Wearable heart rate variability and atrial fibrillation monitoring to improve clinically relevant endpoints in cardiac surgery—a systematic review

Introduction

This systematic review aims to shed light on the prognostic potential of heart rate variability (HRV) metrics and improving atrial fibrillation (AF) diagnosis in cardiac surgery (CS) patients considering the pivotal role of AF, HRV assessment, and current development of continuous health monitoring methods via wearables.

CS patients with an impaired cardiovascular system are particularly prone to preoperative and postoperative life-threatening complications such as arrhythmias, strokes, and sudden cardiac death (1-4). Continuous monitoring of AF are common risk factors and complications after CS. Tools to better predict and eventually prevent associated adverse events (AEs) are thus clearly needed (5,6). The pivotal role of continuous assessment of AF and HRV has been described: commonly available wearables with integrated single-lead electrocardiogram (ECG) and photoplethysmography (PPG) can provide easily accessible continuous data. Alterations in the HRV pattern provides an early marker of compromised health (7-11). In the 1970s, decreases in HRV were first associated with increased mortality and increased incidence of cardiac arrhythmias in patients after myocardial infarction (12). In more recent studies, reduced HRV has been found to independently predict all-cause mortality and sudden cardiac death in patients with acute and chronic heart failure (13,14). Given that research on HRV as a biomarker for AF prediction and other cardiac AEs is still in its early stages—and standard AF monitoring still consists mostly of singular heart rate measurements—we emphasize the importance of continuous data collection to drive both the improvement of AF detection and the progression of HRV as an early marker for AEs in CS.

In brief, frail CS patients may undergo pre-, peri, and post-interventional complications, most commonly AF, which is often subclinical and may be overlooked. Continuous monitoring today via wearables (e.g., Apple, Fitbit, and Samsung watches) can improve patient outcomes. They have already received the CE mark and FDA approval to report sinus rhythm, AF, and inconclusive (15). Moreover, HRV assessments via wearables could predict complications such as post-operative atrial fibrillation (POAF) or sudden cardiac death sudden cardiac death. This systematic review aims to describe the current state of continuous monitoring through wearables in CS patients and outline an outlook of future potential highlighting AF detention and risk evaluation with HRV.

The number of conduced cardiac surgeries per year range from 0.5 per million in low- and lower-middle income countries (average 107±113 per million) to 500 in the upper-middle-income countries (average 107±113 per million). Outside the industrialized world, CS is insufficiently accessible for more than 6 billion people. Therefore, a population ranging from 1.6 to 500 billion per year would benefit from improved AF monitoring and HRV assessments. Obviously, this benefiting population would further increase as worldwide access to CS improves because low-income and middle income countries are still on their way to fully meet their needs (16). We present this article in accordance with the PRISMA reporting checklist (available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-23-19/rc).

Methods

Search strategy

A focused query of the most popular databases (PubMed, Scopus, and Web of Science) was conducted on the 28th of October 2022. There was no restriction on date or language for publication selection. The following keyword combination yielded 33 results: ((Wearable) AND (Cardiac Surgery) AND (Heart Rate Variability)) OR ((Wearable) AND (Cardiac Surgery) AND (Atrial Fibrillation)).

Among the 33 results, there were nine articles that focused on postoperative monitoring, three articles that focused on preoperative assessment and screening, five reviews, two meta-analyses, and five results of low relevance. One publication was removed prior to screening because the study did not focus on CS patients. In addition, nine results were excluded because of a lack of focus on patients or applicability to CS. The focus in the studies excluded from the analysis was on the cognitive load on the team during surgery from HRV assessment, the influence of specific drugs, implantable devices, and specific brands of devices (17-26). Figure 1 shows the PRISMA flow diagram for the selection of outcomes included in this review.

Figure 1 The PRISMA flow chart indicates the process of research, screening, and inclusion of records for this review.

Study biases

Attrition-bias might have impact on the study outcomes because CS patients are often frail, elderly, and undergo an emotionally arousing situation. Therefore, probands with an impaired cardiovascular system and consequently a higher probability to develop arrhythmic events could be more likely to drop out of the studies or show incompliance wearing the wearable eventually impairing sufficient data collection. Intention-to-treat analysis that better preserves randomization than per-protocol-analysis was conducted in most included studies to mitigate attrition-bias. When reported by these studies, intention-to-treat-analysis was often applied in wearables studies for cohorts obligated to wear the device, whereas per protocol analysis was conducted for patients that did not undergo continuous monitoring in most cases.

There are other important biases of concern such as the risk of response bias in cardiovascular symptom reports. Additionally, reporting bias may impact on the study that we could assess for this review because positive results are more likely to be published than negative. This can substantially influence the findings of this systematic review.

Results

The burden of AF in CS

AF plays a crucial role in primary and secondary stroke prevention and is associated with a significant increase in morbidity and mortality (1,10). In North America, an estimated 12–20% of all ischemic strokes are due to AF (27,28). In patients with paroxysmal AF, the risk of stroke increases up to 5-fold, whereas the annual risk of suffering a cardiac-related stroke varies between 1–12%.

To accurately assess the individual stroke risk and thus evaluate the indication for anticoagulation, the 4S scheme is used in patients with AF according to current 2020 ESC guidelines (29). As outlined in Figure 2, the 4S scheme assesses the following parameters: (I) stroke risk, (II) severity of symptoms, (III) severity of AF burden, and (IV) severity of substrate. As part of the 2020 ESC Guidelines, the AF Better Care (ABC) pathway guides AF management (indicated by Figure 3). CS patients whose cardiovascular systems are often severely compromised will benefit from improved AF screening and continuous postoperative long-term monitoring.

Figure 2 The 4S-AF scheme aims to outline a structured representation of AF management by relevant factors grouped into four AF-related domains (stroke risk, symptom severity, AF burden severity, substrate severity). CHA₂DS₂-VASc, congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke, vascular disease, age 65−74 years, sex category (female); EHRA, European Heart Rhythm Association; AF, atrial fibrillation; TEE, transesophageal echocardiography; TOE, transesophageal echocardiography; CT, computed tomography; MR, magnetic resonance.

Figure 3 The ABC pathway aims to streamline an integrated approach to AF management. Hereby, three main pillars are outlined: “A” Avoid stroke (with Anticoagulants); “B” Better symptom management, with patient‐centered decisions on rate or rhythm control; “C” Cardiovascular and Comorbidity risk optimization (29). AF, atrial fibrillation; ABC, Atrial fibrillation Better Care; OAC, oral anticoagulation; NOAC, new oral anticoagulation; VKA, vitamin K antagonists; TTR, time in therapeutic range; Qol, quality of life; CV, cardioversion; AAD2, antiarrhythmic drugs class 2.

Wearable health monitoring devices (wearables) are designed to be worn by patients in their everyday lives and can rely on increasingly accurate and sophisticated artificial intelligence supported measurement methods. This will take preoperative and postoperative AF screening to a new level. A key driver of this evolution is the ability to easily monitor heart rate non-invasively and continuously, which most clearly differentiates wearables from traditional AF screening methods. The predominant technical methods used in wearables for this purpose are PPG and single-lead ECG. It is therefore likely that wearables will not only enable earlier and more accurate AF detection before and after surgery, but, through their continuous long-term monitoring of health data, will provide new and much deeper insights into the postoperative course of patients undergoing CS than ever before (30).

The physiological background behind HRV monitoring

The HR is the number of heartbeats per minute. The HRV expresses the physiological variations of time-intervals between successive heartbeats both at rest and during exercise (31). Although HRV has recently been widely implemented in medical and commercial applications, its scientific investigation is still in an early stage (32). HRV expresses neuro-cardiac function and is composed of heart-brain interactions and dynamic non-linear processes via the autonomic nervous system (ANS). Different interdependent regulatory systems operating on different time scales create HRV and enable us to properly adapt to an everchanging environment. Indeed, the HRV reflects ANS balance, the activity of the central nervous system, endocrine system, heart, gut, blood pressure (BP), and vascular tone (31,33). For instance, information about central nervous activity is reflected in higher levels of vagally-mediated HRV at regular heartbeat. These changes correlate to executive functions such as attention and emotional processing by the prefrontal cortex (1). On the contrary, afferent information produced by the intrinsic cardiac nervous system can influence frontocortical activity and thus affect higher-level functions (34,35). The heart rate itself is mainly regulated by the sympathetic and parasympathetic branches of the ANS. Thus, HRV is not only a relevant parameter for global cardiac health but is also an important indicator of the functional and cardiovascular integrity of the ANS. Often expressed in sympathetic hyperactivity, ANS impairment is linked to hypertonia, arrythmia, and other risk factors of cardiovascular deterioration. Consequently, early ANS assessment after pathological HRV measurements might be of particular importance in elderly and frail CS patients. We thus hypothesized that the extent of dysfunctional cardiac ANS integrity may not only provide essential insight into the optimal timing of the treatment, but might also be able to distinguish beneficial from futile regimens, thus ultimately providing a potential key factor for improved short- and long-term outcomes (36-38).

To understand the different HRV measurement methods and their correlation to either sympathetic or parasympathetic activity, we need to know the tendency of a healthy heart to have a strictly regular rhythm during high-frequency activity. The healthy heart shows lower HRV during physical exertion (strictly homogeneous rhythm) and higher HRV in relaxed mode (increasingly inhomogeneous rhythm). Physiological frequency fluctuations are most likely to be seen in the context of “respiratory sinus arrhythmia” (RSA).

The central ANS controls heart rate mainly via parasympathetic neuronal networks originating from the nucleus ambiguous of the medulla oblongata in synchronization with sympathetic nerve activity (superior cervical ganglia, ganglion stellatum, and thoracic ganglia) (39,40). Autonomic cardiac innervation is further modulated by various factors such as baroreceptors in the arcus aorticus and sinus caroticus or chemoreceptors in the glomus aorticum and caroticum. Best represented by the HF band (0.15–0.4 Hz) described below, the RSA forms the physiological heart rate pattern, which increases during inspiration and decreases during expiration (40,41). The baroreflex supports RSA and links heart rate, BP, and vascular tone. RSA fluctuations are mostly created through parasympathetic activation and inhibition (31). This can be illustrated by the example of a denervated heart, e.g., in the context of a heart transplantation where vagal tone is completely absent. As a result, the heart beats at an intrinsic rate averaging over 80 bpm at rest—this is significantly higher than a healthy heartbeat (42).

Our HR increases when we inhale followed by a BP rise about 4–5 seconds later (43). This BP rise is recognized by baroreceptors that consequently fire more rapidly. On the contrary, a rapid decline in BP recognized by these baroreceptors might activate vasocontraction to sustain sufficient perfusion. Expressed by the law of Hagen-Poiseuille [Q = (ΔP × r4 × π)/(8 × η × l)], the arteriole-diameter adds most to peripheral resistance. BP is determined mainly by peripheral resistance and cardiac output, and thus the link between RSA, baroreflex, and vascular tone is unambiguous (31,43).

Healthy biological systems such as the cardiovascular system are characterized by temporal complexity. A decrease or increase in this complexity above a physiological threshold can be triggered by illness. Although increased HRV is associated with improved cardiac health in most cases, pathological conditions such as conduction disorders sometimes can lead to increased HRV, which in turn is associated with increased risk of death (31).

The different half-lives of norepinephrine and acetylcholine allow us to clearly assign certain frequency bands to sympathetic or parasympathetic activity. While the acetylcholine of the parasympathetic nervous system is metabolized with a high turnover rate and a very short latency period, the norepinephrine of the sympathetic nervous system is metabolized much more slowly. Therefore, the parasympathetic influence on HRV occurs relatively fast and is associated with the high-frequency band. In contrast, the influence of the sympathetic nervous system correlates with the low frequency (LF), very low frequency (VLF), and ultralow frequency (ULF) bands. Consequently, high frequency (HF) metrics, which represent a rapid change in ANS influence on the heart are related to a dominant parasympathetic branch.

In summary, HF parameters, including the frequency range of RSA described above, are a clear indicator of parasympathetic function, whereas LF metrics represent mainly the sympathetic influence on the ANS (<0.15 Hz) (9,44). The assessment of HRV by quantifying the individual frequency components of a frequency spectrum using spectral analysis is shown in Figure 4.

Figure 4 An ECG input signal is partitioned into different component bands by spectral analysis and the frequency domain method. Thereby, high frequency bands correlate with parasympathetic activity, while low frequency bands correlate with sympathetic activity (26). HRV, heart rate variability; ECG, electrocardiogram.

An overview of most relevant HRV metrics

Information on HRV can be extracted through time-domain, frequency-domain, and non-linear metrics. The complexity and unpredictability of a series of heartbeats (R-R intervals) is assessed in non-linear measurements as shown in Figure 5 (31). Time-domain parameters calculate the amount of HRV during monitoring periods ranging from 2 minutes to 24 hours (most relevant time domain metrics are displayed by Figures 6,7). Frequency-domain parameters quantify the amount of signal energy within the components of the frequency spectrum (31). Longer recordings better depict slower HRV fluctuations such as the circadian rhythm, and therefore are more prone to being influenced by environmental stimuli. Thus, these long-term recordings (~24 hours) cannot be directly compared with short- or ultrashort-term recordings. When calculating the variability of the R-R intervals, the indices in the time domain are expressed in original units or as the natural logarithm (Ln) of the original units to improve normal distribution of values (44-47).

Figure 5 Most relevant non-linear parameters for HRV assessment (26). HRV, heart rate variability.

Figure 6 Most relevant time-domain parameters for HRV assessment (26). HRV, heart rate variability; ECG, electrocardiogram.

Figure 7 Additional time-domain parameters for HRV assessment (26). HRV, heart rate variability; ECG, electrocardiogram.

According to the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, frequency-domain measurements are divided by spectral analysis into four different frequencies: HF, LF, VLF, and ULF. The absolute or relative HRV power can be calculated within each band. The techniques that are used the most are the fast Fourier transformation or autoregressive modeling. The power describes the signal energy detected in each frequency band. Thus, absolute power is calculated as meter per second squared divided by cycles per second (ms²/Hz), whereas relative power represents the proportion of a single frequency band to the absolute power of all low and HF bands measured in normal units (nu). Relative power enables easy HRV comparison between two patients, in case of large differences in total or specific band power (48).

Using 24-h recordings, the ULF band (≤0.003 Hz) quantifies fluctuations in R-R intervals within a period of 5 min to 24 h (10). Very slow acting biological processes affect the ULF band, and thus scientists assume that circadian rhythm may be the main driver of ULF alterations besides the body core temperature and certain metabolic events such as the renin-angiotensin system (49,50).

The VLF band (0.0033–0.04 Hz) measures rhythmical alternations within periods between 25 and 300 s. Requiring recording periods of at least five minutes, these recordings typically include 0–12 oscillating cycles. Overall, the VLF value may be most valuable in assessment of general cardiac health because low values on a 24-h VLF bands are associated with greater risk of AEs, and VLF power displays much stronger correlation with all-cause mortality than LF or HF bands. Furthermore, low power in the VLF band is related to low testosterone levels, high circulatory inflammation parameters, arrhythmic death, and post-traumatic-stress-disorder (11,44,51-55). Sympathetic impulse may modulate the amplitude and frequency of these oscillations, whereas parasympathetic modulations seem to be most influential on VLF power; parasympathetic blockage almost eradicates VLF power in general (56).

The LF band (0.04–0.15 Hz) requires recording periods of at least 2 minutes and predominantly reflects baroreceptor activity in resting conditions. That is why the LF band was originally called baroreceptor range. Influenced by both parasympathetic and sympathetic activity during normo-frequent breathing, vagally-mediated efferent respiratory influences are expressed particularly in the LF band during LF breathing (8.5 bpm) (55,57).

The HF band requires recordings over a minimum of 1 minute and previously has been called respiratory band (0.15–0.40 Hz). It is mostly influenced by normo-frequent breathing patters ranging from 9 to 24 bpm and reflects mostly parasympathetic activity (50). Total vagal blockage lowers LF power and completely irradicates HF oscillations. Remarkably, mitigated vagal inhibition is associated with morbidity, and low HF power correlates with mental stress, panic, and anxiety (58).

The LF to HF power ratio (LF/HF ratio) represents the ratio between the sympathetic nervous system (SNS) and the parasympathetic nervous system (PNS) activity. Total power is described by the sum of energy in every frequency band for 24 hours and the VLF, LF, and HF bands for short-term recordings (55). Ultimately, the unpredictability and intrinsic entropy of HRV are indexed by non-linear measurements (59) (Figure 7).

Potential of HRV metrics with wearables to improve clinically relevant endpoints

HRV has already been established as a helpful predictor in certain cardiac scenarios. On the one hand, HRV correlates with outcomes after acute myocardial infarction (AMI) (60,61). On the other hand, HRV-recorded vagal modulation predicts increased risk of sudden cardiac death (62). Thus, early diagnosis of these risks and subsequent implantable cardioverter defibrillator (ICD) implantation might be lifesaving. Moreover, in patients with hypertensive heart failure, root mean square of successive RR interval differences (rMSSD), and pNN50 (time-domain metrics outlined in Figures 6,7) were found to be independent predictors of the occurrence of AF along with traditional risk factors (age, hemodialysis, high PAC burden). Kim et al. concluded that abnormal autonomic cardiac regulation, represented by higher HRV in this study, is correlated with a higher risk of AF development (63). However, in CS, patients with a shorter PR interval and a decrease of the non-linear HRV parameter DFA1 could be identified as significant predictors of POAF (non-linear metrics are outlined in Figure 5) (64). In summary, further research is needed to realize the full potential of readily available HRV data in the face of a future in which a large proportion of people wear a continuous monitoring device. This is still true considering these findings on HRV metrics as predictors of mortality after myocardial infarction, risk of sudden cardiac death, and AF during CS as well as in patients with hypertension.

AF diagnosis + HRV assessment with wearables

Similar to heart rhythm analysis with ECG, the current gold-standard of AF diagnosis, HRV metrics also derive from beat-to-beat (R-R) interval analysis (32). With respect to the 2020 ESC Guidelines, AF is diagnosed by either a 12-lead ECG or an at least 30 seconds recording of a single-lead ECG (65,66).

AF is defined as supraventricular tachyarrhythmia with uncoordinated electrical activity and inefficient contraction of the atrium and is characterized by irregular R-R intervals (with preserved integrity of atrioventricular conduction) and the absence of P waves in the ECG recording (24). AF is often associated with a ventricular rate of approximately 110–160 bpm, and AF can be further distinguished between the two types of rapid ventricular response and slow ventricular response constituting a ventricular rate of >100 and <60 bpm respectively (67). AF can often present as a subclinical phenomenon. In 18% of cases, AF is first detected at the time of stroke, thus highlighting the utility of using wearables for early detection (68). Currently available wearable health monitoring devices with CE mark and/or FDA approval are outlined in Table 1 (30,69-71).

Compared to common chest-mounted devices such as a Holter ECG, watch-like continuous health monitoring devices are convenient and easy to integrate into the patient’s daily routine (72). Despite the 2020 ESC guidelines, which for the first time have also addressed the diagnosis of AF using single-lead ESC, and the increasing use of wearable health monitoring devices by a broad mass of the population, further research is needed to perform screenings for silent AF in the general population and to reduce the associated morbidity and mortality. Currently, wearable health monitoring devices are seen more as medical advisors or detectors of rhythmic irregularities than as diagnostic tools (68).

Standards engineering methods in wearables: PPG + single-lead ECG

Single-lead ECG and PPG are the most common portable health monitoring devices. Therefore, this review focuses only on these two standard engineering methods.

PPG measures pulsatile volumetric changes in microvascular blood flow during the cardiac cycle by converting light absorbed or reflected by blood flow (typically LED or infrared light) within the sensor coverage area into waveforms and measurable signals (73). This amplitude of these waveforms marks the maximum of ventricular contraction during systole and resembles the R-peak in the ECG (68). Disadvantages of PPG measurements include the need for adequate BP, which is not the case, for example, when AF triggers peripheral malperfusion (74). Furthermore, PPG can only serve as an aid in the diagnosis of AF because, unlike the ECG, it is not considered a valid diagnostic method to date. A watch-type single-lead ECG can record the Einthoven I lead with the right index finger on the crown acting as the negative electrode and the watch on the left wrist acting as the positive electrode (75). The most relevant clinical applications of PPG and ECG in wearables are outlined in Figure 8. Consequently, the combination of screening by PPG and diagnosis by single-lead ECG could be a plausible solution to reduce undetected AF and related complications.

Figure 8 Clinical applications of ECG and PPG monitoring by wearable health monitoring devices (71). HRV, heart rate variability; PPG, photoplethysmography; HR, heart rate; HRR, heart rate recovery; BP, blood pressure; ECG, electrocardiogram; CVD, cardiovascular disease.

Pre-surgery

In addition to medical therapy, there are now additional minimal invasive treatment modalities such as left atrial appendage closure and radiofrequency catheter ablation that can both reduce thromboembolic risk and provide long-term freedom from AF (76,77). Earlier diagnosis of AF through continuous monitoring may thus allow these therapies to be applied at an early stage to prevent AEs (30,78,79). In a study of a population of 96 patients, Kim et al. compared continuous long-term measurement methods of a wrist-worn health monitoring device versus conservative, single-point screening methods with respect to the detection rate of AF (30). Paroxysmal arrhythmias often remain undetected during single-point screening using 12-lead ECGs and even Holter monitoring. In contrast, long-term continuous monitoring could detect 53% of all cardiac arrhythmias. Holter monitoring could only detect a significantly lower proportion at 28.1%. In conclusion, this study supports the potential of the wearable health monitoring device in early detection of AF as 24.9% of total cardiac events that have been missed without continuous measurement methods (30).

Periprocedural

Early detection of AEs in high risk-CS patients

Postoperative complications of CS are significantly associated with intrahospital mortality, but recognition of precursor events is often difficult leading to avoidable rescue failures (80-84). In addition, in most cases, signs of deterioration indicative of AEs occur up to 38 hours before the patient is transferred to the ICU (85-88). Therefore, efficient detection of these AE precursor events is critical to improve clinical outcomes.

To date, several wearable health monitoring devices have been tested for tracking vital signs in the hospital setting. However, the major obstacle to monitoring by wearables is the unacceptably high number of false-positive alarms. Therefore, machine learning systems were used to set more precise parameter thresholds for alerting. Approaching this end, Breteler et al. found that simultaneous increases in HR and RR and decreases in SpO₂ were associated with AEs during general and trauma surgery. Within the study population of 31 patients, 11 patients developed AEs, with AF as the most common cause in 20% (89). However, future studies incorporating larger study populations are needed to develop more precise vital sign warning patterns or algorithms for appropriate alerting systems. Similarly, other studies investigated the correlation between the new set AF and postoperative complications (90). Although a considerable number of studies have investigated the recording of vital signs to establish prognostic warning patterns, the number of studies evaluating continuous monitoring using wearable health monitoring devices in the hospital before and after surgery is limited (88,91-95). For example, continuous PPG monitoring could indicate a higher risk of cardiovascular events by assessing cardiac stress using measures such as ejection time index, subendocardial viability ratio, and evaluation of arterial stiffness by assessing pulse wave velocity (96,97).

In conclusion, the scientific community should further explore continuous monitoring by wearable health monitoring devices used in an in-hospital setting both to improve standard monitoring and to prevent serious postoperative complications.

Post surgery

POAF in CS patients

POAF is one of the most common AEs in CS patients within 30 days of discharge and is associated with a significantly increased risk of stroke (25–33% lifetime risk), heart failure, dementia, and death (98,99). In addition, according to the 2020 ESC guidelines, the decision to treat AF is based on the 4S regimen in which other cardiovascular deteriorations are relevant factors. Continued anticoagulation may lead to bleeding events and therefore must be weighed against the benefits of stroke prevention in each individual patient. However, early POAF diagnosis followed by pharmacological therapy could protect a significant proportion of CS patients from fatal strokes.

General risk for AF after CS

The incidence of POAF in coronary artery bypass grafting (CABG) and valve surgery patients is 20–30% and even up to 50%, respectively (28,100-105). To investigate the potential of continuous monitoring in CS, Funk et al. studied 302 patients undergoing CABG and/or valve surgery using wearable health monitoring devices during hospitalization and for an additional 14 days after discharge. The study reported that AF occurred in 42% of subjects with AF occurring after discharge in approximately 33% of these subjects. Of note, up to 70% of AF episodes were subclinical and were not associated with symptoms such as palpitations, fatigue, chest discomfort, or dyspnea. In addition, valve surgery and pulmonary hypertension, AF during index-hospitalization, and discontinuation of ß-blockers were found to be predictive of subsequent AF occurrence after discharge (106). By demonstrating a substantial degree of asymptomatic AF after discharge of CS patients, this study reinforced the importance of improved monitoring strategies with wearable health monitoring devices.

Wearables on the rise for POAF screening

Patch-based POAF screening

To investigate POAF screening by continuous monitoring methods, another study investigated a patch-based wearable health monitoring device applied after hospital discharge following CS. The results showed that continuous monitoring improved POAF detection by 17.9% in 336 CS patients for 30 days after CS compared with usual care (12-lead ECG screening) (107). The inclusion criteria were the absence of preoperative AF and a CHA2DS2-VASc score of greater than or equal to four. In addition, patients who reported AF of >24 h duration during hospitalization were excluded. A total of 19.6% of all patients monitored with the wearable health monitoring device met the criteria for intention to treat whereas only 1.7% of patients monitored with the 12-lead ECG were estimated to need treatment (absolute difference: 17.9%). Based on the inclusion criteria, the risk of stroke in patients diagnosed with POAF is estimated to be 1.3% and would meet the threshold for oral anticoagulation as described above (29). Notably, the relevant AF duration in this study was set at ≥6 minutes, and a substantial 73.3% were diagnosed in the first week after surgery. Consequently, continuous monitoring by wearable health monitoring devices for as little as one week would greatly improve postoperative surveillance and reduce future stroke events.

PPG-based POAF screening

Because of the practicality of wrist PPG monitoring, it can be easily used for screening POAF. Its predictors include nonlinear HRV parameter DFA1 mentioned above; it is seen in the long-term monitoring in CS patients (64). Today, guidelines call for 12-lead ECG recording at 3 and 6 months after index surgery followed by annual follow-ups. In contrast to continuous PPG screening, the major disadvantages of single time point measurements by 12-lead ECG are the non-detection of short AF episodes in measurement-free periods, the need for strong compliance, the cost to patients and healthcare providers, and the impossibility of detection during periods of mobilization (108). In terms of future improvements, continuous monitoring using PPG to monitor arrhythmias could be recommended in addition to standard procedures. Thus, wearable long-term monitoring could detect recurrence of AF or other arrhythmias and eventually lead to better treatment outcomes.

Immediate AF diagnose through machine learning

To diagnose paroxysmal AF even earlier, researchers are trying to improve PPG through machine learning. To develop the algorithms, the investigators provided CS patients with an Apple Watch (109). Thus, 80 patients underwent telemetry monitoring, 12-lead ECG screening, and continuous heart rate monitoring using Apple Watch in training mode over a 14-day period after CS in 2021 and 2022. Regarding the accuracy of the data, small black spots in the detection of AF could be compensated by the applied machine learning model GBDT (Gradient Boosting Decision Tree) and reached an accuracy of 0.94 (sensitivity: 0.909, specificity 0.838). This accuracy rate of the improved PPG system of the Apple Watch is comparable to similar wearable health monitoring devices, which generally achieve an accuracy rate of >0.9 (110-112). The investigators ultimately did not achieve their goal of establishing a more accurate diagnosis of AF. However, precisely distinguishing AF from other arrhythmias under perioperative conditions caused by various cardiac irregularities is inherently complex.

One of the major impediments to continuing progress on the development of an immediate diagnostic tool of AF may be related to the nature of PPG that derives its electronical impulses from pulse pressure curves propagating through the cardiovascular tree. In addition to the substantial cardiac changes and altered physiological processes immediately after CS complicating diagnosis, the compromised cardiovascular system often causes inadequate pulse pressure and intravascular volume, thus making accurate monitoring by PPG difficult (74,113-115). To address the question of whether a device other than the Apple Watch could provide more accurate data, Inui et al. compared the metric accuracy of the Apple Watch 3 and Fitbit, which showed slightly higher accuracy of 75% and 30%, respectively, when comparing pulse rate and heart rate (74). Therefore, techniques other than PPG could probably provide better support for more accurate AF detection in frail CS patients.

Algorithm-supported AF screening in a large population

Recently, the Fitbit Heart Study, published in September 2022 by Lubitz et al., investigated on a novel PPG-based software algorithm aiming to detect irregular heart rhythms (IHR) in the general population to identify undiagnosed AF. The Study population (≥22 years) was monitored by using compatible wearable Fitbit devices and Android or iOS smartphones. In detail, 5-minute pulse windows (tachograms) were assessed, and subsequently participants revealing 11 consecutives irregular tachograms were identified as showing an IHR. These participants with an IHR were invited to a telehealth visit and received a 1-week ambulatory ECG patch monitor. Eventually, the first IHR detection with the ECG patch monitoring was assessed as a predictor for AF as the primary outcome. Results displayed 4,728 IHR in a total of 455,699 enrolled participants representing 1% (4%, 2,070 participants, in the ≥65-year-old). AF was present in 340 (32.2%) of the 1,057 participants who received the ECG patch monitor and in 221 participants of the 225 participants with another IHR during ECG patch monitoring resulting in a positive predictive value of 98.2%. In conclusion, the combination of algorithm-supported AF screening with Fitbit wearables in a large population exhibited a high positive predictive value and may be a strong tool for decreasing undiagnosed AF in the general population (116).

HRV as predictive value in CS

CS is known to alter the ANS system in several ways. First, anesthesia (e.g., propofol) increases sympathetic nerve activity, thus resulting in relaxation of peripheral vessels (117). Second, the pain stimulus, unless adequately suppressed by analgesics, induces an increase in heart rate due to the release of catecholamines resulting in an increase in LF and VLF power. The latter can also be induced by intravenous application by anesthesia (118). In addition, recovery of cardiac activity after cardioplegia affects the ANS. A variety of postoperative complications are related to ANS instability, and thus HRV may play a critical role in detecting ANS imbalances in the future and thus improve the prevention of complications in the early and late postoperative period.

HRV after AMI

Today, guidelines for treatment after AMI rely heavily on the left ventricular ejection fraction (LVEF) to decide between further treatment options such as implantation of an ICD or optimized pharmacological therapy. LVEF is often preserved after a myocardial infarction and revascularization (119,120). Therefore, the following additional metrics are needed to refine treatment options. Wolf et al. discovered that patients with a more pronounced HRV had a lower mortality rate after AMI than patients with a less pronounced HRV (12). In addition, other studies found significant associations between patients with decreased HRV and subsequent myocardial infarction and/or higher risk of sudden cardiac death (121). Consequently, the majority of the scientific community recognizes HRV as a strong marker of arrhythmogenic death and postulates its superiority versus ejection fraction-based metrics (94,122,123).

Decreased HRV often occurs because of increased sympathetic activity—adrenergic activity correlates with electrical instability. This is readily exemplified by AMI-induced myocardial necrosis, which exhibits increased sympathetic activity and drives malignant arrhythmias. Changes in HRV are also observed in patients with chronic heart failure, which is characterized by overactivation of the sympathetic nervous system. Furthermore, relevant correlations between HRV impairment and disease severity have been found especially a correlation between the LF/HF ratio and the risk of AEs, as well as a correlation between the LF parameter and right ventricular dysfunction could be described. Consequently, HRV could be used in addition to standard metrics, such as LVEF, in important decision-making and patient selection to improve clinical outcome (32,94,124-127).

Post-CABG

It is well known that coronary artery bypass leads to a significant reduction in HRV, which recovers in most patients after approximately two months. Versus HRV deterioration after AMI, HRV reduction after CABG is even more pronounced and is not caused by necrosis but is rather related to the use of the heart-lung machine, cardioplegia, anesthesia, and cardiac manipulation. Notably, due to the extreme adrenergic release during both on- and off-pump grafting, CABG invariably leads to a reduction in HRV immediately after the procedure (128,129). However, revascularization apparently outweighs ANS irritation, although HRV does not recover in every patient as shown by long-term follow-up. The prognostic value of HRV in the context of CABG is controversial. The results of the CAST (Cardiac Arrythmia Suppression Trial) suggest that there is no association between HRV recovery and clinically relevant endpoints or mortality. However, HRV metrics assessed by the nonlinear domain measurement method showed a particularly strong correlation between postoperative HRV reduction and the occurrence of AEs as well as the increase in mortality (129,130). In conclusion, the scientific community should conduct additional research to determine whether the persistent decline in HRV in CABG patients has prognostic value and whether restoration of HRV could lead to prognostically favorable outcomes.

HRV in the context of heart valve procedures

To date, HRV reduction after cardiac valve surgery has not been found to have predictive value for long-term outcome or mortality. HRV deteriorated more after mitral valve compared to aortic valve surgery, which might be explained by the larger surgical accesses and the intensity of manipulation in mitral valve procedures. In addition, percutaneous procedures such as transcatheter aortic valve implantation (TAVI) did not cause significant worsening of HRV, whereas transapical procedures resulted in a mean reduction in HRV due to the higher invasiveness of the procedure (131).

Liu et al. investigated continuous monitoring after transcatheter aortic valve replacement (TAVR) by a smartwatch in 100 patients for 30 days after discharge. Cardiac events occurred in 34 patients, and four patients received pacemaker implantation as a result (132). Postoperative complications, such as conduction disorders occurring with latency can be life-threatening and must be detected at all costs in all patients. Continuous monitoring could certainly be helpful—especially in those patients who are scheduled for discharge the next day. In similar studies, a new diagnosis of AF was made in four patients—one of whom received anticoagulation as a result (133-135). We see that continuous monitoring beyond the hospitalized period is key to preventing late postoperative complications in the future.

PPG for HRV assessment

The biggest plus of PPG measurement for HRV monitoring is the convenience for the patient. The drawbacks of the PPG method include that HRV measurement is often compromised by low hypovolemia, that it varies significantly between subjects, and that there are numerous discrepancies between certain subset parameters. While some subset parameters can be adequately assessed by PPG (e.g., SDNN), most parameters belonging to the high-frequency band (LF/HF ratio) have not been adequately validated. One study even reported a measurement accuracy of only about 30% (136). Although PPG certainly has its drawbacks in terms of HRV assessment, PPG-equipped wearable health monitoring devices with sophisticated algorithms provide AF screening that is nearly comparable in quality to gold standard methods as demonstrated by the recent Huawei or Apple Watch study (96).

Overall challenges of wearables in a clinical practice are indicated by Table 2.

Discussion

Discussion on POAF monitoring via wearables

CS patients with a weakened cardiovascular system are vulnerable to AF and consecutive strokes; the 4S and ABC scheme of the ESC 2020 guidelines provide good guidance in the assessment and therapy (described in Figures 2,3). In addition, AF often occurs after discharge and in most cases presents without symptoms. For this reason, patients should be regularly evaluated for AF even after their hospitalization to benefit from earlier medical treatment such as anticoagulation, radiofrequency ablation, or atrial appendage closure. Consequently, the scientific and clinical community should make further efforts to explore and implement accurate, convenient, and affordable screening options via wearable health monitoring devices to promote AF and stroke prevention. These efforts would ultimately not only prevent serious complications but should also be undertaken in the interest of health care costs.

It is particularly important to note that POAF is not limited to the period of hospitalization because a significant percentage of AF has been detected by continuous monitoring using wearables after discharge. POAF is often described as a transient phenomenon attributable to pericardial inflammation, cardiac ischemia, hemodynamic alterations, and increased adrenergic state after surgery (137); however, conclusive studies examining periods after hospitalization of several weeks or even months are scarce. Therefore, the scientific community should investigate long-term monitoring through wearable health monitoring devices after CS, which have recently enabled and simplified this finding. Further research is needed to investigate the unexplored areas of POAF screening and potential therapeutic indications. Not only is post-discharge monitoring after CS rare, it remains unclear whether POAF in this specific setting should be classified as a transient phenomenon or a serious precursor of short- or long-term cardiovascular outcomes (138).

Furthermore, PPG-based recording of AF in the context of the compromised cardiovascular system after CS is complex because influencing factors such as blood volume and BP are often subject to considerable fluctuations and cannot be used for adequate assessment. Using machine learning methods, Hiraoka et al. developed an algorithm with an accuracy of 0.94 for AF detection in the study population of 80 patients (109). Consequently, future studies incorporating larger patient populations are needed to validate such findings in machine learning driven assessment in combination with PPG metrics of wearable health monitoring devices for AF detection in a postoperative setting.

Reduced HRV indicates a higher risk of arrhythmogenic death, adverse outcome after myocardial infarction, and severity of heart failure. Based on these current findings, HRV data may encompass a much broader range of predictive values in cardiac surgical patients, which the scientific community should further explore. However, the accuracy of HRV assessment as obtained by commercial wearable health monitoring devices is still uncertain. Despite the availability of remote and continuous HRV monitoring, ensuring accurate data delivery may be the major obstacle to overcome to realize the full potential of HRV as a prognostic tool to improve clinically relevant outcomes (32).

Bridging the gap between HRV assessment and postoperative AF in CS is important with an overall incidence 15% to 50% of patients undergoing CABG or heart valve surgery (139). One of the main causes of POAF is attributed to sympathetic hyperactivity, which is reflected by reduced HRV as a surrogate parameter. Therefore, HRV could be a valuable predictor for distinguishing between transient and long-lasting AF after CS. In terms of individualized patient care, understanding novel biomarkers such as HRV is essential. For example, HRV has been shown to be a valid surrogate parameter for treatment success of mesenchymal stem cell implantation after myocardial ischemia, as demonstrated by de Morais et al. The improvement in HRV after treatment was correlated with the decrease in infarct size and fibrosis (140).

Conclusions

In conclusion, the potential of HRV as a predictive value for AEs after CS is not just a hypothesis but has already been established as a validated predictor of AEs after myocardial infarction and the occurrence of sudden cardiac death. In addition, continuous monitoring of AF with wearable health monitoring devices has been shown to contribute to significant improvements in early detection. In addition, the researchers presented long-term monitoring of POAF as a promising option. The conclusion is that continuous monitoring with wearable health monitoring devices in CS patients should further explore HRV as a predictor to improve the detection of late-onset AF and, consequently, should be integrated into modern medical practice.

Acknowledgments

Funding: None.

Footnote

Reporting Checklist: The authors have completed the PRISMA reporting checklist. Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-23-19/rc

Peer Review File: Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-23-19/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-23-19/coif). M.M. has received institutional grants, research support, speaker honoraria, and travel compensation from Edwards Lifesciences, Symetis SA, Jena Valve, Boston Scientific, Medtronic, Abbott, and Novartis. M.A. is a proctor/consultant/speaker for Edwards Lifesciences, Abbott Laboratories, Boston Scientific, Medtronic, Zoll and Abbvie and received institutional research grants (Edwards, Abbott, Medtronic, LSI). The other author has no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Kumar A, Avishay DM, Jones CR, et al. Sudden cardiac death: epidemiology, pathogenesis and management. Rev Cardiovasc Med 2021;22:147-58. [Crossref] [PubMed]
2. Heart Surgery - Risks | NHLBI, NIH [Internet]. 2022 [cited 2023 Jul 22]. Available online: https://www.nhlbi.nih.gov/health/heart-surgery/risks
3. Shaban A, Leira EC. Neurologic complications of heart surgery. Handb Clin Neurol 2021;177:65-75. [Crossref] [PubMed]
4. Chikwe J, Athanasiou T. Complications of cardiothoracic surgery. In: Surgical Complications [Internet]. Published By Imperial College Press And Distributed By World Scientific Publishing CO.; 2007 [cited 2023 Jul 22]. p. 245–300. Available online: https://www.worldscientific.com/doi/10.1142/9781860948329_0008
5. GOV.UK [Internet]. 2020 [cited 2023 Jul 22]. Atrial fibrillation prevalence estimates for local populations. Available online: https://www.gov.uk/government/publications/atrial-fibrillation-prevalence-estimates-for-local-populations
6. Lopes LA, Agrawal DK. Post-Operative Atrial Fibrillation: Current Treatments and Etiologies for a Persistent Surgical Complication. J Surg Res (Houst) 2022;5:159-72. [Crossref] [PubMed]
7. Dekker JM, Schouten EG, Klootwijk P, et al. Heart rate variability from short electrocardiographic recordings predicts mortality from all causes in middle-aged and elderly men. The Zutphen Study. Am J Epidemiol 1997;145:899-908. [Crossref] [PubMed]
8. Liao D, Sloan RP, Cascio WE, et al. Multiple metabolic syndrome is associated with lower heart rate variability. The Atherosclerosis Risk in Communities Study. Diabetes Care 1998;21:2116-22. [Crossref] [PubMed]
9. Pumprla J, Howorka K, Groves D, et al. Functional assessment of heart rate variability: physiological basis and practical applications. Int J Cardiol 2002;84:1-14. [Crossref] [PubMed]
10. Dekker JM, Crow RS, Folsom AR, et al. Low heart rate variability in a 2-minute rhythm strip predicts risk of coronary heart disease and mortality from several causes: the ARIC Study. Atherosclerosis Risk In Communities. Circulation 2000;102:1239-44. [Crossref] [PubMed]
11. Tsuji H, Larson MG, Venditti FJ Jr, et al. Impact of reduced heart rate variability on risk for cardiac events. The Framingham Heart Study. Circulation 1996;94:2850-5. [Crossref] [PubMed]
12. Wolf MM, Varigos GA, Hunt D, et al. Sinus arrhythmia in acute myocardial infarction. Med J Aust 1978;2:52-3. [Crossref] [PubMed]
13. La Rovere MT, Pinna GD, Maestri R, et al. Short-term heart rate variability strongly predicts sudden cardiac death in chronic heart failure patients. Circulation 2003;107:565-70. [Crossref] [PubMed]
14. Sessa F, Anna V, Messina G, et al. Heart rate variability as predictive factor for sudden cardiac death. Aging (Albany NY) 2018;10:166-77. [Crossref] [PubMed]
15. BCS [Internet]. corporateName: [site name]; 2021 [cited 2023 Jul 22]. Smartwatches: The New Way to Diagnose Arrhythmias? Available online: https://www.britishcardiovascularsociety.org/resources/editorials/articles/smartwatches-new-way-diagnose-arrhythmias
16. Zilla P, Yacoub M, Zühlke L, et al. Global Unmet Needs in Cardiac Surgery. Glob Heart 2018;13:293-303. [Crossref] [PubMed]
17. Dias RD, Zenati MA, Stevens R, et al. Physiological synchronization and entropy as measures of team cognitive load. J Biomed Inform 2019;96:103250. [Crossref] [PubMed]
18. Ouyang H, Tian J, Sun G, et al. Self-Powered Pulse Sensor for Antidiastole of Cardiovascular Disease. Adv Mater 2017; [Crossref] [PubMed]
19. O'Neill MJ, Yoneda ZT, Crawford DM, et al. 2-Hydroxybenzylamine (2-HOBA) to prevent early recurrence of atrial fibrillation after catheter ablation: protocol for a randomized controlled trial including detection of AF using a wearable device. Trials 2021;22:576. [Crossref] [PubMed]
20. Barron LK, Moon MR. Impact of autonomic regulation on burnout and performance in thoracic surgery residents. JTCVS Open 2022;11:229-40. [Crossref] [PubMed]
21. Kennedy-Metz LR, Dias RD, Srey R, et al. Analysis of Dynamic Changes in Cognitive Workload During Cardiac Surgery Perfusionists' Interactions With the Cardiopulmonary Bypass Pump. Hum Factors 2021;63:757-71. [Crossref] [PubMed]
22. Georgiou KE, Dimov RK, Boyanov NB, et al. Feasibility of a New Wearable Device to Estimate Acute Stress in Novices During High-fidelity Surgical Simulation. Folia Med (Plovdiv) 2019;61:49-60. [Crossref] [PubMed]
23. Pron G, Ieraci L, Kaulback KMedical Advisory Secretariat, Health Quality Ontario. Internet-based device-assisted remote monitoring of cardiovascular implantable electronic devices: an evidence-based analysis. Ont Health Technol Assess Ser 2012;12:1-86. [PubMed]
24. Wei CY, Chen PN, Lin SS, et al. Using machine learning to determine the correlation between physiological and environmental parameters and the induction of acute mountain sickness. BMC Bioinformatics 2022;22:628. [Crossref] [PubMed]
25. Torkamani-Azar M, Lee A, Bednarik R. Methods and Measures for Mental Stress Assessment in Surgery: A Systematic Review of 20 Years of Literature. IEEE J Biomed Health Inform 2022;26:4436-49. [Crossref] [PubMed]
26. Danelich IM, Reed BN, Hollis IB, et al. Clinical update on the management of atrial fibrillation. Pharmacotherapy 2013;33:422-46. [Crossref] [PubMed]
27. Flint AC, Banki NM, Ren X, et al. Detection of paroxysmal atrial fibrillation by 30-day event monitoring in cryptogenic ischemic stroke: the Stroke and Monitoring for PAF in Real Time (SMART) Registry. Stroke 2012;43:2788-90. [Crossref] [PubMed]
28. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22:983-8. [Crossref] [PubMed]
29. Hindricks G, Potpara T, Dagres N, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): The Task Force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) Developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. Eur Heart J 2021;42:373-498. [Crossref] [PubMed]
30. Kim YG, Choi JI, Kim HJ, et al. A Watch-Type Electrocardiography Is a Reliable Tool for Detecting Paroxysmal Cardiac Arrhythmias. J Clin Med 2022;11:3333. [Crossref] [PubMed]
31. Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Front Public Health 2017;5:258. [Crossref] [PubMed]
32. Nenna A, Lusini M, Spadaccio C, et al. Heart rate variability: a new tool to predict complications in adult cardiac surgery. J Geriatr Cardiol 2017;14:662-8. [PubMed]
33. Shaffer F, Meehan ZM. A Practical Guide to Resonance Frequency Assessment for Heart Rate Variability Biofeedback. Front Neurosci 2020;14:570400. [Crossref] [PubMed]
34. Lane RD, McRae K, Reiman EM, et al. Neural correlates of heart rate variability during emotion. Neuroimage 2009;44:213-22. [Crossref] [PubMed]
35. McCraty R, Shaffer F. Heart Rate Variability: New Perspectives on Physiological Mechanisms, Assessment of Self-regulatory Capacity, and Health risk. Glob Adv Health Med 2015;4:46-61. [Crossref] [PubMed]
36. Graham A, Brown CH 4th. Frailty, Aging, and Cardiovascular Surgery. Anesth Analg 2017;124:1053-60. [Crossref] [PubMed]
37. Makary MA, Segev DL, Pronovost PJ, et al. Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 2010;210:901-8. [Crossref] [PubMed]
38. Lee JS, Auyeung TW, Leung J, et al. Transitions in frailty states among community-living older adults and their associated factors. J Am Med Dir Assoc 2014;15:281-6. [Crossref] [PubMed]
39. Shen MJ, Zipes DP. Role of the autonomic nervous system in modulating cardiac arrhythmias. Circ Res 2014;114:1004-21. [Crossref] [PubMed]
40. Quintana DS, Elstad M, Kaufmann T, et al. Resting-state high-frequency heart rate variability is related to respiratory frequency in individuals with severe mental illness but not healthy controls. Sci Rep 2016;6:37212. [Crossref] [PubMed]
41. Skytioti M, Elstad M. Respiratory Sinus Arrhythmia is Mainly Driven by Central Feedforward Mechanisms in Healthy Humans. Front Physiol 2022;13:768465. [Crossref] [PubMed]
42. Critchley WR, Yonan N, Shaw SM, et al. Heart rate after cardiac transplantation-lessons from the tortoise and the shrew. Transplantation 2013;95:259-65. [Crossref] [PubMed]
43. Physiologie. Pape H, Kurtz A, Silbernagl S, Hrsg. 9., vollständig überarbeitete Auflage. Stuttgart: Thieme; 2019. doi:10.1055/b-006-163285 [Internet]. [cited 2022 Dec 9]. Available online: https://eref-thieme-de.ez.srv.meduniwien.ac.at/ebooks/cs_10278468#/ebook_cs_10278468_cs1383 10.1055/b-006-163285
44. Berntson GG, Bigger JT Jr, Eckberg DL, et al. Heart rate variability: origins, methods, and interpretive caveats. Psychophysiology 1997;34:623-48. [Crossref] [PubMed]
45. HRV - Measuring Parameter - ANS Analysis [Internet]. [cited 2022 Nov 20]. Available online: https://www.ans-analysis.com/hrv/hrv-measuring-parameter.html
46. ResearchGate [Internet]. [cited 2022 Nov 19]. Table 1 . Definition and indication of HRV parameters assessed in this... Available online: https://www.researchgate.net/figure/Definition-and-indication-of-HRV-parameters-assessed-in-this-study_tbl1_329935249
47. Tarvainen MP, Niskanen JP, Lipponen JA, et al. Kubios HRV--heart rate variability analysis software. Comput Methods Programs Biomed 2014;113:210-20. [Crossref] [PubMed]
48. Kamath MV, Watanabe M, Upton A, editors. Heart Rate Variability (HRV) Signal Analysis: Clinical Applications. Boca Raton: CRC Press; 2012. 532 p.
49. Bigger JT Jr, Fleiss JL, Steinman RC, et al. Frequency domain measures of heart period variability and mortality after myocardial infarction. Circulation 1992;85:164-71. [Crossref] [PubMed]
50. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation 1996;93:1043-65. [Crossref] [PubMed]
51. Tsuji H, Venditti FJ Jr, Manders ES, et al. Reduced heart rate variability and mortality risk in an elderly cohort. The Framingham Heart Study. Circulation 1994;90:878-83. [Crossref] [PubMed]
52. Hadase M, Azuma A, Zen K, et al. Very low frequency power of heart rate variability is a powerful predictor of clinical prognosis in patients with congestive heart failure. Circ J 2004;68:343-7. [Crossref] [PubMed]
53. Schmidt H, Müller-Werdan U, Hoffmann T, et al. Autonomic dysfunction predicts mortality in patients with multiple organ dysfunction syndrome of different age groups. Crit Care Med 2005;33:1994-2002. [Crossref] [PubMed]
54. Shah AJ, Lampert R, Goldberg J, et al. Posttraumatic stress disorder and impaired autonomic modulation in male twins. Biol Psychiatry 2013;73:1103-10. [Crossref] [PubMed]
55. Shaffer F, McCraty R, Zerr CL. A healthy heart is not a metronome: an integrative review of the heart's anatomy and heart rate variability. Front Psychol 2014;5:1040. [Crossref] [PubMed]
56. Taylor JA, Carr DL, Myers CW, et al. Mechanisms underlying very-low-frequency RR-interval oscillations in humans. Circulation 1998;98:547-55. [Crossref] [PubMed]
57. Brown TE, Beightol LA, Koh J, et al. Important influence of respiration on human R-R interval power spectra is largely ignored. J Appl Physiol (1985) 1993;75:2310-7. [Crossref] [PubMed]
58. Thayer JF, Yamamoto SS, Brosschot JF. The relationship of autonomic imbalance, heart rate variability and cardiovascular disease risk factors. Int J Cardiol 2010;141:122-31. [Crossref] [PubMed]
59. Stein PK, Reddy A. Non-linear heart rate variability and risk stratification in cardiovascular disease. Indian Pacing Electrophysiol J 2005;5:210-20. [PubMed]
60. Lombardi F, Mäkikallio TH, Myerburg RJ, et al. Sudden cardiac death: role of heart rate variability to identify patients at risk. Cardiovasc Res 2001;50:210-7. [Crossref] [PubMed]
61. Huikuri HV, Stein PK. Heart rate variability in risk stratification of cardiac patients. Prog Cardiovasc Dis 2013;56:153-9. [Crossref] [PubMed]
62. Bauer A, Kantelhardt JW, Barthel P, et al. Deceleration capacity of heart rate as a predictor of mortality after myocardial infarction: cohort study. Lancet 2006;367:1674-81. [Crossref] [PubMed]
63. Kim SH, Lim KR, Seo JH, et al. Higher heart rate variability as a predictor of atrial fibrillation in patients with hypertension. Sci Rep 2022;12:3702. [Crossref] [PubMed]
64. Kališnik JM, Avbelj V, Vratanar J, et al. Cardiac autonomic regulation and PR interval determination for enhanced atrial fibrillation risk prediction after cardiac surgery. Int J Cardiol 2019;289:24-9. [Crossref] [PubMed]
65. Calkins H, Hindricks G, Cappato R, et al. 2017 HRS/EHRA/ECAS/APHRS/SOLAECE expert consensus statement on catheter and surgical ablation of atrial fibrillation: executive summary. J Interv Card Electrophysiol 2017;50:1-55. [Crossref] [PubMed]
66. Charitos EI, Stierle U, Ziegler PD, et al. A comprehensive evaluation of rhythm monitoring strategies for the detection of atrial fibrillation recurrence: insights from 647 continuously monitored patients and implications for monitoring after therapeutic interventions. Circulation 2012;126:806-14. Correction appears in Circulation 2012;126:e216.
67. Burns E, Buttner R, Buttner EB and R. Atrial Fibrillation [Internet]. Life in the Fast Lane • LITFL. 2018 [cited 2023 Jul 25]. Available from: https://litfl.com/atrial-fibrillation-ecg-library/
68. Raja JM, Elsakr C, Roman S, et al. Apple Watch, Wearables, and Heart Rhythm: where do we stand? Ann Transl Med 2019;7:417. [Crossref] [PubMed]
69. Medical Product Outsourcing [Internet]. [cited 2022 Nov 20]. CE Mark Granted for Wearable ECG Recorder. Available online: https://www.mpo-mag.com/contents/view_breaking-news/2022-01-10/ce-mark-granted-for-wearable-ecg-recorder/
70. MedTech Dive [Internet]. [cited 2022 Nov 20]. Withings gets FDA nod for AFib-detecting wearable, taking on Apple. Available online: https://www.medtechdive.com/news/withings-fda-nod-afib-detecting-wearable-apple-smartwatch/608119/
71. Bayoumy K, Gaber M, Elshafeey A, et al. Smart wearable devices in cardiovascular care: where we are and how to move forward. Nat Rev Cardiol 2021;18:581-99. [Crossref] [PubMed]
72. Hanne-Paparo N, Kellermann JJ. Long-term Holter ECG monitoring of athletes. Med Sci Sports Exerc 1981;13:294-8. [Crossref] [PubMed]
73. Kamišalić A, Fister I Jr, Turkanović M, et al. Sensors and Functionalities of Non-Invasive Wrist-Wearable Devices: A Review. Sensors (Basel) 2018;18:1714. [Crossref] [PubMed]
74. Inui T, Kohno H, Kawasaki Y, et al. Use of a Smart Watch for Early Detection of Paroxysmal Atrial Fibrillation: Validation Study. JMIR Cardio 2020;4:e14857. [Crossref] [PubMed]
75. Samol A, Bischof K, Luani B, et al. Single-Lead ECG Recordings Including Einthoven and Wilson Leads by a Smartwatch: A New Era of Patient Directed Early ECG Differential Diagnosis of Cardiac Diseases?. Sensors (Basel) 2019;19:4377. [Crossref] [PubMed]
76. Yang PS, Kim D, Sung JH, et al. Reduction of mortality by catheter ablation in real-world atrial fibrillation patients with heart failure. Sci Rep 2021;11:4694. [Crossref] [PubMed]
77. Ganesan AN, Shipp NJ, Brooks AG, et al. Long-term outcomes of catheter ablation of atrial fibrillation: a systematic review and meta-analysis. J Am Heart Assoc 2013;2:e004549. [Crossref] [PubMed]
78. Dai H, Zhang Q, Much AA, et al. Global, regional, and national prevalence, incidence, mortality, and risk factors for atrial fibrillation, 1990-2017: results from the Global Burden of Disease Study 2017. Eur Heart J Qual Care Clin Outcomes 2021;7:574-82. [Crossref] [PubMed]
79. Kim YG, Han KD, Kim DY, et al. Different Influence of Blood Pressure on New-Onset Atrial Fibrillation in Pre- and Postmenopausal Women: A Nationwide Population-Based Study. Hypertension 2021;77:1500-9. [Crossref] [PubMed]
80. Ghaferi AA, Birkmeyer JD, Dimick JB. Complications, failure to rescue, and mortality with major inpatient surgery in medicare patients. Ann Surg 2009;250:1029-34. [Crossref] [PubMed]
81. Jakobson T, Karjagin J, Vipp L, et al. Postoperative complications and mortality after major gastrointestinal surgery. Medicina (Kaunas) 2014;50:111-7. [Crossref] [PubMed]
82. Global patient outcomes after elective surgery: prospective cohort study in 27 low-, middle- and high-income countries. Br J Anaesth 2016;117:601-9. [Crossref] [PubMed]
83. Taenzer AH, Pyke JB, McGrath SP. A review of current and emerging approaches to address failure-to-rescue. Anesthesiology 2011;115:421-31. [Crossref] [PubMed]
84. DeVita MA, Smith GB, Adam SK, et al. "Identifying the hospitalised patient in crisis"--a consensus conference on the afferent limb of rapid response systems. Resuscitation 2010;81:375-82. [Crossref] [PubMed]
85. Goldhill DR, White SA, Sumner A. Physiological values and procedures in the 24 h before ICU admission from the ward. Anaesthesia 1999;54:529-34. [Crossref] [PubMed]
86. Buist M, Bernard S, Nguyen TV, et al. Association between clinically abnormal observations and subsequent in-hospital mortality: a prospective study. Resuscitation 2004;62:137-41. [Crossref] [PubMed]
87. Churpek MM, Yuen TC, Huber MT, et al. Predicting cardiac arrest on the wards: a nested case-control study. Chest 2012;141:1170-6. [Crossref] [PubMed]
88. Cuthbertson BH, Boroujerdi M, McKie L, et al. Can physiological variables and early warning scoring systems allow early recognition of the deteriorating surgical patient? Crit Care Med 2007;35:402-9. [Crossref] [PubMed]
89. Breteler MJM. Are current wireless monitoring systems capable of detecting adverse events in high-risk surgical patients? A descriptive study. Injury 2020;51:S97-S105. [Crossref] [PubMed]
90. Seesing MFJ, Borggreve AS, Ruurda JP, et al. New-onset atrial fibrillation after esophagectomy for cancer. J Thorac Dis 2019;11:S831-4. [Crossref] [PubMed]
91. Subbe CP, Kruger M, Rutherford P, et al. Validation of a modified Early Warning Score in medical admissions. QJM 2001;94:521-6. [Crossref] [PubMed]
92. Gardner-Thorpe J, Love N, Wrightson J, et al. The value of Modified Early Warning Score (MEWS) in surgical in-patients: a prospective observational study. Ann R Coll Surg Engl 2006;88:571-5. [Crossref] [PubMed]
93. Churpek MM, Adhikari R, Edelson DP. The value of vital sign trends for detecting clinical deterioration on the wards. Resuscitation 2016;102:1-5. [Crossref] [PubMed]
94. Vaishnav S, Stevenson R, Marchant B, et al. Relation between heart rate variability early after acute myocardial infarction and long-term mortality. Am J Cardiol 1994;73:653-7. [Crossref] [PubMed]
95. Kellett J, Murray A, Woodworth S, et al. Trends in weighted vital signs and the clinical course of 44,531 acutely ill medical patients while in hospital. Acute Med 2015;14:3-9. [Crossref] [PubMed]
96. Pereira T, Tran N, Gadhoumi K, et al. Photoplethysmography based atrial fibrillation detection: a review. NPJ Digit Med 2020;3:3. [Crossref] [PubMed]
97. Pereira T, Correia C, Cardoso J. Novel Methods for Pulse Wave Velocity Measurement. J Med Biol Eng 2015;35:555-65. [Crossref] [PubMed]
98. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. J Am Coll Cardiol 2014;64:e1-76. [Crossref] [PubMed]
99. Lloyd-Jones DM, Wang TJ, Leip EP, et al. Lifetime risk for development of atrial fibrillation: the Framingham Heart Study. Circulation 2004;110:1042-6. [Crossref] [PubMed]
100. Passman R, Beshai J, Pavri B, et al. Predicting post-coronary bypass surgery atrial arrhythmias from the preoperative electrocardiogram. Am Heart J 2001;142:806-10. [Crossref] [PubMed]
101. Mayr A, Knotzer H, Pajk W, et al. Risk factors associated with new onset tachyarrhythmias after cardiac surgery--a retrospective analysis. Acta Anaesthesiol Scand 2001;45:543-9. [Crossref] [PubMed]
102. Kalman JM, Munawar M, Howes LG, et al. Atrial fibrillation after coronary artery bypass grafting is associated with sympathetic activation. Ann Thorac Surg 1995;60:1709-15. [Crossref] [PubMed]
103. Deliargyris EN, Raymond RJ, Guzzo JA, et al. Preoperative factors predisposing to early postoperative atrial fibrillation after isolated coronary artery bypass grafting. Am J Cardiol 2000;85:763-4, A8.
104. Creswell LL, Schuessler RB, Rosenbloom M, et al. Hazards of postoperative atrial arrhythmias. Ann Thorac Surg 1993;56:539-49. [Crossref] [PubMed]
105. Borzak S, Tisdale JE, Amin NB, et al. Atrial fibrillation after bypass surgery: does the arrhythmia or the characteristics of the patients prolong hospital stay? Chest 1998;113:1489-91. [Crossref] [PubMed]
106. Funk M, Richards SB, Desjardins J, et al. Incidence, timing, symptoms, and risk factors for atrial fibrillation after cardiac surgery. Am J Crit Care 2003;12:424-33; quiz 434-5. [Crossref] [PubMed]
107. Ha ACT, Verma S, Mazer CD, et al. Effect of Continuous Electrocardiogram Monitoring on Detection of Undiagnosed Atrial Fibrillation After Hospitalization for Cardiac Surgery: A Randomized Clinical Trial. JAMA Netw Open 2021;4:e2121867. [Crossref] [PubMed]
108. Baman JR, Passman RS. The future of long-term monitoring after catheter and surgical ablation for atrial fibrillation. J Cardiovasc Electrophysiol 2022;33:1911-8. [Crossref] [PubMed]
109. Hiraoka D, Inui T, Kawakami E, et al. Diagnosis of Atrial Fibrillation Using Machine Learning With Wearable Devices After Cardiac Surgery: Algorithm Development Study. JMIR Form Res 2022;6:e35396. [Crossref] [PubMed]
110. Tison GH, Sanchez JM, Ballinger B, et al. Passive Detection of Atrial Fibrillation Using a Commercially Available Smartwatch. JAMA Cardiol 2018;3:409-16. [Crossref] [PubMed]
111. Harju J, Tarniceriu A, Parak J, et al. Monitoring of heart rate and inter-beat intervals with wrist plethysmography in patients with atrial fibrillation. Physiol Meas 2018;39:065007. [Crossref] [PubMed]
112. Aschbacher K, Yilmaz D, Kerem Y, et al. Atrial fibrillation detection from raw photoplethysmography waveforms: A deep learning application. Heart Rhythm O2 2020;1:3-9. [Crossref] [PubMed]
113. Rawles JM. What is meant by a "controlled" ventricular rate in atrial fibrillation?. Br Heart J 1990;63:157-61. [Crossref] [PubMed]
114. Samet P. Hemodynamic sequelae of cardiac arrhythmias. Circulation 1973;47:399-407. [Crossref] [PubMed]
115. Clark DM, Plumb VJ, Epstein AE, et al. Hemodynamic effects of an irregular sequence of ventricular cycle lengths during atrial fibrillation. J Am Coll Cardiol 1997;30:1039-45. [Crossref] [PubMed]
116. Lubitz SA, Faranesh AZ, Selvaggi C, et al. Detection of Atrial Fibrillation in a Large Population Using Wearable Devices: The Fitbit Heart Study. Circulation 2022;146:1415-24. [Crossref] [PubMed]
117. Howell SJ, Wanigasekera V, Young JD, et al. Effects of propofol and thiopentone, and benzodiazepine premedication on heart rate variability measured by spectral analysis. Br J Anaesth 1995;74:168-73. [Crossref] [PubMed]
118. Roberts AJ, Niarchos AP, Subramanian VA, et al. Systemic hypertension associated with coronary artery bypass surgery. Predisposing factors, hemodynamic characteristics, humoral profile, and treatment. J Thorac Cardiovasc Surg 1977;74:846-59. [Crossref] [PubMed]
119. Moss AJ, Zareba W, Hall WJ, et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med 2002;346:877-83. [Crossref] [PubMed]
120. Kusmirek SL, Gold MR. Sudden cardiac death: the role of risk stratification. Am Heart J 2007;153:25-33. [Crossref] [PubMed]
121. Kurths J, Voss A, Saparin P, et al. Quantitative analysis of heart rate variability. Chaos 1995;5:88-94. [Crossref] [PubMed]
122. Farrell TG, Bashir Y, Cripps T, et al. Risk stratification for arrhythmic events in postinfarction patients based on heart rate variability, ambulatory electrocardiographic variables and the signal-averaged electrocardiogram. J Am Coll Cardiol 1991;18:687-97. [Crossref] [PubMed]
123. Zuanetti G, Neilson JM, Latini R, et al. Prognostic significance of heart rate variability in post-myocardial infarction patients in the fibrinolytic era. The GISSI-2 results. Gruppo Italiano per lo Studio della Sopravvivenza nell' Infarto Miocardico. Circulation 1996;94:432-6. [Crossref] [PubMed]
124. Nolan J, Batin PD, Andrews R, et al. Prospective study of heart rate variability and mortality in chronic heart failure: results of the United Kingdom heart failure evaluation and assessment of risk trial (UK-heart). Circulation 1998;98:1510-6. [Crossref] [PubMed]
125. Mäkikallio TH, Huikuri HV, Hintze U, et al. Fractal analysis and time- and frequency-domain measures of heart rate variability as predictors of mortality in patients with heart failure. Am J Cardiol 2001;87:178-82. [Crossref] [PubMed]
126. Mäkikallio TH, Huikuri HV, Mäkikallio A, et al. Prediction of sudden cardiac death by fractal analysis of heart rate variability in elderly subjects. J Am Coll Cardiol 2001;37:1395-402. [Crossref] [PubMed]
127. Nikolic VN, Jevtovic-Stoimenov T, Stokanovic D, et al. An inverse correlation between TNF alpha serum levels and heart rate variability in patients with heart failure. J Cardiol 2013;62:37-43. [Crossref] [PubMed]
128. Kalisnik JM, Avbelj V, Trobec R, et al. Assessment of cardiac autonomic regulation and ventricular repolarization after off-pump coronary artery bypass grafting. Heart Surg Forum 2006;9:E661-7. [Crossref] [PubMed]
129. Lakusic N, Slivnjak V, Baborski F, et al. Heart Rate Variability after Off-Pump versus On-Pump Coronary Artery Bypass Graft Surgery. Cardiol Res Pract 2009;2009:295376. [Crossref] [PubMed]
130. de Godoy MF, Takakura IT, Correa PR, et al. Preoperative nonlinear behavior in heart rate variability predicts morbidity and mortality after coronary artery bypass graft surgery. Med Sci Monit 2009;15:CR117-22. [PubMed]
131. Compostella L, Russo N, Compostella C, et al. Impact of type of intervention for aortic valve replacement on heart rate variability. Int J Cardiol 2015;197:11-5. [Crossref] [PubMed]
132. Liu X, Fan J, Guo Y, et al. Wearable Smartwatch Facilitated Remote Health Management for Patients Undergoing Transcatheter Aortic Valve Replacement. J Am Heart Assoc 2022;11:e023219. [Crossref] [PubMed]
133. Skaf M, Makki N, Coressel A, et al. Rhythm Disturbances After Transcatheter Aortic Valve Replacement: A Strategy of Surveillance. Cardiovasc Revasc Med 2020;21:475-8. [Crossref] [PubMed]
134. Winter JL, Healey JS, Sheth TN, et al. Remote Ambulatory Cardiac Monitoring Before and After Transcatheter Aortic Valve Replacement. CJC Open 2020;2:416-9. [Crossref] [PubMed]
135. Rodés-Cabau J, Urena M, Nombela-Franco L, et al. Arrhythmic Burden as Determined by Ambulatory Continuous Cardiac Monitoring in Patients With New-Onset Persistent Left Bundle Branch Block Following Transcatheter Aortic Valve Replacement: The MARE Study. JACC Cardiovasc Interv 2018;11:1495-505. [Crossref] [PubMed]
136. Hoog Antink C, Mai Y, Peltokangas M, et al. Accuracy of heart rate variability estimated with reflective wrist-PPG in elderly vascular patients. Sci Rep 2021;11:8123. [Crossref] [PubMed]
137. Maesen B, Nijs J, Maessen J, et al. Post-operative atrial fibrillation: a maze of mechanisms. Europace 2012;14:159-74. [Crossref] [PubMed]
138. Piccini JP. Postoperative and Nonvalvular Atrial Fibrillation-Not So Different After All. JAMA Cardiol 2019;4:1147-8. [Crossref] [PubMed]
139. Steinberg JS. Postoperative atrial fibrillation: a billion-dollar problem. J Am Coll Cardiol 2004;43:1001-3. [Crossref] [PubMed]
140. de Morais Sdel B, da Silva LE, Lataro RM, et al. Mesenchymal Stem Cells Improve Heart Rate Variability and Baroreflex Sensitivity in Rats with Chronic Heart Failure. Stem Cells Dev 2015;24:2181-92. [Crossref] [PubMed]