Decreased HRV has been shown to be a predictor of death in CAD patients [22]. CABG improves prognosis in this group of patients [23], however little is known about the effect of the procedure on HRV. Although CABG leads to a positive clinical outcome, it is unclear whether CABG surgery simultaneously improves HRV. Published studies of the short-term changes in HRV post-CABG are listed in Table 4 (Ref. [18, 24, 25, 26, 27, 28, 29, 30]). In addition, some of these studies recorded HRV changes beyond 6 weeks post-CABG, these results are discussed in the section following this one.

All studies reported a decrease in traditional HRV parameters immediately after surgery, reaching their lowest value at 3–6 days post-CABG and remaining depressed even 6-weeks follow up after surgery [24, 25, 26, 27, 28, 29]. Thanh et al. [27] showed that decreased HRV before surgery was present in 28.6% of patients in 51.8% after seven days, 19.6% after three months, and 12.7% after six months. However, in contrast the finding of Bellwon et al. [30] showed that HF and percent of power in the HF band (%HF) increased significantly at 6-weeks post-CABG while in a standing position, but results of this study must be interpreted with caution, because of the brevity of the data- recording time (10 minutes) and because it is not clear that the organization of the HRV signal in the HF power was assessed and that makes it impossible to determine if this increase in HRV reflected a more disorganized cardiac autonomic functioning which has been called “erratic sinus rhythm” [31].

Summary of results: HRV decreases acutely after CABG. A study comparing HRV in patients undergoing uncomplicated CABG surgery with patients undergoing non-thoracic vascular surgery found that HRV indices were 40–50% lower in the CABG patients, suggesting that CABG itself reduces HRV [26]. Decreases in HRV 6 weeks after surgery may be associated with the acute effects of the operation including: surgical manipulations, prolonged aortic cross-clamping, use of ice slush, ischemia or dysfunction of cardiac ANS regulation post-CABG [32, 33, 34]. Anxiety about the danger and the pain of the surgery are also likely also to contribute to the reduction in HRV [35].

It has now been established, as described in the first section of this review, that HRV decreases acutely after CABG [24, 25, 26, 27, 28, 29]. However, the effect on HRV beyond 6-weeks post-CABG is still unclear. N = 4 studies, listed in Table 5 (Ref. [18, 24, 26, 27]), suggest a recovery of HRV after the initial drop. Demirel et al. [26] reported, in N = 12 patients, that standard HRV indices, which decreased after surgery, came back to pre-operative levels after CABG and then improved during the 1-year follow-up. At the 3-year follow-up, all HRV indices recovered to normal values. Soares et al. [24], in their study had a CABG group (N = 13) and control groups (patients who refused the operation (N = 9), and healthy subjects (N = 9). HRV was recorded during a 15-min period of paced breathing at 18 cycles/min one day before and 3, 6, 15, 30, 60, and 90 days after surgery. They found that paced-breathing-based HRV reached its lowest value three to six days after CABG and returned to its pre-operative levels at about 2 months post-CABG. Thanh et al. [27] demonstrated that, 6 months after surgery, the reperfusion and majority of HRV parameters reflecting injury to the ANS had improved.

Summary of results: The report of a later recovery of HRV to levels like those found in healthy subjects, regardless of the pre-operative status of the subjects and their post-operative clinical condition, suggest that decrease in HRV parameters 6 weeks after surgery is associated with acute effects of CABG and its improvement $>$6 weeks may be related to success of revascularization, although this too needs to be tested [24].

Atrial fibrillation (AF) is the most common complication after CABG and is related to a greater risk of heart failure, in-hospital stroke, longer hospital stay and rehospitalization [24, 36, 37, 38, 39]. Importantly, it has been shown that patients at high risk for AF may be recognized using HRV. However, assessment of relationships between HRV and AF risk have been limited to time or frequency domain parameters [40, 41]. Studies (N = 7) describing HRV as a predictor of complications including AF post-CABG are shown in Table 6 (Ref. [40, 42, 43, 44, 45, 46, 47]).

Research by Hogue et al. [40] based on post-CABG inpatient ECG monitoring and analyzing HRV indices during the hour before the onset of AF, showed that subjects who had AF after surgery had diminished heart rate complexity and more frequent atrial ectopy before the arrhythmia started. They also found that lower ApEn and greater heart rate were independently related to AF. Similarly, another study by Tarkiainen et al. [42] found that those who developed AF post-CABG had consistently reduced pre-operative DFA${}_{\alpha 1}$, reflecting a breakdown in fractal R-R interval dynamics and the greater randomness of R-R interval dynamics when breathing normally. These results suggest that the nonlinear HRV parameters evaluated pre-operatively may tell us some extra data about risk assessment of post-operative AF. Interestingly, Park et al. [43] showed that pre-operative abnormal heart rate turbulence (HRT) was related to worse both short-term and long-term outcomes after surgery.

In contrast, two other studies found that HRV analysis provided little value in predicting AF after CABG. Hakala et al. [44] concluded that the subjects at risk of AF development after surgery could not be recognized with used pre-operative short-term HRV parameters. In their study, assessed HRV indices did not differ between AF and sinus rhythm patients. However, they found that older age and higher body mass index were predictors of AF after surgery. Also, Chamchad et al. [45] in a study that involved 50 patients who underwent off-pump CABG, concluded that nonlinear HRV parameters may provide minimal value in prediction of AF after cardiac operation. In their study only the LF/HF ratio was decreased pre-operatively in subjects with new onset AF after CABG when compared to subjects without AF. Nevertheless, a more sophisticated real-time analysis of HRV from intensive care unit (ICU) ECG signals, as in the Hogue et al. [40] paper cited above may provide a powerful additional tool in identifying at-risk patients and permit timely preventive interventions.

Post-surgical HRV has also been explored as a predictor of length of stay in the ICU. Laitio et al. [46] assessed post-operative states of patients throughout their ICU stay. In this study, continuous 24-hr ECG recordings were obtained in 40 adult patients after CABG. They divided the patients into two groups, by length of stay in the ICU: Group A included 33 patients with (ICU stay $\le$48 hr) and Group B included 7 patients (ICU stay $>$48 hr). N = 20 patients had pre-operative (12–40 hr before surgery) continuous 24-hr ECG recordings. Three patients in Group A (short IUC stay) and two in Group B (longer stay) had incidents of AF. As expected, SD of RR intervals and the VLF power, ULF power, LF power, and HF power parameters diminished from pre-operative levels early after CABG. A Poincaré plot analysis, a marker for the organization of beat-to-beat HR patterns, showed greater randomness in beat-to-beat heart rates. in the longer ICU stay group Also, they found that DFA${}_{\alpha 1}$ dropped early after operation. Diminished DFA${}_{\alpha 1}$ (24-hrs after surgery) was the best HRV parameter for distinguishing subjects with short or prolonged ICU stays. They also reported that ApEn tended to increase after CABG. These data imply that subjects with less organized fractal-like HR dynamics may have more frequent post-operative complications, including AF or the need to require prolonged ICU care, but that this marker only appears after surgery.

Wu et al. [47], in their study, randomized 86 CABG patients into a control and an ischemic preconditioning group (IP). The IP group had induced 2-min ischemia and 3-min reperfusion periods. 24-hr ECG recordings were done both before and after CABG. They found that both standard and nonlinear HRV indices, i.e., DFA${}_{\alpha 1}$ were decreased after CABG both in control and IP group. Diminished pre-operative and post-operative DFA${}_{\alpha 1}$ predicted a greater incidence of post-operative AF and worse post-operative prognosis. However, the reduction in DFA${}_{\alpha 1}$ was less marked in the IP group, suggesting a possible measurable benefit of IP. No other differences were observed in HRV or ectopy parameters between the IP group and controls.

Summary of results: From the studies cited above, we can conclude that nonlinear HRV parameters may be better in detection of subtle abnormalities in cardiac autonomic control that traditional HRV indices and may give improved prediction of post-operative MI and post-operative AF in subjects after CABG [33, 34, 42, 46, 47]. Even though the precise mechanism is yet unclear, ischemic preconditioning appears to be effective in improving cardiac performance, decreasing cardiac troponin T levels, decreasing incidence of post-operative arrhythmias and improving post-operative outcome [44, 48, 49]. Although, the clinical use of nonlinear measures of HRV for risk assessment in patients undergoing major cardiac surgery needs to be further confirmed in prospective studies with larger samples, and the use of AI technologies could uncover additional measures, we believe that HRV analysis has shown promise of becoming a clinically useful new prognostic tool for post-operative clinical complications in the assessment of patients scheduled to undergo major cardiac surgeries. Further studies could combine HRV, clinical factors, ECG, peripheral blood biomarkers and/or myocardial properties to improve prediction of adverse clinical outcomes after CABG, as in other cardiovascular diseases [50, 51, 52, 53, 54]. Interestingly, progress in AF detection tools and methods, including wearable and mobile devices may be crucial also for CABG patients [55, 56, 57]. To study the agreement of ECG and photoplethysmographic signals (PPG)-derived HRV, increasingly available on personal devices like smart watches, Chen et al. [58] compared their agreement in patients over 1 year after CABG and demonstrated that pulse rate variability measures, might potentially be able to be useful in assessment the ANS regulation of patients after CABG because of the good agreement between the smart watch and ECG-derived for the majority of standard HRV parameters.

Multiple studies have demonstrated that CABG results in an acute decrease in HRV [26, 27, 59, 60]. Lower HRV is also seen in patients undergoing other cardiac surgery, e.g., valve reparation [61]. The probable reasons for the decrease in HRV after surgery include: a surgical intervention on the heart and adjacent structures, extended anesthesia, and cardioplegia. Importantly, as mentioned above, anxiety about the danger and pain of the surgery is likely also to contribute to the reduction in HRV [29]. Studies (N = 3) describing HRV as a predictor of mortality post-CABG are described in Table 7 (Ref. [62, 63, 64]).

Kalisnik et al. [29] concluded that changes in HRV parameters after off-pump CABG are related to adrenergic mobilization which is similar to that seen in on-pump surgeries. Lakusic et al. [61] demonstrated that in long-term follow up after CABG surgery HRV parameters were similar both in patients undergoing CABG off-pump and on-pump. Some studies have demonstrated that decreased HRV has no prognostic value in CABG patients, unlike in post-MI patients [62, 63, 65, 66]. Stein et al. [63] concluded that excluding post-CABG and diabetes mellitus subjects from HRV analysis strengthens the association between decreased HRV and risk of death. On the other hand, the recent study performed by Lakusic et al. [64] compared differences in mortality in patients after CABG with normal vs. decreased post-operative HRV. In their study, 24-hr Holter ECGs were obtained on all patients. Unlike some prior reports, results indicated that decreased post-operative SDNN was related to higher risk of death in CABG patients. However, it should be mentioned that patients with diminished HRV after CABG had lower left ventricular ejection fractions, smaller functional capacity and had more bypassed vessels (meaning that a higher degree of CAD, longer operation, and duration of cardioplegia) than patients with normal HRV. All of this may be related to decreased HRV per se. Moreover, one of the limitations of this study is that the HRV was not measured before and early after operation (the mean time from the CABG to measuring HRV was 3.7 $\pm$ 1.4 months).

Summary of results: It has been shown, using both short-term and 24-hr based measures that, in most subjects, HRV declined early after cardiac operation and recovered after a few months of surgery. However, to establish whether diminished post-operative HRV after CABG has prognostic implications, additional studies in a larger, well-characterized sample of subjects are needed. Lakusic et al. [64], as do other investigators, emphasized that patients with diminished HRV observed during long term follow-up after CABG should have access to long-term monitoring, diagnostic procedures, and appropriate treatment such as angiotensin-converting enzyme inhibitors (ACE-I), amiodarone and ivabradine [67, 68, 69].

Patients are exposed to autonomic dysfunction early after their operations, and this makes them more prone to early post-operative complications [26]. Increased parasympathetic tone has been shown to prevent cardiac arrhythmias [70, 71]. Patients with CAD who have decreased parasympathetic nervous activity are at risk of sudden cardiac death, making parasympathetic nervous activity (PNA) a marker with prognostic value [71]. Improvements in measures of parasympathetic functioning have been associated with lower cardiovascular risk [72]. However, sympathetic nervous activity may reflect the severity of heart failure and has prognostic value as well [73]. It has been shown that soon after surgery, sympathetic nervous activity (measured as plasma norepinephrine concentrations) recovers but parasympathetic nervous activity recovery (measured as HF power) lags it [74]. Studies have demonstrated that long-term outpatient cardiac rehabilitation (CR) can modify HRV in a favorable way in patients post-CABG [74, 75, 76, 77, 78]. CR has been shown to have positive effects on subjects with decreased parasympathetic tone before the beginning of rehabilitation post-CABG [78]. Exercise training induces adaptations in the autonomic demand needed during exercise. This adaptation occurs for central as well as peripheral neural pathways [75, 79]. Rehabilitation is aimed at prevention of complications post-CABG, shortening the length of hospital stay and to motivate patients to continue rehabilitation programs in an outpatient setting [80, 81]. Reviewing the scientific evidence today is not possible to sustain that CR as no role after CABG [82]. One study has shown that a short-term CR program, also significantly improves autonomic cardiac regulation at hospital discharge [80]. Amjadian et al. [83] demonstrated that both religious practice (Islamic and Qur’an), accompanied by doing homework, and rehabilitation at home with HRV biofeedback therapies, may be useful in subjects after cardiac surgery. It was also shown that various CR programs impacted cardiac autonomic control and length of hospital stay in subjects after surgery [80]. Some studies have shown no reduction in HRV attenuation in subjects after surgery who did not have any CR. Another study attributed this finding in part to prolonged bed rest, warranting early mobilization after CABG [84]. The intensity of the exercise regimen being administered has also been shown to impacts HRV. In one study [84] patients were randomized to an intensive training regimen post-CABG, characterized by supervised CR for one hour, three times a week and a home bicycle program for another three days a week for three months. Results showed that maximal workload capacity increased significantly and persisted one year later. SDNN and SDANN increased significantly, also persisting one year after the CR. A meta-analysis conducted by Kushwaha et al. [85] confirmed and reinforced the finding that exercise training improved selected HRV indices, e.g., rMSSD, SDNN, HF power, and LF/HF ratio. Studies (N = 8) describing use of HRV to assess effects of post-CABG interventions are shown in Table 8 (Ref. [74, 75, 76, 77, 78, 80, 81, 84]).

Summary of results: Different kinds of CR improve cardiac autonomic control in patients after CABG. However, in the analyzed studies, there are some methodological discrepancies regarding assessment of HRV, therefore future studies are needed to clarify HRV changes reflecting autonomic adaptation after exercise training in CABG patients and identify risk factors that prevent successful adaptation.

CABG improved the quality of life of CAD patients [86]. Nevertheless, about 6–46% of these subjects have been reported to have been exposed to psychological distress and complication related to it [87, 88]. Hallas et al. [89] in their study recruited 22 patients undergoing elective CABG for the first time and used the Hospital anxiety and Depression Scale (HAD), the Global Mood Scale (GMS) and the Dispositional Resilience Scale (DRI). Twelve-hour ECG recordings were obtained one week before surgery and two months after CABG, and HRV was assessed. They found that about 40% of subjects were anxious and depressed before their operation, but only 27% after CABG. Depression was the best predictor of reduced HRV assessed both before and after surgery.

Summary of results: Depression is related to decreased HRV. It is important to also recognize the role of psychological factors in prognosis after CABG operation and in non-emergent cases, tools to manage this distress might be beneficial to outcome.

As stated in sections 3.1 and 3.2, HRV decreases acutely after CABG and later recovers to pre-operative values. However, the process of undergoing CABG presents many confounding factors that might affect HRV beyond the effect of CABG itself. These confounding factors can be split into three general categories: (1) the perioperative medical management of CABG patients, (2) the process of undergoing cardiac surgery, and (3) characteristics of CABG patients that might affect HRV.

The goal of perioperative medical management of CABG is to minimize perioperative and short-term complications. According to the most recent AHA guidelines [90], medical management should include: insulin infusion, beta blockers and amiodarone. Insulin infusion is indicated to keep blood glucose levels under 180 mg/dL and reduce sternal wound infection. Glycemic control has been shown to improve HRV. In a larger cross-sectional investigation, higher fasting glucose was associated with lower time-domain measures of HRV (RR, rMSSD, SDNN) [91], and in a double-blind randomized controlled trial (RCT) with healthy individuals, manipulation of glucose levels outside the 70–90 mg/dL range caused decreases in SDNN, rMSSD and pNN50 [92].

In addition to insulin infusion, perioperative medical management for CABG also includes preoperative beta-blocker therapy since beta-blockers significantly reduce the risk of AF after surgery [93]. There is general agreement, in the literature, that beta-blockers improve both time-domain and frequency-domain measures of HRV in patients [93]. For example, in a large retrospective study of post-MI patients, use of beta-blockers was related to several increased time-domain HRV parameters (average NN, SDNN, SDANN and SDIDX) [65]. In subjects with decompensated heart failure, beta-blockers were associated with better frequency-domain HRV parameters (total power and ULF) [65, 94]. An experimental study with a rat model of chronic heart failure also reported similar findings, with atenolol and pindolol increasing both LF and HF power [95]. This study revealed augmentation of HRV with the use of beta-blockers supporting their use in patients with heart failure [95, 96]. Another element of perioperative medical management is amiodarone, which also reduces the risk of post-operative AF. There are few studies exploring the impact of amiodarone on HRV, but one experimental study found that amiodarone decreases frequency-domain measures (total power, LF power, LF/HF ratio) in a rabbit model [97].

Finally, traditionally, patients were given aspirin after surgery, to prevent graft occlusion and adverse cardiac events after CABG. The latest AHA guidelines recommend the use of dual antiplatelet therapy (DAPT) up to one year after CABG. However, available evidence is limited to small RCTs and the choice between the use of aspirin vs. DAPT remains unclear in CABG [98]. There are two RCTs that characterize the effect of aspirin on HRV, both done on healthy participants: one found that aspirin leads to higher HF and lower LF [99], while another found that aspirin does not have a significant effect on rMSSD [98], although the second may be underpowered. In another RCT, clopidogrel and ticagrelor did not have significant effects on either time- or frequency-domain measures of HRV [100].

Although HRV might be affected by perioperative medical management, it is also affected by the stress of the cardiovascular surgery process itself. For example, anesthetics generally depress both time- and frequency-domain measures of HRV, although different types of anesthetics have different effects. Inhaled anesthetics have been demonstrated to increase HR and decrease HRV (SDNN, LF and HF) [101], while the benzodiazepine remimazolam has been shown to have no effect on LF and HF [102]. Propofol has been shown to decrease HF across two studies, although it is unclear whether it has an effect on LF: one study concludes that it increases LF [102], while another does not find an effect [103]. Another IV anesthetic, thiopental, has similarly been shown to decrease HF and increase LF [104]. The choice of cardioplegia also has an impact on HRV. In an experimental study, warm blood cardioplegia led to significantly higher total power, LF and HF compared to cold crystalloid cardioplegia, although both predictably caused a decrease in HRV [104]. There was also a difference between on-pump versus off-pump CABG: with HRV measured one week and one month. Off-pump CABG led to significantly higher total power, LF and HF compared to on-pump [105], although after a mean of 3.7 months of follow up, there was no difference in HRV by surgery type [106].

There are potentially confounding factors associated with clinical characteristics in CABG patients that need to be considered when assessing changes in HRV parameters. In patients with CAD different HRV changes were observed among subjects with left ventricular ejection fraction (LVEF) over 50%, between 40–50% and below 40% [107]. Differences included normalized HF and VLF as well as LF peak during 5:00–6:00 and 18:00–19:00 [107]. The at-risk group with LVEF 40% had the highest normalized values of these HRV parameters. In another study, positive correlations were observed between SDRR, SDANN, SD, pNN50, LF and HF values and LVEF suggesting that better cardiac function is reflected in better HRV [108]. Traditional linear HRV, multiscale entropy parameters and DFA$\alpha$${}_1$ were decreased in subjects with heart failure compared to controls [109]. HRV was also related to the presence (patients with CAD vs. controls) and extent (one-, two-, three-vessel CAD vs. controls) of CAD as well as dependent on the Gensini score, a scale used for quantifying angiographic atherosclerosis [110]. A number of other cardiovascular risk factors, disorders and comorbidities have been linked to abnormal HRV parameters, including: older age, hypertension, diabetes, peripheral arterial disease, lacunar and atherosclerotic lesions, chronic obstructive pulmonary disease, chronic kidney disease, neuropathy and psychiatric disease [111, 112, 113]. Interestingly, time and frequency domain HRV parameters did not correlate with body mass index in young healthy volunteers [114]. However, in this group which consisted of young healthy adults, waist circumference correlated negatively with SDNN, rMSSD, pNN50, HF normalized units and positively with LF normalized units [114].

Summary of results: HRV can be affected by perioperative medical management (insulin infusion, beta-blockers and amiodarone), the cardiovascular surgery itself (anesthetics, cardioplegia) and clinical characteristics of the patients.