| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Departments of *Anesthesiology and Intensive Care and Clinical Physiology, Turku University Hospital; Departments of Biostatistics and ¶Pharmacology and Clinical Pharmacology, Turku PET Centre, University of Turku; and Division of Cardiology, Department of Medicine, Oulu University Hospital, Finland
Address correspondence and reprint requests to Timo Laitio, MD, Department of Anesthesiology and Intensive Care, Turku University Hospital, P.O.B. 52, FIN-20521 Turku, Finland. Address e-mail to timo.laitio{at}tyks.fi.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
1), approximate entropy, and power-law relationship of relative risk interval variability (ß-slope). The high, low, very low, and ultra low frequency powers decreased significantly after the operation and remained at a significantly decreased level 6 wk and 6 and 12 mo after the operation than before (P = 0.01, P < 0.001, P < 0.001, and P < 0.001 for overall difference between the time points, respectively). The fractal scaling exponent 1 was at significantly more decreased 6 wk after (P < 0.05) CABG than before surgery but recovered to the preoperative level 6 mo after the operation. Long-term fractal organization (ß-slope) remained stable, but the overall complexity (approximate entropy) decreased toward more predictable HR dynamics during the study period (P < 0.01 after 1 yr). The predictive value of temporary and persistent long-term changes of the HR dynamics after CABG surgery for long-term outcome is not clear. |
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The immediate changes of HR variability (HRV) after CABG surgery have been studied (4,8,9). Time and frequency domain measures decreased significantly after CABG. Fractal scaling exponent 1 decreased significantly in the immediate postoperative phase of CABG surgery, but the power law slope ß, i.e., a measure of long-term fractal HR dynamics, and approximate entropy (ApEn) did not change significantly in the immediate postoperative phase (4).
The purpose of this explorative study was to extend the evaluation of the temporal pattern of HRV to 1 yr after CABG surgery. Time domain measures and a wide range of spectral components from high (HF) to ultra low frequencies (ULF) were used to assess the magnitude of HRV. In addition, methods derived from nonlinear dynamics (chaos theory) and fractal analysis, providing information about quantitative properties of HRV, were applied. The long-term temporal pattern of short-term fractal correlation properties and the overall complexity have not yet been studied.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The patients received their regular antianginal medication until the time of surgery. Anesthesia was induced and maintained with a fentanyl-midazolam combination and isoflurane. Moderate systemic hypothermia (core temperature 30°C–33°C), -stat pH management, and pump flow rate 2.4 L · min–1 · m2 were used. Antegrade cold cardioplegia with topical iced "slush" or antegrade intermittent cold blood (+10°C) cardioplegia were used during cardiopulmonary bypass. After the operation, patients were transferred to the ICU, and there lungs were mechanically ventilated until they had stable hemodynamic status and had recovered from anesthesia. ß-blockers were not routinely administered during the ICU stay.
Two-channel (bipolar leads CC5 and modified CM5) continuous 24-h ECG recording with a digital Holter (Oxford Medilog, Oxford Medical, Ltd., Woking, United Kingdom) device with a temporal resolution of 1024 Hz was performed in all patients. A blood sample for determination of creatine kinase MB isoenzyme (CK-MB) was obtained on arrival at the ICU, in the morning and evening of the first two postoperative days, and at discharge from the hospital or 7 days after surgery, whichever occurred first. An increase of CK-MB activity to >100 µg/L at any time after surgery or to >70 µg/L at any time after the first postoperative 12 h indicated perioperative MI. A 12-lead ECG was obtained before surgery, after arrival to the ICU, in the first and second postoperative mornings, and at discharge from hospital. All new Q-waves were identified. The occurrence of Q-wave was determined in one or more of the three lead groups: anterolateral (I, AVL, V6), posteroinferior (II, III, AVF), and anterior (V1-V5). A non–Q-wave MI was defined as the absence of a Q-wave but an increase of CK-MB activity to the above level. An additional ECG was obtained if perioperative MI was suspected or if there was an increase of CK-MB activity. ST segment depression >1.0 mm (0.1 mV) or increase >2.0 mm (0.2 mV) lasting 
1 min at 0.06 after the J-point in Holter data were defined as an ischemic change. The clinical data were analyzed by an experienced cardiologist who was blinded to the Holter data (4,6). Holter data were sampled digitally and then transferred from the Oxford Medilog scanner (Oxford Medical) to a computer for further analysis of HRV. A careful editing of the relative risk (RR) intervals was performed as previously described (4–7). All segments with >85% of qualified sinus beats were included, and at least 21 h of usable data were required for inclusion (4,10,11).
Mean HR and standard deviation of normal-to-normal RR intervals (SDNN) of 24-h data were used as conventional time-domain indices of HRV (12). An autoregressive model was used to estimate the power spectrum densities of HRV. The power spectra were quantified by measuring the areas in the following frequency bands: ULF power <0.0033 Hz, very low frequency (VLF) power from 0.0033 to 0.04 Hz, low frequency (LF) power from 0.04 to 0.15 Hz, and HF power from 0.15 to 0.4 Hz, as suggested (12).
Power-law relationship of RR interval variability was calculated for slow HR fluctuation using a previously described method (13). Briefly, a power spectrum was first computed. The resulting power spectrum was logarithmically smoothed in the frequency domain. Finally, a robust line-fitting algorithm of log (power) on log (frequency) was applied to the power spectrum over the frequency range of 10–4 to 10–2 Hz, and the slope of this line (ß) was calculated. This frequency band was chosen on the basis of previous observations regarding the linear relationship between log (power) and log (frequency) in this range (13). Fractal HR dynamics has a ß value of 1.0, and totally random dynamics has ß = 0. Another extreme example is a ‘Brownian noise‘ in which the ß-exponent is 2 (see also below). In this case, there are short-range correlations indicating that the RR interval at any given instant is strongly, and only, correlated to the previous interval (2,3,14).
ApEn is a variable that quantifies the regularity or predictability of time series data. It measures the logarithmic likelihood that runs of patterns, which are close to each other, will remain close in the next incremental comparisons. A higher regularity and predictability of HR dynamics (i.e., a greater likelihood of consecutive beats remaining close) produces smaller ApEn values, and conversely, random data produce higher values. Two input variables, m (number of observations) and r (filter level), were fixed to compute ApEn, and m = 2 and r = 20% of the standard deviation of the data sets were used for time series, as have been recommended based on previous findings of good statistical validity (13). The details of this method have been discussed elsewhere (14).
De-trended fluctuation analysis was used to quantify short-term fractal-like correlation 1 (window size 
11 beats) (1,2). The details of this method have been previously described (1–4,6). An 1-exponent value of 1.0 contains both random and highly correlated characteristics in RR interval time series and has been interpreted to indicate fractal HR dynamics, which is the case in healthy individuals (1–3). An 1-exponent value of 1.5 (i.e., ß = 2, see above) indicates high interbeat correlations, i.e., highly predictable and less complex HR dynamics. The value of 0.5 indicates totally random HR dynamics (1,2).
The ULF power and ß-slope of the power-law relationship of HRV were analyzed as a whole epoch of 24 h, as previously described (12,13). The remaining HRV measures were analyzed in segments of 8000 beats.
An average of 24 h was calculated whenever segments of 8000 beats were used. The Friedman’s nonparametric test for repeated measurements was used to test the overall difference between the HRV values in separate time points (only complete cases were used). The Wilcoxon’s matched pairs test with Bonferroni corrections was used in the post hoc comparisons between the preoperative and the postoperative values at different time points. The SAS System for Windows (SAS Institute, Cary, NC), release 8.2/2001, was used in the calculations. A P value <0.05 was considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
|
The postoperative HR did not differ significantly from the preoperative level at any time point. Time and frequency domain measures decreased significantly after the operation. The SDNN was significantly decreased 6 wk after the operation and recovered during the follow-up. The HF, LF, VLF, and ULF powers remained significantly decreased 6 and 12 mo after the operation than before (Table 2). A representative example of power spectrum is displayed in Figure 1.
|
|
The scaling exponent 1 was significantly lower 6 wk after surgery compared to the preoperative level but recovered to the preoperative level 6 mo after surgery (Table 2; Fig. 1). The ß-slope did not change after the CABG at any time point (Table 2). ApEn tended to decrease during the follow-up, and it was significantly lower 12 mo after surgery than before (Table 2).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We have evaluated the perioperative changes of HR dynamics in CABG patients (4). The time (i.e., SDNN) and frequency domain measures were significantly lower in the immediate postoperative phase after the surgery compared with the preoperative level. The LF and VLF powers of the first and second postoperative days were 80% less than the preoperative values, which were the largest reductions in the frequency domain measures. The HF power decreased 50% from the preoperative level. The fractal scaling exponent 1 decreased significantly immediately after the surgery, indicating increased randomness of HR dynamics, and was associated with prolonged ICU stay (i.e., more than 48 hours) and postoperative myocardial ischemia (4–6). The power-law slope ß and ApEn did not change in the immediate postoperative phase compared with the preoperative level.
An earlier study has suggested that the time and frequency domain measures recover completely within six months after CABG surgery (15). In the present study, the postoperative decrease of frequency domain measures persisted throughout the follow-up. The main reason for this discrepancy could be the smaller number of patients (i.e., 14 patients) and short-term measures (i.e., 15 minutes) of HRV in the earlier study. A recent study by Cygankiewicz et al. (16) of one-year follow-up with 56 patients showed that SDNN of the time domain and ULF, VLF, and HF powers of the frequency domain measures recover after one year in CABG patients, but LF power does not. In the present study, the largest decrease and the least recovery occurred in the LF power. One possible reason for the partly different results could be that 26% of their study patients had diabetes mellitus, which was an exclusion criterion in the present study. Therefore, these studies cannot be compared directly. Furthermore, baseline values of HF and ULF were markedly lower in their study. Thus, these two studies clearly suggest that there are some long-term changes in the autonomic nervous system after CABG. Furthermore, the predictive value of the persistently low LF power for long-term mortality requires further study, because decreased LF power has been shown to be the only independent predictor for all-cause mortality in 1028 patients of the Framingham study (17).
The physiological explanation for long-term attenuation of frequency domain measures is speculative. However, there are a few possible explanations: first, anesthesia can be excluded because of the persistent alterations of HR dynamics. Cerebral microembolism has been shown to occur during cardiopulmonary bypass, and various neurological deficits occur in 25% of CABG patients one year after surgery (18,19). It has been suggested that such cerebral consequences could modulate frequency domain measures after CABG surgery (20). Evidently, changes in the central nervous system contribute to the link between impaired baroreflex sensitivity and depressed HRV in elderly subjects because it normally triggers reflex adjustments after baroreceptor afferent discharge (21). In addition, there is strong evidence that the LF component of the power spectrum is influenced by the baroreflex gain because the LF power of both RR interval and arterial blood pressure variability is reduced after baroreceptor deafferentation (22,23). There is also evidence that common central mechanisms regulate both vagal and sympathetic modulation of the heart (24). In addition, the LF component may be modulated by sympathetic activity of central origin (25,26). Therefore, changes in the central nervous system and impaired baroreflex sensitivity resulting in altered feedback systems could cause a long-term reduction of HRV. This is supported by findings that baroreflex function is impaired after CABG surgery (27). The association of impaired baroreflex sensitivity and LF power is also supported by the study by Cygankiewicz et al. (16). Such changes may also be caused by long-term bed-rest, which is common in CABG patients. Furthermore, the long-term alterations of HR dynamics may be because of damage of the sinus node or nerve endings caused by the surgery or perfusion. Nevertheless, the explanation of the long-term HR dynamic alterations after CABG remains to be explored. In addition, the possible long-term effect of persistent reduction of LF power and possibly impaired baroreflex sensitivity on outcome requires further study.
Analysis methods derived from nonlinear dynamics, including chaos theory and fractals, have opened a novel approach for studying and understanding the characteristics of dynamic phenomena (1,2). Fractal organization seems to play a fundamental role in physiological structure and function (1–3). The normal HR dynamics have been shown to be fractal because they display scale-invariant fluctuations over a wide range of time scales. This indicates a long-range correlation between RR intervals, i.e., interbeat interval at every point is partially dependent on the intervals at all previous points. Therefore, changes from scale-invariant behavior toward behavior resembling either random fluctuations (white noise) or toward highly predictable less complex behavior might be physiologically deleterious (1). Furthermore, the short-term fractal scaling exponent 1 and the ß-slope have been shown to be powerful predictors of cardiac morbidity and mortality for cardiac causes in numerous recent studies (1,4–7,10,11). Our current results clearly showed a recovery of scaling exponent 1 to the preoperative level six months after surgery, and the ß-slope remained stable throughout the study period. These findings suggest that the fractal organization is able to recover after stressful events such as CABG surgery, although the overall HRV remains low, and the complexity is decreased. It has to be acknowledged, however, that our sample size was relatively small and was not based on any formal power analysis.
The later decrease of ApEn indicating decreased complexity of HR dynamics after CABG is also an interesting preliminary finding. In previous studies, physiological aging has been associated with a loss of complexity in previous studies (28,29). In the present study, the increased predictability indicated by decreased ApEn was surprisingly similar, as has been shown in physiological aging of healthy middle-aged (40–60 years old) to elderly (>60 years old) subjects (28). However, the physiological explanation of this finding remains unresolved.
It is evident that abnormalities in fractal HR behavior and LF power are associated with hemodynamic compromise and thereby indicate increased risk of mortality (4–7,10,11). CABG itself should improve rather than impair cardiovascular function. In this respect, the recovery of normal fractal HR dynamics after CABG is logical. In addition, it can be hypothesized that there is an increased risk for cardiac complications the first few weeks after CABG surgery, or as long as the short-term fractal correlation properties are decreased. Therefore, the predictive value of these short- and long-term changes of HRV after CABG surgery for long-term outcome requires further studies with a larger patient population.
|
Footnotes |
|---|
Accepted for publication November 15, 2005.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. S. Heffernan, J. J. Sosnoff, C. A. Fahs, K. K. Shinsako, S. Y. Jae, and B. Fernhall Fractal scaling properties of heart rate dynamics following resistance exercise training J Appl Physiol, July 1, 2008; 105(1): 109 - 113. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Laitio, J. Jalonen, T. Kuusela, and H. Scheinin The Role of Heart Rate Variability in Risk Stratification for Adverse Postoperative Cardiac Events Anesth. Analg., December 1, 2007; 105(6): 1548 - 1560. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
