and to optimize their pharmacotherapy to prevent decompensated HF recurrencies and hospitalizations [10][11][12]. Despite the need for human resources, telemedicine visits (phone call, videoconference, email) for HF patients during pandemic are proven to be effective in the reduction of adverse clinical outcomes [12]. Whether and to what extent HF patients' physical activity varies as a response to lockdown and pandemic is still unclear. In addition, the feasibility of a remote monitoringguided approach to provide medical support, reassurance and clinical benefits to HF patients is so far unexplored, now more than ever representing an unmet need.
The purposes of our study are as follows: firstly, to investigate the direct impact of COVID-19 pandemic on HF patients' physical activity levels and vital signs; secondly, to determine the contribution of the indirect effect of the lockdown (i.e., anxiety and depression) on HF patients' health status; thirdly, to explore whether telemedicine and remote monitoring can play a role in the management of chronic HF patients amidst circumstances discouraging outpatient visits.

Study population
Patients were considered eligible for the inclusion in our study if they met the study inclusion criteria: (1) Age 18 years old; (2) Chronic HF with reduced left ventricular ejection fraction (HFrEF); (3) Implantation with a CIED (ICD or CRT) and periodic outpatient follow-up at our facility; (4) Active Biotronik Home Monitoring (Biotronik SE&Co. KG, Berlin, Germany) providing daily highquality data on electrical parameters, vital signs and physical activity levels.
To minimize heterogeneity and to select daily high-quality remote monitoring data on the variables of interest, the inclusion in the current study was limited to patients with active Home Monitoring. Patients with low remote monitoring transmission rates (defined as <20% of programmed transmissions) were excluded from our analysis.

Study design
Demographic data, clinical indication to CIED implantation and remote monitoring records were extracted from the Home Monitoring system and subsequently entered an interim database for data sorting and aggregation. To investigate the effects of lockdown and to detect potential variations across the study period, individual patient data from 16 weeks at the turn of the lockdown date were collected and categorized into two groups: (1) lockdown period (from March 9th to May 3rd, as of Italy government decree) [13]; (2) control period (from January 18th to March 8th, defined as a comparable timeframe before restrictive measures were issued). During the lockdown period, people could not leave their home except for serious reasons, including health problems, work purposes and necessities.
For the purposes of this analysis, individual patient data were aggregated and summarized to derive two central tendency measures for each variable (i.e., median values during the control and lockdown periods). These figures were finally collected into the final database for between-periods comparisons. To explore the variability of the parameters of interest within the lockdown period, a week-to-week analysis was performed, selecting three representative weeks for comparisons, i.e. week 7 (before lockdown, from February 24th to March 1st), week 11 (during lockdown, from March 23rd to March 29th) and week 16 (final lockdown, from April 27th to May 3rd).
Furthermore, to investigate the network among forced lockdown, physical activity variations and patients' mindset, study par-ticipants were asked to undergo a phone-call interview, during which they were asked to rate their physical activity levels before lockdown (from 0 [complete inactivity] to 4 [athletic lifestyle]) and to disclose whether a decline (mild, moderate or severe) was noticed during the lockdown period. As part of the interview, study participants were also asked information about their compliance to cardiovascular pharmacotherapy and were administered two specific and broadly validated questionnaires (i.e., the Zung selfrating depression and anxiety scales) to assess their psychological attitude towards pandemic and lockdown [14].

Endpoint definitions
The primary endpoint of our study was the variation between control and lockdown periods in terms of physical activity levels. Physical activity was assessed by remote monitoring as the percentage time during which any type of exercise was detected with regards to the whole monitored daily time (24 hours).
Secondary outcomes of interest included the following: variation in the mean heart rate between control and lockdown periods; variation in thoracic impedance between control and lockdown periods; patients' perceptions on physical activity across the lockdown period and their matching with observed modifications; increase in anxiety and/or depression levels during the lockdown, assessed by the Zung self-rating questionnaires [14].

Statistical analysis
The distribution of each variable of interest was explored using the Kolmogorov-Smirnov test. Continuous variables were summarized as means with standard deviation in case of normal distribution, or medians with interquartile range if the Kolmogorov-Smirnov test showed skewed data. To detect mean ranks differences among repeat measurements of non-Gaussian variables in a single sample (paired data), the Wilcoxon signed-rank test was used [15]. Similarly, to ensure the identification of differences across multiple test attempts, the week-to-week analysis availed from the Friedman test [16]. Correlation between variables of interest was explored with the Spearman's rank correlation test [17]. All p values were based on two-sided tests. A p value less than 0.05 was considered to be statistically significant for all analyses. For multiple comparisons, the Bonferroni correction was applied [18]. All the analyses were performed using the Statistical Package for Social Sciences (SPSS Inc., Chicago, Illinois), version 25.

Study population
Patients' screening and selection process was conducted as showed in the study flow-chart (Fig. 1). Overall 2225 consecutive patients implanted with a CIED and followed-up at Policlinico ''G. Rodolico -S. Marco" (Catania, Italy) were initially screened for the inclusion in the current study. Patients without active remote monitoring were excluded, thus restricting our population to 388 subjects. To minimize heterogeneity and to select patients with daily high-quality data transmission, the inclusion in the current study was restricted to 261 patients with active 1Home Monitoring (Biotronik SE&Co. KG, Berlin, Germany). To keep consistency in device characteristics, remote monitoring parameters and clinical indication behind CIED implantation, we further excluded 2 patients implanted with a pacemaker, 26 with an implantable loop recorder and 5 with ICD or CRT for secondary prevention purposes (i.e., patients without HF diagnosis). In addition, 17 patients were excluded due to low remote monitoring transmission rate (defined as <20% of the programmed transmissions). As a result, 211 HF patients with daily high-quality remote monitoring data were deemed to be eligible for the inclusion in the current analysis, therefore representing the study population.

Baseline characteristics
Study participants displayed a mean age of 68.6 ± 9.5 years and were predominantly male (N = 161, 76.3%). A mean time of 36 ± 24 months has elapsed since device implantation.
With respect to CIED type, we can distinguish different subgroups: 86 patients (40.8%) were implanted with a CRT device, while 12 (5.7%) and 113 (53.5%) with a dual-and single-chamber ICD, respectively. Among single-chamber ICD patients, 28 patients (24.8%) have a ventricular lead with DX technology, therefore enabling atrial sensing through a floating dipole into the right atrium [19].

Patient-reported outcomes
The phone-call interview was accepted and successfully completed by 125 (59.2%) study participants ( Table 2). Two out of three included subjects (66.9%) reported a moderate-to-high physical activity level at baseline. In addition, a reduction in exercise levels during the lockdown was perceived by a large proportion of After the administration of the Zung self-rating questionnaire for anxiety, a mean value of 34.2 ± 7.8 was found (normal anxiety level). The majority of study participants (N = 114; 91.2%) reported normal anxiety levels, whereas 11 subjects (8.8%) scored for a level of mild anxiety; no patient disclosed moderate to high anxiety levels. No correlation was found between the anxiety score and variations in physical activity (evaluated as difference between control and lockdown periods), mean heart rate, mean resting heart rate and thoracic impedance.
Likewise, the median depression score by the Zung self-rating questionnaire for was 32.0 (interquartile range 28.0-38.5). While 115 participants (92.0%) scored in the normal range, 9 (7.2%) and 1 (0.8%) patient could be considered to suffer from mild or moderate depression, respectively. Severe depression was reported by none of the participants. Similar to anxiety, no correlation was found between depression levels and other study variables.
The vast majority of the study participants (71.0%) reported an optimal compliance to prescribed pharmacotherapy (i.e., no missed doses for any of the medications).

Discussion
The main findings of the current study are as follows. Firstly, patients' physical activity levels significantly decreased during the lockdown period as compared to the control period. Secondly, small reductions in mean heart rate, mean resting heart rate and thoracic impedance were noted during the lockdown period. Thirdly, patients' perceptions about physical activity poorly correlated with the remote monitoring-assessed physical activity levels. Finally, anxiety and depression levels did not correlate with patients' physical activity.

Study implications
This study assessed main modifications occurred during the COVID-19 pandemic in a population of HF patients implanted with a CIED. Forced homestay and physical activity levels during the lockdown offered a peculiar overview of the acute effects that physical inactivity may exert in the real-life setting, playing as a potential risk factor for HF recurrencies and delay in the access to care. A recent survey of 1,047 subjects highlighted the impact of the lockdown on physical activity levels, with an estimated reduction of 24% [20]. Preliminary reports showed that hospitalizations due to decompensated HF significantly declined during the COVID-19 pandemic, yet hospitalized patients displaying more severe symptoms at admission [21]. This may be surrounded by a significant delay in hospital admissions due to patients' fear of contagion and crowded areas, as noted for other acute events [22]. Similarly, psychological factors influence the choice of longterm cardiovascular pharmacotherapy. Preliminary warnings arose about the safety of angiotensin converting enzyme inhibitors and angiotensin II receptor blockers with regards to the spreading of COVID-19. However, subsequent studies and their meta-analyses disavowed these concerns, supporting the recommendation for continuing these drugs by European and American cardiovascular societies [23]. Physical activity levels, mean heart rate, resting mean heart rate and thoracic impedance from CIED remote monitoring are broadly known to be helpful in predicting clinical outcomes in HF patients.
Physical inactivity has been largely studied as a crucial risk factor in HF patients, with lower physical activity levels associated with a higher risk for recurrencies and hospitalization [24]. Interestingly, our study showed that the extent of physical activity reduction varied across the lockdown period, with a greater reduction across the first two weeks. Similarly to physical activity, mean heart rate plays a key role as a predictor of poor prognosis in HF patients [25][26][27]. Thoracic impedance has been used in the management of HF patients as well, acting as a proxy measure for pulmonary fluid status [28]. Indeed, it is well-known to be inversely correlated with pulmonary fluid balance and capillary pressures, with a reduction occurring before HF recurrencies, symptoms onset and requirement for hospital admission [29]. A routine monitoring of thoracic impedance may signal an upcoming fluid overload suggesting the potential for optimization of medical therapy [30]. The MOMOTARO II (Monitoring and Management of OptiVol Alert to Reduce Heart Failure Hospitalization II) was a prospective  randomized trial investigating the role of thoracic impedance in patients with CIEDs to explore the role of lifestyle modification and pharmacological prevention of acute HF. In particular, patients assigned to the lifestyle modification group (reduction of sodium, water intake and daily activity for 1 week) showed an increase in thoracic impedance of 6.2% (from 63.3 ± 9.6 to 67.2 ± 10.1; P < 0.001), signaling a reduction of lung congestion. In our study, the percentage decrease of thoracic impedance was slightly lower than in the MOMOTARO II trial, probably due to the observational design of our anaylsis [31].

Study limitations
Our study should be interpreted in the light of several limitations. Firstly, this study represents a retrospective analysis exploring data extracted from CIED remote monitoring of only a manufacturer. Secondly, no active management of pharmacotherapy was performed based on the results of remote monitoring. Thirdly, 17 patients were not included in our analysis due to the low quality of remote monitoring transmissions; we cannot exclude differences in baseline characteristics of these patients, as a result of a selection bias. Finally, no clinical information was available for study participants due to COVID-19 pandemic discouraging outpatient follow-up visits.

Conclusions
Forced homestay during the COVID-19 pandemic played a key role in modifying vital signs and physical activity levels of HF patients, who tended to display poor insights about their physical activity, with a very low correlation degree between patientreported and remote monitoring-assessed physical activity levels.
Based on these findings, telemedicine and remote monitoring may be of crucial importance in the prediction of HF worsening during circumstances discouraging outpatient visits.

Declaration of Competing Interest
The authors report no relationships that could be construed as a conflict of interest.