| 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, Turku, Finland; the
Departments of Biostatistics, and
Pharmacology and Clinical Pharmacology, and Turku PET Centre, University of Turku, Turku, Finland;
||Division of Cardiology, Department of Medicine, Oulu University Hospital, Oulu, Finland
Address correspondence and reprint requests to Timo Laitio, MD, Department of Anesthesiology and Intensive Care, Turku University Hospital, Turku, Finland. POB 52, FIN-20521 Turku, Finland. Address email to timo.laitio{at}tyks.fi
Abstract
Patients with myocardial ischemia after noncardiac surgery have a three- to ninefold increased risk of adverse cardiac events. In this study we tested the hypothesis that altered preoperative heart rate variability (HRV) predicts postoperative prolonged myocardial ischemia (>10 min) in elderly surgical patients. Thirty-two patients, age 60 yr or older, admitted to hospital for surgical repair of a traumatic hip fracture with preoperative night and daytime Holter recordings were included. Holter monitoring was initiated at arrival at hospital and continued until the third postoperative morning. Conventional HRV measures along with analysis of short-term fractal scaling exponent (
1) of RR intervals were assessed for night (from 2 AM to 5 AM) and day (7 AM to 12 AM) periods in each patient. Preoperative 1 was significantly lower (i.e., increased randomness in HRV) during the nighttime compared with daytime (mean ± SEM; 0.92 ± 0.08 versus 1.03 ± 0.06; P = 0.002) in patients with postoperative myocardial ischemia. Patients without ischemia had no such difference. In stepwise multivariate logistic regression analysis, increased preoperative night-day difference of 1 was the only independent predictor of postoperative prolonged ischemia. The odds ratio for an increase of 0.16 U in night-day difference of 1 (corresponding to interquartile range) was 7.7 (95% confidence interval, 1.9–51.4; P = 0.0018). Breakdown of fractal-like heart rate dynamics is predictive for postoperative prolonged myocardial ischemia in elderly patients having emergency surgery for traumatic hip fracture.
IMPLICATIONS: Night and daytime Holter recordings before surgical repair of traumatic hip fracture were analyzed with linear and nonlinear heart rate variability methods. Preoperatively increased randomness in heart rate variability was predictive for postoperative, silent prolonged myocardial ischemia. Prolonged myocardial ischemia increases the risk for adverse cardiac events.
Patients with myocardial ischemia after nonvascular and noncardiac vascular surgery have a three- to ninefold risk of adverse cardiac events, respectively, and cardiac complications account for more than half of the deaths (1–3). Especially, prolonged ischemia over 10 min is a strong predictor for postoperative death and myocardial infarction (4). The prevalence of perioperative myocardial ischemia in unselected hip fracture patients has been reported to be more than 30% (2). Complications are mainly attributable to ischemic events, pneumonia, and lung embolism. The 3-year mortality rate is more than 30%, and almost half of those who survive are permanently institutionalized (5). The American Heart Association has issued guidelines to identify patients at increased risk for postoperative adverse cardiac outcome preoperatively but diagnostic tools with better performance in risk stratification are still needed (6).
The autonomic nervous system plays a significant role in the pathophysiology of perioperative ischemia (7). There is evidence that sympathetic activation has an important role in the onset of adverse cardiac events (7). Adrenergic activity and plasma catecholamine levels change considerably in the postoperative period, which may predispose to myocardial ischemia by altering the relationship between myocardial oxygen demand and supply (1). Furthermore, increased sympathetic activation during rapid eye movement (REM) sleep has been suggested to be associated with the circadian pattern of ischemia occurring most frequently during early morning hours (8,9). Heart rate variability (HRV) measures from ambulatory electrocardiograph (ECG) recordings are widely used in the assessment of cardiovascular autonomic regulation. Recent studies suggest that newer measures of HRV, such as fractal analysis methods, can complement the traditional, time, and frequency domain HRV measures in risk stratification of patients with heart disease (10–15). These new dynamic analysis methods describe qualitative rather than quantitative properties of HRV. Fractal correlation properties exhibit long-range correlations between RR-intervals (i.e., interbeat interval at every time point is partially dependent on the intervals at all previous time points) (10). A change of fractal heart rate dynamics toward more random behavior has been able to predict morbidity and mortality in various populations (10–16). However, the predictive value of fractal analysis for myocardial ischemia has not been studied in surgical or nonsurgical patients.
This study was undertaken to test the hypothesis that preoperative alterations in heart rate dynamics predict postoperative prolonged ischemic episodes in patients undergoing surgery for traumatic hip fracture.
Methods
The joint Ethics Committee of Turku University Hospital and University of Turku approved the study protocol. All patients provided written informed consent. Thirty-two patients, age 60 yr or older, admitted to Turku University Hospital for surgical repair of a traumatic hip fracture with preoperative night and daytime Holter recordings were included in this study. The patients were from our published study (17) that evaluated the incidence of perioperative ischemia in 59 traumatic hip fracture patients treated with continuous epidural infusion or conventional parenteral opiates. Twenty-seven patients, not included in the present study, were operated soon after hospital intake and therefore had no preoperative night and day ECG recordings. Patients with other than sinus rhythm or with significant conduction abnormalities were excluded. All patients were operated under spinal anesthesia. Cardiac medications (excluding diuretics) were continued normally throughout the study period.
Preoperative 2-channel continuous Holter recording with an analog device with temporal resolution of 128 Hz (Series 8500; Marquette Electronics Inc., Milwaukee, WI) was initiated immediately after recruitment and continued until the third postoperative morning. Two bipolar leads were used: a modified V5 lead (fifth intercostal space at the left anterior axillary line) and a modified aVF lead (sixth intercostal space at the left midclavicular line). The corresponding reference electrodes were in the right and left first intercostal space at the midclavicular lines. A horizontal or down-sloping ST segment depression 
1.0 mm (0.1 mV) or an elevation 2.0 mm (0.2 mV) at 0.06 s after the J-point with over 10 min duration in Holter data were defined as reversible prolonged ischemic changes as previously described (4). All data were also analyzed with short ischemic episodes of at least 1 min. For each ischemic episode the maximum ST-deviation, its duration, and the area under the ST deviation x time curve (AUC) were determined. The ECG Holter data were sampled digitally and then transferred from the scanner (Oxford Medical Ltd., Clearwater, FL) to a computer for further analysis of HRV. A careful manual editing of the RR-interval series with inspection of the ECG data by deleting premature beats and noise was performed. All RR-intervals of suspected portions were printed on a 2-channel ECG at a paper speed of 25 mm/s to confirm the sinus origin of the RR-interval data as previously described (11–14,18).
Heart rate and standard deviation of all RR-intervals (SDNN) of 24-h data were used as conventional indices of HRV (18). An autoregressive modeling with a model order 20 was used to estimate power spectral densities of RR-interval time series. The power spectra were quantified by measuring the areas in the following frequency bands: very low frequency (VLF) power (0.0033–0.04 Hz), low frequency (LF) power (0.04–0.15 Hz), and high frequency (HF) power (0.15–0.4 Hz), as recently suggested (18). Detrended fluctuation analysis (DFA) was used to quantify fractal-like scaling properties of the time-series (10,19,20). The mathematical details of this method have been described previously (10,19,20). The deviations of each RR intervals from the average RR-interval are integrated over the selected window (1000 beats). Then the window is divided into smaller windows (time scales) and a least squares line fit is applied to the data in each window. This produces a "local" trend that is subtracted from the overall integrated time series, producing detrended time series. A root mean square fluctuation from this integrated and detrended time series is then repeatedly calculated using different time scales. Typically, there is a linear relationship between the logarithm of the fluctuation and the logarithm of the size of the time scale. The scaling exponent represents the slope of this line, which relates (log)fluctuation to (log)window size. The present heart rate correlation was defined for short-term fractal-like correlation 1 (window size 
11 beats) of RR-interval data, based on a previous finding of altered short-term heart rate behavior among elderly subjects (14). An exponent value of 0.5 means that there are no correlations between the RR-intervals as a result of random heart rate dynamics. An exponent value of 1.0 contains both random and highly correlated characteristics in RR-interval time series and has been interpreted to indicate fractal heart rate dynamics and has been documented for healthy heart rate dynamics (10,19,20).
All preoperative ECG data with ischemic ST segment changes were excluded from the HRV analysis. Association of the preoperative HRV measures and the postoperative ischemia were analyzed. Mean values of the HRV variables for preoperative RR-intervals as a whole epoch (i.e., 16 h of data before surgery), and night (from 2 AM to 5 AM) and day (7 AM to 12 AM, all patients were awakened at 7 AM) periods were calculated as an average of 1000 beat segments for each patient. The night-day difference was also calculated (i.e., the night value minus the day value). The period of nighttime was selected based on earlier findings that the majority of REM sleep occurs during late night between 2 AM and 5 AM (9,21).
Occurrence of at least one prolonged (over 10 min) postoperative ischemic episode was used as the criterion for division of the patients into two groups. The comparisons of the ischemia and the nonischemia group patient characteristics were performed using the Fisher’s exact test or the two-sample Student’s t-test. Associations of different HRV measures with postoperative ischemia were tested using a univariate logistic regression model. The predictive value of preoperative HRV measures for postoperative ischemia was also tested with multivariate analyses. First, all preoperative HRV measures were included in a stepwise multivariate logistic regression analysis. Statistically significant predictors were then included in further multivariate analyses including other potential explanatory factors divided into several subgroups (i.e., demographic factors, concomitant diseases, and concomitant medications, clinical variables listed in Table 1, analgesic regimen, and quality of sleep and pain scores using visual analog scale). Because of the small sample size, the significance for the final model was confirmed using exact techniques in calculations. The results were quantified with odds ratios (OR) and 95% confidence intervals (CI). The ORs were calculated corresponding to a change equal to interquartile range in the predictor variable. The goodness of fit of the final model was tested using the Hosmer-Lemeshow goodness-of-fit test. The area under the receiver operating characteristic (ROC) curve (c-index) was calculated as the criterion for sensitivity of the final logistic model. The SAS System for Windows, release 8.2/2001 (SAS, Chicago, IL)was used for the calculations. A P value of 0.05 was used as the threshold for significance.
|
One patient was excluded because of a continuous ischemia and 3 patients were excluded because of technical recording problems. Thus, 28 patients were included in the final analysis. There were 16 patients without perioperative prolonged ischemia, and 12 patients with postoperative prolonged ischemia. Five patients had preoperative ischemia and all of them also had postoperative ischemia. Patient characteristics and details of ischemia are shown in Tables 1 and 2.
|
1 measures were not found to be associated with ischemia. Preoperative 1 was significantly lower during the nighttime compared with daytime (mean ± SEM; 0.92 ± 0.08 vs 1.03 ± 0.06, P = 0.002) (Table 3) in patients with postoperative prolonged myocardial ischemia and the night-day difference of 1 was significantly associated with postoperative ischemia. Time and frequency domain measures did not show any significant association with ischemia.
|
1 was the only independent predictor of prolonged postoperative ischemia. The goodness-of-fit test of the model indicated a good fit (P = 0.163). The OR for an increase of 0.16 U in night-day difference of 1 (corresponding to interquartile range) was 7.7 (95% CI, 1.9–51.4; P = 0.0018). The respective values for all ischemia in 13 patients (one patient had one short ischemic episode of 1.7 min) were as follows: OR, 11.7; 95% CI, 2.3–108.5; P < 0.001.
Scaling exponent 1 was lower during nighttime than during daytime in 11 out of 12 patients with prolonged postoperative ischemia. The sensitivity, specificity, and predictive values of negative and positive tests of the negative night-day difference of 1 for postoperative prolonged ischemia were 92%, 69%, 92% and 69%, respectively. The area under ROC-curve (c-index) was 0.85, indicating high sensitivity of the final logistic model.
Discussion
The main finding of this study was that preoperatively increased nocturnal random heart rate dynamics (reduced short-term fractal exponent 1) had a predictive value for postoperative prolonged myocardial ischemia in patients with traumatic hip fracture. There were no predictive changes in the time and frequency domain measures of HRV preoperatively. This supports the concept that dynamical analysis methods of HRV can reveal clinically important abnormalities in heart rate dynamics that are not detected by conventional methods (10–15).
The normal heart rate time series have been shown to be fractal-like (1 = 1.0) (i.e., a subunit of RR-interval time series resembles the larger time scale). This indicates long-range correlation between RR-intervals; i.e., interbeat intervals are partially dependent on the intervals at previous points. Change from this scale-invariant behavior toward behavior resembling more random fluctuations (white noise, 1 = 0.5) with no correlation between interbeat intervals has been shown to be physiologically deleterious (10–16,19). Increased vulnerability to illness in elderly patients with increased randomness in the heart rate dynamics has been suggested to be attributable to a loss of physiologic responsiveness (10). This is also supported by one study that showed increased random short-term heart rate dynamics to predict cardiac mortality in a general elderly population (14). Current results showed an independent predictive value of increasing preoperative night-day difference of short-term fractal correlation properties for postoperative ischemia in traumatic hip fracture patients. Further studies are needed to explore possible predictive value of altered preoperative fractal heart rate dynamics for adverse cardiac outcome (e.g., myocardial infarction, death) in a larger population of high-risk surgery patients. Nevertheless, current results support earlier findings that breakdown of fractal organization is physiologically deleterious.
In this study, 12 of 13 patients with postoperative ischemia fulfilled the criterion of Landesberg et al. (4) of prolonged ischemia for more than 10 minutes. Previous studies have revealed improved HRV (i.e., HF power) and better outcome in patients treated with ß-adrenergic blockers (22). Furthermore, ß-adrenergic blockers decrease the incidence of postoperative myocardial ischemia and death in noncardiac surgery patients (23,24). One study showed preoperatively impaired HRV to predict non-fatal myocardial infarction or cardiac death in peripheral vascular surgery patients (25). In addition, the HF power has been shown to decrease during 60 minutes before ischemic events in nonsurgery patients (26). Thus, current results further support the strong predictive value of impaired HRV and an association between impaired HRV and prolonged postoperative myocardial ischemia. Prolonged ischemia has strong predictive value for postoperative death and myocardial infarction in vascular surgery patients (4). Although most of the predictive value of ischemia has been shown with vascular surgery patients, one study has shown perioperative myocardial ischemia to be associated with 2.6-fold risk for postoperative adverse cardiac events in elective hip arthroplasty patients (2). It has been suggested that alterations in HRV could partly explain the pathophysiology of perioperative myocardial infarction and also the protective effect of ß-adrenergic blockers (25). Therefore, the present preliminary results of the preoperative nighttime alterations in fractal heart rate dynamics may have clinical value in preoperative risk stratification for postoperative adverse cardiac events in high-risk patients.
From a mathematical point of view, conventional time and frequency domain measures describe the magnitude of HRV and fractal scaling measures describe relative changes in the characteristics of heart rate fluctuations. The conventional measures correlate significantly with fractal scaling measures when analyzed during strictly controlled external conditions and with fixed respiratory rate (27). However, the relationship is weaker during "free running" ambulatory ECG recordings because measurement of scaling exponents by the DFA method provide information on the scaling properties of heart rate fluctuations over several segmented time windows, whereas conventionally computed measures describe heart rate fluctuations only in one (e.g., HF, LF, and VLF powers) or in 2 (i.e., LF/HF ratio) predetermined time windows (27).
Fractal indices have quite small interindividual variation and short-term exponent values less than 1.0 are not usually seen among healthy subjects (28). In this study, most of the patients with postoperative myocardial ischemia had a scaling exponent 1 value less than 1.0 during the preoperative night, which was significantly less than in the daytime. On the contrary, fractal correlation properties of heart rate dynamics have been shown to be higher during sleep in healthy elderly subjects older than 60 years of age (28). Also in this study, patients without ischemia tended to have higher fractal correlation properties during the nighttime than during the daytime.
Physiological mechanisms for increased nocturnal randomness of heart rate dynamics in patients with postoperative ischemia are speculative. Increasing evidence supports the role of sympathetic activation behind this random heart rate dynamic. In a previous study, high norepinephrine levels, indicating sympathoexcitation, have been shown to be associated with random heart rate dynamics in heart failure patients (29). In addition, in young adults, IV infusion of norepinephrine has been shown to lead to increased randomness in heart rate dynamics (30). One study showed a decrease of fractal scaling exponent 1 during high levels of circulating norepinephrine levels (27). The increased sympathetic activation is supported by higher LF power during nighttime than during daytime in patients with ischemia (18). However, the difference did not reach statistical significance in our small sample. There was no such difference in patients without ischemia. There is also evidence for increased sympathetic activation during non-REM and REM sleep in elderly healthy subjects and in patients after myocardial infarction, whereas in healthy young subjects increased vagal dominance during non-REM sleep is well documented (9,31,32). Increased sympathetic activation during REM sleep could be involved in triggering ischemic events (7,9). In addition, the selected period of nighttime recording in the present study was based on earlier findings that the majority of REM sleep occurs during late night (i.e., between 2 AM and 5 AM) (9,21). Unfortunately, information of various sleep stages was not gathered in the present study. Further studies are therefore needed to specify temporal changes in fractal organization of heart rate dynamics and its association to autonomic regulation during different sleep stages in patients with increased risk for cardiac adverse events.
There are several limitations to this study. First, the sample size was rather small and thus the predictive value of altered fractal heart rate dynamics for postoperative myocardial infarctions, arrhythmias, and cardiac death has to be confirmed with a larger patient population. Second, although the current results support earlier findings that the newer dynamic measures complement the conventional measures, it is possible that some other HRV measure, in addition to the scaling exponent 1, could have predictive value for postoperative complications in a larger patient population. Third, there is no absolute reference of standard for detection of myocardial ischemia (1). Thus, the true predictive value of ischemic ST segment changes is possible to establish only by assessing long-term outcome (1).
In conclusion, this study shows preoperatively decreased nighttime fractal correlation properties of heart rate dynamics in patients who are vulnerable to developing myocardial ischemia in the early postoperative phase after nonvascular surgery. This observation may provide a new possibility to recognize such patients preoperatively. The clinical applicability of this measurement in risk stratification and for online monitoring of vulnerable patients should be further explored.
Acknowledgments
Supported, in part, by an unrestricted study grant from the Instrumentarium Science Foundation, Helsinki, Finland.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
