Download Heart rate variability and baroreflex sensitivity in subjects without

HEART RATE VARIABILITY
AND BAROREFLEX
SENSITIVITY IN SUBJECTS
WITHOUT HEART DISEASE
SIRK KU
PIK KUJÄ MSÄ
Department of Internal Medicine
Effects of age, sex and cardiovascular risk factors
O UL U 1 9 9 9
SIRKKU PIKKUJÄMSÄ
HEART RATE VARIABILITY AND
BAROREFLEX SENSITIVITY IN
SUBJECTS WITHOUT HEART
DISEASE
Effects of age, sex and cardiovascular risk factors
Academic Dissertation to be presented with the assent
of the Faculty of Medicine, University of Oulu, for public
discussion in Auditorium 10 of the University Hospital of
Oulu, on May 8th, 1999, at 10 a.m.
O U LU N Y LI O P IS T O , O U LU 1 99 9
Copyright © 1999
Oulu University Library, 1999
Manuscript received 13.4.1999
Accepted 15.4.1999
Communicated by
Docent Juha Hartikainen
Docent Markku Mäkijärvi
ISBN 951-42-5227-6
ALSO AVAILABLE IN PRINTED FORMAT
ISBN 951-42-5226-8
ISSN 0355-3221
(URL: http://herkules.oulu.fi/issn03553221/)
OULU UNIVERSITY LIBRARY
OULU 1999
Abstract
Healthy subjects show wide interindividual variation in their heart rate behavior, but the factors
affecting heart rate dynamics are not well known. This research was undertaken to evaluate heart rate
variability (HRV) and baroreflex sensitivity (BRS) in a large random sample of subjects without
evidence of heart disease, and to estimate the relation of heart rate behavior to age, sex and
cardiovascular risk factors.
Short-term HRV was analyzed from 15-minute periods of standardized recording in supine and
upright positions using time and frequency domain measures, and BRS was calculated using the
Valsalva maneuver in an original randomly selected population of 600 hypertensive and 600 control
middle-aged subjects. In addition, HRV was analyzed from the same segments using new measures
based on fractals and complexity (chaos theory) of R–R interval dynamics in the same random
population, and from 24-hour period in 114 healthy subjects aged from 1 to 82 years.
Large interindividual variation was observed in the measures of HRV and BRS in middle-aged
subjects; coefficient of variation (CV) of the standard deviation of R–R intervals (SDNN) 39%
(54 ± 21 ms) and CV of BRS 49% (9.9 ± 4.9 ms/mmHg). In healthy middle-aged men, SDNN was
weakly related to age (r = -0.19, p < 0.01), HDL cholesterol (0.19, p < 0.01), serum insulin (-0.23,
p < 0.001) and triglyceride (-0.25, p < 0.001) levels. In women, SDNN was only related to insulin
levels (r = -0.23, p < 0.001). BRS was related to systolic blood pressure (r = -0.31 and -0.30, in men
and women respectively, p < 0.001 for both) and blood glucose (r = -0.25, p < 0.01) and serum
insulin levels (r = -0.34, p < 0.001) in women. Lesser intersubject variation was observed in the nonlinear measures of HRV; CV 14% of short-term scaling exponent (a1), a measure of fractal-like
correlation properties of HRV, (1.21 ± 0.17) and CV 12% of approximate entropy, a measure of
complexity, (1.13 ± 0.14). Neither a1 or ApEn was related to any risk factors. Women had lower
overall short-term HRV (p < 0.01) and BRS (p < 0.001), but a higher spectral high-frequency
component of HRV, higher ApEn and lower a1 (p < 0.001 for all) compared to men. The impairment
in overall HRV was confined to the hypertensive subjects with metabolic features of the insulin
resistance syndrome (IRS, n = 69), but the BRS and spectral high-frequency component were also
impaired in hypertensive subjects without IRS compared to normotensive subjects. The 24-hour
cardiac interbeat interval dynamics changed markedly from childhood to old age. Children showed
similar complexity and fractal correlation properties of R–R intervals as young adults. Healthy aging
resulted in R–R interval dynamics with higher regularity and predictability and altered fractal
scaling.
The traditional measures of HRV and BRS are weakly related to many cardiovascular risk factors in
subjects without heart disease, but the interindividual variation of HRV and BRS is only partly
explained by these factors, suggesting a genetic background of the intersubject variation in
cardiovascular autonomic regulation. The new dynamical measures of HRV show less
interindividual variation than the conventional measures of HRV in healthy subjects and are not
related to cardiovascular risk variables, suggesting that these dynamical measures quantify the
"intrinsic" capacity of a healthy cardiovascular control system without the significant influence of
life-style, metabolic or demographic variables. However, there are sex and age-related differences
also in the fractal and complexity measures of heart rate behavior.
Keywords: cardiovascular regulation, electrocardiography, non-invasive methods.
Acknowledgements
This work was carried out at the Department of Internal Medicine, University of Oulu,
during the years 1992-1999.
I owe my deep gratitude to Professor Antero Kesäniemi, M.D., who both as the Head
of the Department and as an inspiring co-worker has supported this project in many ways.
I am deeply grateful to Docent Juha Hartikainen, M.D., from the University of
Kuopio, and Docent Markku Mäkijärvi, M.D., from the University of Helsinki, for their
thorough and efficient review of this manuscript and for all insightful and valuable
comments and suggestions to improve it.
I can hardly put into words my gratitude to Professor Heikki Huikuri, M.D., my
doctoral supervisor, without whom this project would never have been finished - if even
started. These years, full of both laughter and tears, have taught me so much of the
scientific world and of myself. It has been a great privilege having him as a guide.
The expertise, new ideas and, most of all, the broad supporting shoulders of Professor
Juhani Airaksinen, M.D., my other supervisor, have substantially contributed to this
project and are most warmly acknowledged.
I sincerely thank all colleges and the stuff in the Cardiovascular Laboratory for both
clinical and scientific guidance. The multiple help from Docent Markku Ikäheimo, M.D.,
and Markku Linnaluoto, M.Sc., have been of enormous value.
All help (especially statistical and spiritual) from my co-workers in the research
laboratory of the Department of Medicine, Professor Markku Savolainen, M.D., Docent
Mauno Lilja, M.D., Heikki Kauma, M.D., Asko Rantala, M.D., and also the other staff, is
greatly acknowledged.
I would not have survived without our special research team: Timo Mäkikallio, M.D.,
Juha Perkiömäki, M.D., Mikko Tulppo, Ph.D., Jari Tapanainen, M.D., Aino-Maija Still,
M.D., and Tapio Seppänen, Ph,D. Especially, I want to thank Antti Ylitalo, M.D., for
sharing all the highs and lows with me during this project. Both working and gossiping
with Päivi Karjalainen, R.N., Pirkko Huikuri, R.N., Miss Anne Lehtinen, Mirja Salo,
M.Sc., Marja Hietaniemi, R.N. and Miss Tuija Ranta has been most relaxing and fruitful.
The expertise of my co-authors Professor Ary Goldberger, M.D., Chung-Kang Peng,
Ph.D., Professor Leif Sourander, M.D., Ismo Räihä, M.D., Pauli Puukka, M.Sc., and
Antti Reunanen, M.D., is most sincerely acknowledged.
I warmly thank Sirkka-Liisa Leinonen, Phil.Lic., for promptly and efficiently revising
the English language of this summary.
I want to express my warmest thanks to my mother and father, Saara and Tauno Ojala,
for all the love, support and encouragement, which they have unselfishly given me and
my sister Johanna all our lives. The continuous baby-sitting during the last two and a half
years has also been both warmly welcomed and essential to complete this thesis.
The understanding love and support from my husband Mikko has been irreplaceable
during these years. Both he and our 2- and 1-year old sons, Aapo and Tuomas, have
certainly given me many, most rewarding roles in life beyond the field of medicine and
science.
This work was financially supported by the Aarne and Aili Turunen Foundation, the
Ida Montin Foundation, the Finnish Medical Foundation, the Finnish Foundation for
Cardiovascular Research, the Medical Council of the Academy of Finland and the
Finnish Life and Pension Insurance Companies.
Oulu, April 1999
Sirkku Pikkujämsä
Abbreviations
α1
α2
β
ApEn
BRS
CAD
ECG
HF
HRV
IRS
LF
SDANN
SDNN
ULF
VLF
short-term scaling exponent of fractal-like correlation properties
intermediate-term scaling exponent of fractal-like correlation properties
slope of the power-law relationship
approximate entropy
baroreflex sensitivity
coronary artery disease
electrocardiogra/m, -phic, -phy
high frequency
heart rate variability
insulin resistance syndrome
low frequency
standard deviation of average R-R intervals of analyzed segments
standard deviation of all R-R intervals
ultra low frequency
very low frequency
List of original communications
This thesis is based on the following publications, which are referred to in the text by
their Roman numerals:
I
Pikkujämsä SM, Huikuri HV, Ikäheimo MJ, Airaksinen KEJ, Rantala AO, Lilja
M, Savolainen MJ, Reunanen A & Kesäniemi YA (1996) Relationship between
heart rate variability and cardiovascular risk factors in middle-aged males. Ann
Noninvas Electrocardiol 1:354-362.
II
Huikuri HV, Pikkujämsä SM, Airaksinen KEJ, Ikäheimo MJ, Rantala AO, Kauma
H, Lilja M & Kesäniemi YA (1996) Sex-related differences in autonomic
modulation of heart rate in middle-aged subjects. Circulation 94:122-125.
III
Pikkujämsä SM, Huikuri HV, Airaksinen KEJ, Rantala AO, Kauma H, Lilja M,
Savolainen MJ & Kesäniemi YA (1998) Heart rate variability and baroreflex
sensitivity in hypertensive subjects with and without metabolic features of insulin
resistance syndrome. Am J Hypertens 11:523-531.
IV
Pikkujämsä SM, Mäkikallio TH, Sourander LB, Räihä IJ, Puukka P, Skyttä J,
Peng CK, Goldberger AL & Huikuri HV (1999) Cardiac interbeat interval
dynamics from childhood to senescence: Comparison of conventional and new
measures based on fractals and chaos theory. Circulation, in press.
V
Pikkujämsä SM, Airaksinen KEJ, Mäkikallio TH & Huikuri HV (1999)
Determinants of complexity and correlation properties of R-R intervals in middleaged subjects. Submitted for publication.
Contents
Abstract
Acknowledgements
Abbreviations
List of original communications
1. Introduction ................................................................................................................. 15
2. Review of the literature ............................................................................................... 16
2.1. Anatomy and physiology of cardiovascular autonomic regulation ....................... 16
2.2. Assessment of cardiac autonomic nervous function ............................................. 17
2.2.1. Measurement of HRV.................................................................................... 18
2.2.1.1. General ................................................................................................... 18
2.2.1.2. Time domain measures of HRV ............................................................. 18
2.2.1.3. Frequency domain measures of HRV ..................................................... 19
2.2.1.4. Dynamical analysis of HRV ................................................................... 20
2.2.1.5. Physiological background of different measures of HRV....................... 21
2.2.2. Measurement of BRS..................................................................................... 22
2.2.3. Reproducibility of measures of autonomic function ...................................... 23
2.2.4. Relation to heart rate and correlations between different measures of
autonomic function ........................................................................................ 23
2.3. Autonomic cardiovascular regulation in healthy subjects..................................... 24
2.3.1. Autonomic function in relation to changes in body posture and activity ....... 24
2.3.1.1. Effects of body posture on measures of autonomic function .................. 24
2.3.1.2. Acute effects of exercise on measures of autonomic function ................ 25
2.3.1.3. Circadian rhythm of measures of autonomic function ............................ 25
2.3.2. Interindividual variation of autonomic function in healthy subjects .............. 25
2.3.2.1. Effects of aging on autonomic function .................................................. 26
2.3.2.2. Sex-related differences in autonomic function........................................ 26
2.3.2.3. Effects of lifestyle on autonomic function .............................................. 26
2.3.2.4. Clinical and laboratory measures and autonomic function ..................... 27
2.4. Effect of drugs and disease states on autonomic nervous function ....................... 27
2.4.1. Effect of drugs on HRV and BRS.................................................................. 27
2.4.2. HRV and BRS in hypertension...................................................................... 28
2.4.3. HRV and BRS in CAD .................................................................................. 29
2.4.4. HRV and BRS in other disease states............................................................ 29
2.4.5. Prognostic significance of HRV and BRS ..................................................... 29
3. Purpose of the present study........................................................................................ 31
4. Subjects and methods .................................................................................................. 32
4.1. Study populations ................................................................................................. 32
4.2. Collection of clinical data..................................................................................... 36
4.3. Blood pressure measurements .............................................................................. 36
4.4. Laboratory methods.............................................................................................. 36
4.5. Echocardiographic measurements ........................................................................ 37
4.6. ECG recordings for HRV analysis........................................................................ 37
4.7. Measurement of HRV........................................................................................... 38
4.7.1. Time and frequency domain measures........................................................... 38
4.7.3. Fractal scaling and complexity measures....................................................... 39
4.8. Measurement of BRS............................................................................................ 39
4.9. Statistics................................................................................................................ 40
5. Results ......................................................................................................................... 42
5.1. Time and frequency domain measures of HRV in healthy middle-aged
subjects ................................................................................................................. 42
5.1.1. Interindividual variation in measures of HRV ............................................... 42
5.1.2. Sex-related differences in measures of HRV ................................................. 42
5.1.3. Relation of HRV to demographic, lifestyle and cardiovascular risk
factors ............................................................................................................ 44
5.2. BRS in healthy middle-aged subjects ................................................................... 45
5.2.1. Interindividual variation and relations to demographic, lifestyle and
cardiovascular risk factors of BRS ................................................................ 45
5.2.2. Sex-related differences in BRS...................................................................... 45
5.3. HRV and BRS in hypertensive subjects with and without insulin resistance
syndrome .............................................................................................................. 46
5.3.1. Overall HRV and BRS in hypertensive subjects with and without
metabolic features of insulin resistance syndrome and normotensive
controls .......................................................................................................... 46
5.3.2. Relationships of HRV and BRS to features of insulin resistance
syndrome ....................................................................................................... 46
5.4. Fractal correlation properties and complexity of HRV......................................... 48
5.4.1. Interindividual variation and relations to demographic, lifestyle and
cardiovascular risk factors in short-term scaling exponent and
approximate entropy ...................................................................................... 48
5.4.2. Effects of age on 24-hour fractal correlation properties and complexity
of HRV .......................................................................................................... 48
6. Discussion ................................................................................................................... 57
6.1. Interindividual variation of HRV and BRS in healthy middle-aged subjects ....... 57
6.2. Relations of HRV and BRS to demographic, lifestyle and laboratory factors
in healthy middle-aged subjects............................................................................ 57
6.3. Sex-related differences in HRV and BRS............................................................. 59
6.4. HRV and BRS in hypertensive subjects with and without insulin resistance
syndrome .............................................................................................................. 60
6.5. Dynamical measures of HRV in healthy subjects ................................................. 61
6.5.1. Dynamical analysis of R-R intervals.............................................................. 61
6.5.2. Effect of aging from childhood to advanced age on R-R interval
dynamics........................................................................................................ 61
6.5.3. Determinants of correlation properties and complexity of R-R intervals
in healthy middle-aged subjects..................................................................... 62
6.5.4. Physiological interpretation of R-R interval dynamics .................................. 63
6.6. Limitations of the study ........................................................................................ 63
7. Conclusions ................................................................................................................. 65
8. References ................................................................................................................... 66
1. Introduction
Heart rate is controlled most importantly by the autonomic nervous system to maintain
cardiovascular homeostasis. A healthy cardiovascular control system is able to adapt to
physiologic perturbations and changing conditions. During the past three decades,
impairment of autonomic cardiovascular regulation has been observed in many disease
states, including hypertension, diabetes and CAD (Airaksinen et al. 1987; Eckberg 1979;
Ewing et al. 1981). The relevance of this impairment for clinical cardiology was realized
in the late 1980s, when the results of impaired autonomic function were found to
independently predict mortality in postinfarction subjects (Kleiger et al. 1987; La Rovere
et al. 1988). The clinical importance of assessing cardiac autonomic function so far lies
in the risk stratification of postinfarction subjects.
The clinically applicable methods of assessing autonomic function are the
measurement of HRV and the evaluation of BRS. HRV analyzed as time and frequency
domain reflects primarily the physiological level of tonic autonomic regulation, whereas
BRS indicates the capacity of reflex autonomic regulation. New methods based on the
complexity and fractal scaling of R-R interval variability have been developed constantly
to uncover subtle abnormalities or alterations in time series data that are not otherwise
apparent (Lipsitz et al. 1992). Both low HRV and low BRS are associated with increased
cardiovascular risk (Kleiger et al. 1987; La Rovere et al. 1988). The new dynamical
measures of HRV may be more powerful predictors of adverse outcome in various
populations than the conventional measures (Bigger et al. 1996; Huikuri et al. 1998;
Mäkikallio et al in press).
Large interindividual variation of autonomic function has been described in healthy
subjects (Abdel-Rahman et al. 1994; Huikuri et al. 1990; Kupari et al. 1993; Molgaard et
al. 1994), but the factors affecting the autonomic control of heart rate are still largely
unknown. It is not known, for instance, whether alterations in HRV and BRS are
specifically associated with factors related to cardiovascular risk or determined by
genetic or other factors. Interindividual variation or possible relations to the risk factors
of the new dynamical measures of HRV have not been studied in detail.
The purpose of the present study was to investigate the interindividual variation and
its possible relations to age, sex and cardiovascular risk factors of HRV analyzed with
traditional and new dynamical measures and of BRS in subjects without heart disease.
2. Review of the literature
2.1. Anatomy and physiology of cardiovascular autonomic regulation
Heart rate is determined by the depolarization of the cardiac pacemaker, normally the
sinoatrial node. To maintain cardiovascular homeostasis, the sinoatrial node discharge
rate is carefully controlled by various regulatory systems in close connection with the
regulation of blood pressure. Beat-to-beat fluctuations in heart rate and blood pressure
reflect the dynamic response of the cardiovascular control systems to naturally occurring
physiological changes and challenges. The most important control system is the
autonomic nervous system, the function of which is organized on the basis of a reflex
arch (Opie 1998). The sensory receptors, which are abundantly found in the carotid sinus
and aortic arch, the lungs and the atria, sense the possible occurring perturbations. The
afferent pathways from the sensory organs travel within the vagus and sympathetic nerves
to the cardiovascular center in the medulla oblongata. The efferent vagal and
sympathetic pathways relay impulses from the cardiovascular centre to the effector
organs, impinging on both cardiac function (heart rate, atrioventricular conduction,
contractility) and the peripheral vasculature (arterial and venous vasomotor tone) (Opie
1998).
The sinus node is richly innervated with postganglionic sympathetic and
parasympathetic nerve terminals, which exert a continuous, though varied influence upon
heart rate. The two branches of the autonomic nervous system work in a co-ordinated
way, usually acting reciprocally, but sometimes synergistically on heart rate. The effects
of the vagal and sympathetic systems do not ordinarily summate algebraically, but show
complex interactions (Levy 1971; Opie 1998). Under resting conditions like during the
night, vagal tone prevails (Levy 1971) and the variations in heart period are largely
dependent on vagal modulation (Chess et al. 1975). Vagal stimulation by a release of
acetylcholine slows the sinus nodal discharge rate. Adrenergic stimulation mediated by
norepinephrine and epinephrine speeds up the sinus discharge rate (Opie 1998). Cardiac
responses to brief vagal bursts begin after a slow latency and dissipate quickly (in a few
milliseconds); in contrast, responses to sympathetic stimulation both commence and
dissipate slowly (in seconds). The rapid onset and offset of the responses to efferent
vagal stimulation allow for dynamic vagal modulation of heart rate, whereas the slow
17
temporal response to sympathetic stimulation precludes any beat-to-beat regulation at
high frequencies by sympathetic activity (Berne & Levy 1993).
The cardiovascular regulatory systems take compensatory actions when internal
(afferent) stimuli signal a need for regulation. Respiratory sinus arrhythmia is the term
used to describe the spontaneous fluctuation of R-R interval with respiration. Typically,
heart rate accelerates during inspiration and decelerates during expiration. Both the
frequency and amplitude of fluctuation vary, depending on the rate and depth of
respiration (Hirsch et al. 1981). The sensory organs may involve stretch receptors in the
lung tissue or the chest wall (Hirsch et al. 1981). Respiratory sinus arrhythmia correlates
with the vagal efferent traffic (Eckberg 1983; Katona et al. 1975). The Baroreceptor
reflex is a negative feedback system with many control loops, by which transient changes
in blood pressure are countered. A rise in blood pressure is sensed by the baroreceptors,
which are pressure-sensitive nerve endings mainly found in the wall of the aortic arch and
in the carotid sinuses. The baroreceptor reflex is stimulated, which results in a reduction
of heart rate and cardiac contractility and, thus, a fall in blood pressure. An initial
decrease in blood pressure has opposite effects. The baroreceptor reflex depends mainly
on reflex vagal activity (Eckberg et al. 1971; Eckberg 1980; Smyth et al. 1969). Heart
rate fluctuations with lower frequencies, the afferent stimuli of which are somewhat
unclear, may be related to thermoregulation and vasomotor tone (Akselrod et al. 1981;
Hyndman et al. 1971; Lindqvist et al. 1989; Sayers 1973). However, there are data to
suggest that they depend primarily on the presence of vagal outflow (Taylor et al. 1998).
All circulatory reflexes interact in a complex way and it may thus be difficult to
differentiate between the actions of the different control systems (Hainsworth 1998).
2.2. Assessment of cardiac autonomic nervous function
Various indirect quantitative measurements of cardiac autonomic function have been
developed, since no microneurographic techniques are available for direct measurements
of the actual cardiac autonomic nerve traffic. However, all of these indirect measures
may be subject to complex interactions with all circulatory reflexes. The development of
new technical equipment, i.e. digital, high-frequency, 24-hour potential, multichannel
ECG recorders, devices measuring continuous noninvasive blood pressure on a beat-tobeat basis, and microcomputers with sufficient capacity, have enabled the use of more
sophisticated measures of cardiac autonomic function, including spontaneous HRV and
BRS, in addition to the simple traditional bedside tests.
18
2.2.1. Measurement of HRV
2.2.1.1. General
The spontaneous fluctuations of R-R intervals (intervals between adjacent QRS
complexes resulting from sinus node depolarizations) are estimated using different
methods of HRV. The original continuous ECG signals are analogue-to-digital converted
into a microcomputer, which stores the length of R-R intervals versus cumulative time.
The tachogram generated is the source and basis for all analysis of HRV.
The original ECG data should meet some requirements. Artifact and ectopic beats
during the ECG recording can seriously interfere with many measures estimating HRV.
Although techniques have been developed to automatically exclude the abnormal
intervals, human editing of the tachogram is still usually required. However, none of the
methods used to exclude abnormal intervals are optimal. Frequent ectopy or artifacts
during the ECG recording may invalidate the HRV measurement either by interfering
with the measures or by a loss of normal intervals because of overediting.
The analyses of HRV with many linear (especially spectral) methods require the R-R
interval data to be stationary and sufficiently long (Malliani et al. 1991; Öri et al. 1992).
However, stationary conditions are unknown to biology, since the mechanisms
responsible for heart rate modulation do not remain unchanged for long periods. A
practical compromise has to be found between the length of event series and the
theoretical mathematical requirements. Linear detrending and filtering are preprocessing
techniques that make the data more stationary (Penaz 1978).
HRV can be analyzed from short-term (usually 5-minute) and long-term (usually 24hour) ECG recordings. The advantage of analyzing short-term HRV from short ECG
recordings is the limited duration of ECG monitoring, while the limitations include the
possible necessity of interventions and patient co-operation. The measures of short-term
HRV may fail to detect low-frequency oscillations and long-term trends that may have
clinical importance.
2.2.1.2. Time domain measures of HRV
The time domain measures of HRV are based on statistical or geometric analyses of the
lengths of intervals between successive normal complexes (Task Force 1996). The
statistical measures using the length of R-R intervals include the mean R-R interval, the
standard deviation of all R-R intervals (SDNN), the standard deviation of the average RR intervals calculated over short periods (SDANN) and the mean of the 5-minute
standard deviations of R-R intervals (SDNN index). SDNN reflects all cyclic
components in the analyzed period and thus, depends on the length of the period. It is
inappropriate to compare SDNN measures obtained from recordings of different
durations. The measures derived from differences between R-R intervals (RMSSD, the
square root of the mean squared differences of successive R-R intervals, NN50, the
number of interval differences of successive R-R intervals greater than 50 ms, and
19
pNN50, the proportion derived by dividing NN50 by the total number of R-R intervals)
estimate the high-frequency variations in heart rate and are correlated closely (Bigger et
al. 1989).
The series of R-R intervals can also be converted into a geometric pattern. The
variability is judged on the basis of the geometric properties of the resulting pattern by
using a basic measurement of the geometric pattern (i.e. quantitative analysis of Poincaré
plots (Tulppo et al. 1996)), interpolating the geometric pattern by a mathematically
defined shape (i.e. HRV triangular index (Malik et al. 1989)) and classifying the
geometric shape into several pattern-based categories (i.e. qualitative analysis of
Poincaré plots (Woo et al. 1994)). The major advantage of the geometric methods lies in
their relative insensitivity to artifacts and ectopic beats, while the major disadvantage is
the need for a reasonable number of R-R intervals and a longer ECG recording.
The methods for expressing overall HRV and its long- and short-term components are
not interchangeable. The use of SDNN (an estimate of overall HRV), SDANN (an
estimate of the long-term components of HRV) and RMSSD (an estimate of the shortterm components of HRV) as time domain measures of HRV are recommended (Task
Force 1996).
2.2.1.3. Frequency domain measures of HRV
A power spectral density analysis of the tachogram provides the basic information of how
power (variance) distributes as a function of frequency. The heart rate signal is
decomposed into its frequency components and quantified in terms of their relative
intensity (power) (Akselrod et al. 1981; Sayers 1973).
Both nonparametric (fast Fourier transform, FFT) and parametric (autoregressive
model estimation) methods are used. In most instances, the two methods provide
comparable results. The FFT spectra are characterized by discrete peaks for several
frequency components. The advantages of the FFT method are the simplicity of the
algorithm used and the high processing speed. However, FFT requires strict periodicity
of the data and a priori selection of the number and frequency range of the oscillatory
components. The advantages of the autoregressive method include smoother spectral
components and easy postprocessing of the spectrum. An accurate estimation of power
spectral density even on a small number of samples provides an additional advantage,
since the shorter time periods are more likely to meet the requirement of stationarity
(Malliani et al. 1991; Öri et al. 1992). It is recommended that the duration of the
recording should be at least 10 times the wavelength of the lowest frequency bound of the
spectral component investigated (Task Force 1996).
The power spectrum of healthy subjects consists of four major frequency bands. They
do not have fixed periods and the central frequencies may vary considerably. The limits
for the spectral components usually used (Task Force 1996) are: HF component 0.15-0.4
Hz, LF component 0.04-0.15 Hz, VLF component 0.003-0.04 Hz and ULF component
<0.003 Hz. The components estimate fluctuations with a periodicity of 2.5-7 s, 7-25 s, 25
s-6 min, and > 6 minutes respectively. Total power is represented by the total area under
the power spectral curve, and the power of individual frequency components by the area
20
under the proportion of the curve related to each component. The power of the ULF,
VLF, LF and HF components is usually expressed in absolute units (ms2). The ratio
between LF and HF components (LF/HF ratio) is also used (Malliani et al. 1991; Pagani
et al. 1986). LF and HF components may also be measured in normalized units by
dividing the power of the LF and HF components (in ms2) by the total power from which
the power <0.04 Hz has been subtracted and then multiplying by 100 (Malliani et al.
1991; Pagani et al. 1986). The normalization tends to minimize the effect of the changes
in total power on the values of the HF and LF components. Nevertheless, normalized
units should always be quoted with absolute values in order to describe completely the
distribution of power in the spectral components.
2.2.1.4. Dynamical analysis of HRV
Chaos theory and fractal behavior. Chaos refers to a system which has characteristics of
both periodicity and randomness. Periodic behavior is easily predictable, since it repeats
itself over a finite time interval. Random behavior, on the contrary, never repeats itself
and is unpredictable and disorganized. Although chaotic behavior looks disorganized, it
is really deterministic (Denton et al. 1990). A fractal system is a specific form of chaos.
It has the same structure on many measurement scales (i.e. self-similar, scale-invariant
structures) (Goldberger 1996). Nonlinear phenomena are involved in the genesis of HRV
(Sayers 1973). It has been suggested that healthy heart beat is chaotic and shows a fractal
form, which may break down with disease (Goldberger 1996). One of the characteristics
of a nonlinear system is that its components interact in a complex way. Multiple
mechanisms (sympathetic tone, parasympathetic tone, hormones, preload, afterload),
most of which interact and have long feedback loops, control the sinus node. Thus, one
may think that a healthy cardiovascular system is a near-perfect substrate for the
generation of chaos (Denton et al. 1990).
Nonlinear measures of HRV. The linear inverse power-law relationship of (log) power to
(log) frequency describes the distribution of the power spectral density (Bigger et al.
1996; Kobayashi et al. 1982; Saul et al 1987). The steeper (i.e. more negative) the slope
(β) of this power-law relationship is, the greater is the relative power in the lower
frequency ranges compared to the higher frequency ranges. β stands for the fractal-like
correlation properties of R-R interval data over very-low and ultra-low frequency bands
(10-4 to 10-2 Hz).
Detrended fluctuation analysis (DFA) quantifies the fractal-like correlation properties
of time series data (Iyengar et al. 1996; Peng et al. 1995). The scaling exponent α, with a
value near 1 indicating fractal-like behavior, can also be taken as an indicator of the
“roughness” of the time series: the smoother the time series, the larger α. One of the
advantages of the DFA method is that it minimizes the noise effects and, by removing
local trends remains relatively unaffected by nonstationarities.
Approximate entropy (ApEn) is a measure quantifying the regularity and complexity
of time series (Pincus et al. 1992; Pincus et al. 1994). ApEn quantifies the amount of
information needed to predict the future state of a system. Lower values of ApEn indicate
21
a regular (less complex) signal while higher values indicate irregularity (greater
complexity).
In addition many other nonlinear methods have been developed for assessing the
complexity and fractal-like properties of R-R interval time series (i.e. Lyapunov
exponents, Kolmogorov entropy, correlation dimension, Coarse Graining Spectral
Analysis). So far, none of the methods has been superior to the others or gained general
popularity.
2.2.1.5. Physiological background of different measures of HRV
HRV reflects oscillations in the activity of various regulatory systems associated with a
variety of factors, including respiration, baroreceptor reflexes, vasomotor control, and
thermoregulatory processes. However, the physiological mechanisms behind the different
measures of HRV vary. Especially, by using the spectral components of HRV, it may be
possible to discriminate between the effects of various cardiovascular control
mechanisms.
Time domain measures of HRV correlate well with vagal activity (Bailey et al. 1996;
Coker et al. 1984; Eckberg 1983; Hayano et al. 1991; Katona et al. 1975).
The spectral HF component is associated with respiration, and both amplitude and
peak frequency vary with respiration (Hirsch et al. 1981; Pagani et al. 1986). Both
clinical and experimental studies, including muscarinic receptor blockade, vagotomy and
electrical vagal nerve stimulation, have been shown that parasympathetic activity is the
major contributor to the HF component (Akselrod et al. 1981; Akselrod et al. 1985;
Bailey et al. 1996; Hayano et al. 1991; Pagani et al. 1986; Pomeranz et al. 1985).
Interpretation of the LF component is controversial. Vagal activity largely contributes
to the LF component based on the results of parasympathetic blockade (Akselrod et al.
1981; Akselrod et al. 1985; Chess et al. 1975). Experimental and clinical studies with
sympathomimetic or sympatholytic agents have yielded variable results. While an
association between the LF component and sympathetic activity has been found in some
studies (Akselrod et al. 1981; Akselrod et al. 1985; Pomeranz et al. 1985), no correlation
has been found in others (Chess et al. 1975). Interventions to increase sympathetic
activity, such as postural change, moderate exercise and mental stress, have increased the
LF component (Pagani et al. 1986; Pagani et al. 1991; Pomeranz et al. 1985;). The LF
component may be more interpretable if studied under conditions where activity and
body posture are controlled. LF oscillation of heart rate is reduced in subjects with very
high sympathetic activity, such as during heavy exercise (Arai et al. 1989; Breuer et al.
1993) and in heart failure (van de Borne et al. 1997). This is speculated to be due to
saturation of the LF oscillatory systems caused by the high sympathetic drive or the
mechanism to include a central effect of neurohumoral excitation (van de Borne et al.
1997). Some authors have interpreted the LF component to reflect sympathetic activity
either in absolute or normalized units and the LF/HF ratio to indicate sympathetic
outflow in humans or to mirror sympathovagal balance (Malliani et al. 1991; Pagani et
al. 1986). The idea of sympathovagal balance has been seriously criticized recently using
22
both mathematical and physiological arguments (Eckberg 1997). For instance, changes in
LF/HF ratio or in normalized LF and HF power have been interpreted as if sympathetic
and parasympathetic activity always change reciprocally. However, since sympathetic
and vagal nervous system act in parallel in some situations (i.e. cold immersion of the
face), the interpretation may not be physiologically justified in all situations (Eckberg
1997). Changes in baroreceptor activity caused by periodic changes in blood pressure
(Madwed et al. 1989; Penaz 1978) are thought to be the predominant determinants of the
LF component (Akselrod et al. 1985; Pagani et al. 1986; Pomeranz et al. 1985), although
there is data to suggest a central origin (Cooley et al. 1998). Recent studies where direct
muscle sympathetic nerve activity has been used as a reference index have suggested that
the LF and HF spectral components analyzed in normalized units may provide
information on sympathetic and vagal outflow, respectively, in subjects without structural
heart disease (Montano et al 1998; Pagani et al. 1997).
The background of the VLF and ULF components is not well known, although these
components account for about 95% of 24-hour total power. Atropin abolishes almost
completely all variation of heart rate, suggesting that vagal activity also largely
modulates these components (Akselrod et al. 1981; Pagani et al. 1986; Taylor et al.
1998). The renin-angiotensin system may play a minor role (Akselrod et al. 1981; Taylor
et al. 1998). The VLF and ULF components may also reflect changes in
thermoregulation due to peripheral blood flow adjustments (Lindqvist et al. 1989; Sayers
1973).
It must be borne in mind that HRV reflects the fluctuations in autonomic inputs to the
heart rather than the mean level of autonomic tone. Thus, both autonomic withdrawal
(blockade) and a saturating level of autonomic input lead to diminished HRV, although
the effects of the two situations on heart rate are reciprocal (Malik et al. 1993).
The physiological mechanisms behind the fractal-like R-R interval dynamics are still
unknown. The finding that denervated hearts are associated with steeper slopes (β) of the
power-law relationship (Bigger et al. 1996) supports the contention that at least this
scaling index is substantially influenced by the autonomic input of the heart.
In summary, parasympathetic control is the dominant contributor to all HRV. The
other possible mechanisms include the sympathetic nervous system, the renin-angiotensin
system, and humoral and thermoregulatory factors.
2.2.2. Measurement of BRS
BRS can be calculated with various indirect techniques by measuring the changes in heart
rate (in ms) against the changes in blood pressure (in mmHg). Most of the techniques
require laboratory conditions and specific interventions. Both blood pressure and heart
rate are measured on beat-to-beat basis. Changes in blood pressure are achieved by
vasoactive agents (nitroglycerin, phenylephrine), Valsalva maneuver, external neck
suction or neck pressure (Bristow et al. 1969; Goldstein et al. 1982; Gribbin et al. 1971).
The response of heart rate to these changes is then measured. The slope of the linear
relationship between the length of R-R intervals and blood pressure represents BRS.
There are no commonly approved recommendations as to which methods should be used,
23
although invasive testing with phenylephrine has been considered the “gold standard”.
Different techniques may measure different aspects of baroreceptor function and hence
give variable results. The average correlation coefficient between the different techniques
is 0.36 (Goldstein et al. 1982) and that between the phenylephrine and Valsalva methods
0.56 (Airaksinen et al. 1993). Both an inadequate blood pressure rise after the
interventions or frequent ectopic beats may invalidate BRS analysis. BRS reflects the
capacity of baroreceptor activation to increase vagal efferent activity, i.e. vagal reactivity
(Eckberg et al. 1971; Eckberg 1980; Smyth et al. 1969). Also sympathetic activity
(plasma norepinephrine levels) have been related to BRS (Hartikainen et al. 1995).
2.2.3. Reproducibility of measures of autonomic function
Intraindividual reproducibility of HRV measurements obtained from two 24-hour ECG
recordings is good in both normal subjects and subjects with heart disease (Hohnloser et
al. 1992; Huikuri et al. 1990; Kleiger et al. 1991; Van Hoogenhuyze et al. 1991). The
observed intraclass correlation coefficients have been high for both time and frequency
domain measures of HRV: 0.7-0.9 (mean R-R interval), 0.6-0.9 (SDNN), 0.8-0.9 (HF
power), 0.8-0.9 (LF power) (Hohnloser et al. 1992; Kleiger et al. 1991; Van
Hoogenhuyze et al. 1991). The intraindividual coefficient of variation for SDNN was 7±
6 % and that for the mean R-R interval 5 ± 5% (Huikuri et al. 1990). The reproducibility
of BRS measurements have also been high (r=0.73 between two successive tests
(Airaksinen et al. 1993), no significant differences emerged between two tests made up
to 15 months apart (Gribbin et al. 1971).
2.2.4. Relation to heart rate and correlations between different
measures of autonomic function
The time and frequency domain measures of HRV show significant correlations with the
mean R-R interval (r between 0.3-0.8 (Bigger et al. 1995; Molgaard et al. 1994; Kleiger
et al. 1991; Kupari et al. 1993; Van Hoogenhuyze et al. 1991)). BRS and nonlinear
measures of HRV do not significantly correlate with the resting heart rate (Airaksinen et
al. 1993; Gribbin et al. 1971; La Rovere et al. 1988; Mäkikallio et al. 1996).
Because of both mathematical and physiological relationships, there are significant
correlations between many time and frequency domain variables. The measures of pure
tonic vagal activity, i.e. HF, pNN50, RMSSD, correlate strongly (r between 0.94-0.97
(Bigger et al. 1989). The correlations between the other time and frequency domain
measures are weaker (r between 0.4 to 0.9 (Bigger et al. 1995; Kleiger et al. 1991)) and
those with nonlinear measures of HRV even weaker, if not nonexistent (Bigger et al.
1996; Huikuri et al. 1998; Molgaard et al. 1994). BRS and HRV correlate only
moderately (r = 0.6 (Bigger et al. 1989)) or have non-significant correlations (Farrell et
al. 1991; Hohnloser et al. 1994). Thus, despite the moderate degrees of correlation, the
24
different measures of autonomic function may give complementary information and are
hence not interchangeable.
The spectral measures of HRV analyzed from short ECG recordings (2 to 15 minutes)
correlate well (r between 0.58 and 0.88) with the values analyzed from 24-hour
recordings in post-MI patients (Bigger et al. 1993), while there is no data on healthy
subjects.
2.3. Autonomic cardiovascular regulation in healthy subjects
2.3.1. Autonomic function in relation to changes in body posture and
activity
The cardiovascular regulatory systems of healthy subjects are constantly responding to
changes in both internal and external conditions in a highly adaptive way. The changes in
the activity of various control systems due to changing conditions are reflected in the
measures of autonomic function. Thus, the possible changes in body posture and activity
should be considered in the study design when analyzing autonomic function, or at least
short-term HRV. On the other hand, the changes in the spectral components of HRV in
response to body posture or activity have encouraged the interpretation that various
spectral components or their derivates represent the two branches of autonomic system
separately. This may be inappropriate, however.
2.3.1.1. Effects of body posture on measures of autonomic function
Alterations in the autonomic function in an upright compared to supine position are
generally characterized by a decrease in parasympathetic activity and an increase in
sympathetic activity. A marked reduction in the HF component in both absolute and
normalized units has been observed in the standing position and during a passive tilt
compared to the supine position (Pagani et al. 1986; Pomeranz et al. 1985; Vybiral et al.
1989), but not in all studies (Lipsitz et al. 1990). The LF component in absolute and
normalized units and the ratio between the LF and HF components have markedly
increased in the upright position (Montano et al. 1994; Pagani et al. 1986; Pomeranz et
al. 1985; Vybiral et al. 1989), but, again, not in every study (Lipsitz et al. 1990). The
results on total power or total variance have varied even more from a decrease with tilt
(Pagani et al. 1986) through no difference with tilt (Vybiral et al. 1989) to an increase
with tilt (Lipsitz et al. 1990). Although the results have been inconsistent, some authors
have interpreted the increases in the LF component expressed in normalized units or the
ratio between the LF and HF components to reflect an increased sympathetic modulation
of heart rate or sympathovagal balance (Montano et al. 1994; Pagani et al. 1986; Pagani
et al 1991).
25
2.3.1.2. Acute effects of exercise on measures of autonomic function
A rapid vagal withdrawal occurs upon exercise (Robinson et al. 1966). The HF
component of HRV is markedly reduced during exercise (Arai et al. 1989; Breuer et al.
1993; Casadei et al. 1995; Yamamoto et al. 1991), further supporting the vagal origin of
the measure. However, the LF component and the total R-R interval variance are also
markedly reduced during heavy exercise (Arai et al. 1989; Breuer et al. 1993; Casadei et
al. 1995; Yamamoto et al. 1991), suggesting that even these measures are markedly
influenced by vagal activity in healthy subjects. Although sympathetic activity certainly
increases during exercise (Robinson et al. 1966), neither the LF component in
normalized units nor the ratio between the LF and HF components has increased
consistently, possibly reflecting the inability of these measures to assess sympathetic tone
in healthy subjects (Arai et al. 1989; Casadei et al. 1995). ApEn has been shown to
increase during exercise (Tulppo et al. 1996)
2.3.1.3. Circadian rhythm of measures of autonomic function
Autonomic activity shows a circadian rhythm with an increase in sympathetic tone during
the day and in parasympathetic tone at night in healthy subjects. The circadian rhythms of
different measures of HRV have varied in different studies. Higher values of pNN50 and
the HF component (Goldsmith et al. 1992; Molgaard et al. 1991; Molgaard et al. 1994)
have been observed at night. LF showed no difference between day and night in some
studies (Molgaard et al. 1994), was higher at night in some (Goldsmith et al. 1992) and
lower (in normalized units) at night in others (Huikuri et al. 1994). Total HRV was
increased at night in some studies (Huikuri et al. 1990; Huikuri et al. 1994; JensenUrstad et al. 1997), and decreased in one study (Molgaard et al. 1991). BRS was
increased during the nighttime (Hartikainen et al. 1993; Parati et al. 1995; Smyth et al.
1969). In summary, most data suggest that the indexes related to vagal activity are
increased at night. The results concerning the indexes possibly measuring sympathetic
activity are controversial. The possible effect of circadian variation must be taken into
account, especially while cardiac autonomic regulation is assessed from short-term
recordings.
2.3.2. Interindividual variation of autonomic function in healthy
subjects
Although healthy subjects are considered to have a highly adaptive cardiovascular
control system, they show marked interindividual variation in autonomic function
(Abdel-Rahman et al. 1994; Huikuri et al. 1990; Jensen-Urstad et al. 1997; Kupari et al.
1993). The coefficients of variation with different measures of HRV have been 12-15%
for the mean R-R interval, 24% for SDANN, 41-155% for LF power, 70-162% for HF
power and 20-63% for BRS (Abdel-Rahman et al. 1994; Huikuri et al. 1990; Jensen-
26
Urstad et al. 1997; Kupari et al. 1993, Töyry et al. 1995). Thus, more intersubject
variation has been found in HRV than in the mean heart rate. The reasons and
mechanisms of this extensive interindividual variation are not exactly known. Many
studies have tested the relations of various clinical, lifestyle and laboratory factors to the
alterations in autonomic cardiovascular regulation, usually on a univariate, but sometimes
on a multivariate basis.
2.3.2.1. Effects of aging on autonomic function
The effect of age on autonomic function has been uniformly observed in many studies.
Increased variation of heart rate occurs during childhood (Finley et al. 1995; Korkushko
et al. 1991). On the other hand, increasing age during adult life is associated with a
reduction in overall HRV (Bigger et al. 1995; Hayano et al. 1991; Hellman et al. 1976;
Korkushko et al. 1991; O'Brien et al. 1986; Shannon et al. 1987), a reduction in BRS
(Gribbin et al. 1971; Laitinen et al. 1998), a loss of complexity and altered fractal scaling
(Iyengar et al. 1996; Kaplan et al. 1991; Lipsitz et al. 1992; Ryan et al. 1994). Age has
also remained an independent determinant of autonomic function in studies with
multivariate analysis (Molgaard et al. 1994; Tsuji et al. 1996b). The age-related changes
may be due either to structural factors, such as a loss of sinoatrial pacemaker cells in
advanced age (Goldberger 1996) or a loss of arterial distensibility (Gribbin et al. 1971)
and functional changes, e.g. altered coupling between regulatory components
(Goldberger 1996).
2.3.2.2. Sex-related differences in autonomic function
There are several studies on gender-related differences in the autonomic control of heart
rate. Women have decreased BRS compared to men (Abdel-Rahman et al. 1994; Laitinen
et al. 1998). In studies on HRV, men have usually had higher overall HRV at lower
frequencies compared to women (Cowan et al. 1998; Jensen-Urstad et al. 1998; Liao et
al. 1995; Molgaard et al. 1994). Women may have higher complexity (ApEn) compared
to men (Ryan et al. 1994). The effects of various factors on autonomic function may
differ between the sexes (Kupari et al. 1993; Stein et al. 1997). Thus, although younger
men have higher overall HRV compared to women, the HRV in older subjects is
comparable between the sexes (Stein et al. 1997). The mechanisms of the gender-related
differences have not been widely discussed.
2.3.2.3. Effects of lifestyle on autonomic function
Many lifestyle factors may affect autonomic cardiovascular function. The effects of longterm physical activity on autonomic cardiovascular regulation have been addressed in
27
several studies, but the results have been inconsistent. In some studies, physical training
was associated with increased HRV and BRS in healthy subjects (Barney et al. 1988;
Dixon et al. 1992; Goldsmith et al. 1992; Molgaard et al. 1991; Molgaard et al. 1994).
However, no correlation has been found in some studies (Kupari et al. 1993; Seals et al.
1989). The possible beneficial effects of physical activity on autonomic cardiac
regulation are assumed to result from the vagotonic effects and parasympathetic
dominance produced by long-term physical training.
Smoking (Hayano et al. 1990; Jensen-Urstad et al. 1998; Molgaard et al. 1994;
Kupari et al. 1993) and abundant alcohol consumption (Malpas et al. 1991) have been
related to impaired autonomic function. However, the results have been variable between
the sexes, showing harmful effects of smoking only in women (Jensen-Urstad et al. 1998)
and possible beneficial effects of moderate alcohol consumption in women (Kupari et al.
1993). Subjects expressing features of type A personality show an increased ratio
between the LF and HF components compared to type B (Kamada et al. 1992).
2.3.2.4. Clinical and laboratory measures and autonomic function
Many clinical and laboratory measures have also been related to measures of autonomic
function in healthy subjects. The results have varied much, depending on the population
and the factor tested.
No correlations between blood pressure and HRV measures were found in young
healthy adults (Kupari et al. 1993). BMI had no correlations with HRV in a population
sample (Kupari et al. 1993), but had a negative correlation with HRV in women (JensenUrstad et al. 1998). An inverse relation of R-R interval variation and body fat was found
in obese males after beta blockade (Peterson et al. 1988).
Many laboratory values, including serum triglycerides, LDL cholesterol, and
leukocyte count, have univariate negative correlations with HRV. HDL cholesterol has
been directly related to HRV (Jensen-Urstad et al. 1998; Kupari et al. 1993). No
correlations were found with serum triglyceride or insulin levels (Kupari et al. 1993), but
total, VLF and LF power had negative univariate correlations with triglycerides (JensenUrstad et al. 1998).
2.4. Effect of drugs and disease states on autonomic nervous function
2.4.1. Effect of drugs on HRV and BRS
Many drugs interfere with autonomic function and thus possibly affect the measures of
HRV and BRS. Beta-blockers seem to enhance HRV in healthy subjects (Cook et al.
1991; Pagani et al. 1986), in patients with CAD (Niemelä et al. 1994) and in
postinfarction patients (Molgaard et al. 1993). The increase is most marked in the
measures of pure vagal activity (HF power, RMSSD), while an increase has been also
28
seen in total and LF power (Cook et al. 1991). Beta-blockers may increase BRS in
subjects with mild hypertension (Lucini et al. 1993), while no effect on BRS in subjects
with CAD was found (Airaksinen et al. 1994). ACE inhibitors may improve BRS in heart
failure patients (Osterziel et al. 1990) and HRV in postinfarction patients (Bonaduce et
al. 1994), while no effects on healthy subjects were observed (Kaufman et al. 1993).
Amiodarone seems to have no effect on HRV in patients with ventricular arrhythmias,
while an attenuation of HRV was seen with propafenone and flecainide (Zuanetti et al.
1991). Digoxin improved HRV in healthy subjects (Kaufman et al. 1993). Transdermal
scopolamine increased overall HRV, HF power and BRS in healthy subjects (DibnerDunlap et al. 1985; Vybiral et al. 1990), while diltiazem had no effect on HRV (Cook et
al. 1991).
2.4.2. HRV and BRS in hypertension
Previous studies on HRV and blood pressure have yielded variable results. No
differences compared to normotensives were found in some studies (Mancia et al. 1983).
Decreased measures of pure parasympathetic control (pNN50, HF power) have been
found in unmedicated hypertensive subjects (Langewitz et al. 1994) and in hypertensive
subjects with left ventricular hypertrophy (Chakko et al. 1993; Petretta et al. 1995),
suggesting decreased vagal tone in hypertension. However, other measures of overall
HRV, including SD of R-R intervals (Chakko et al. 1993; Huikuri et al. 1996), LF power
(Huikuri et al. 1996; Petretta et al. 1995) and VLF power (Huikuri et al. 1996) in
absolute units, have also been decreased in hypertensive subjects, possibly suggesting
altered function of other control mechanisms (sympathetic, renin-angiotensin system) as
well. A decreased HF component and an increased LF component, expressed as
normalized units were found in hypertensive subjects (Guzzetti et al. 1988), which
finding was interpreted by the authors as an increased sympathetic tone in hypertensive
subjects. Alterations in the circadian rhythm of HRV (Chakko et al. 1993; Dassi et al.
1991) and blunted responses of HRV measures to an upright posture (Guzzetti et al.
1988; Huikuri et al. 1996; Radaelli et al. 1994) have also been observed in hypertensive
subjects.
The results concerning the association between BRS and increased blood pressure
have been more uniform. An inverse relation between blood pressure and BRS has been
observed (Bristow et al. 1969; Gribbin et al. 1971; Sleight 1979). Decreased BRS was
seen in unmedicated hypertensive subjects (Eckberg 1979; Bristow et al. 1969; Radaelli
et al. 1994) as well as in patients with long-standing drug-treated hypertension (Ylitalo et
al. 1997).
In summary, altered cardiovascular autonomic control seems to be common in
hypertension. Most of the previous results on HRV and BRS indicate a decrease in both
tonic and reflex vagal activity with elevated blood pressure. However, even other
mechanisms may be involved in impaired autonomic cardiovascular modulation. There
are no studies where the effects on elevated blood pressure vs. the metabolic factors
possibly accompanying hypertension on HRV or BRS have been tested.
29
The alterations of autonomic cardiovascular control characterized by reduced vagal
activity and thus resulting in relative sympathetic dominance might contribute to the
development of essential hypertension (Eckberg 1979), and hence be one of the
pathophysiological mechanisms behind elevated blood pressure. However, the alterations
may also be secondary to long-standing hypertension and, for instance, sign of structural
alterations of the vessel wall (medial hypertrophy, endothelial damage) associated with
long-standing elevated blood pressure (Kingwell et al. 1995; Zanchetti et al. 1991).
2.4.3. HRV and BRS in CAD
Many studies have addressed the impaired autonomic modulation of heart rate in CAD.
Both HRV (Airaksinen et al. 1987; Hayano et al. 1990; Rich et al. 1988) and BRS
(Eckberg et al. 1971) are reduced in patients with CAD, and heart rate dynamics seem to
be altered in CAD patients (Mäkikallio et al. 1998). The circadian rhythm of HRV is
altered in patients with CAD (Huikuri et al. 1994). The alterations in autonomic function
have correlated with the severity of CAD in some studies (Hayano et al. 1991; La Rovere
et al. 1988), but not in others (Airaksinen et al. 1987; Rich et al. 1988). BRS and
measures of HRV are lower in post-MI subjects (Bigger et al. 1995) and their heart rate
dynamics is altered (Mäkikallio et al. 1996). The exact mechanisms underlying this
impairment in CAD are not known. Denervation of the afferent and efferent sympathetic
or parasympathetic fibers or alteration of their function after ischemic episodes or
infarction have been suggested (Zipes 1990).
2.4.4. HRV and BRS in other disease states
Impaired autonomic cardiovascular regulation has been observed in various other
pathologies of either the heart itself or the autonomic regulatory systems. Decreased
HRV has been seen in patients with congestive heart failure (Casolo et al. 1989; Nolan et
al. 1992; Saul et al. 1988; Van Hoogenhuyze et al. 1991; van de Borne et al. 1997), in
subjects vulnerable to life-threatening arrhythmias (Huikuri et al. 1992; Martin et al.
1987; Myers et al. 1986), post transplantation (Sands et al. 1989), and in patients with
diabetes (Ewing et al. 1981), alcoholism (Malpas et al. 1991), chronic renal failure
(Cloarec-Blanchard et al. 1992) or one of a variety of neurological disorders, such as
stroke (Korpelainen et al. 1996) or brain damage (Lowensohn et al. 1977).
2.4.5. Prognostic significance of HRV and BRS
The most important current application of HRV and BRS in clinical cardiology lies in the
risk stratification of postinfarction patients. Wolf reported in 1978 that postinfarction
subjects with sinus arrhythmia present had better prognosis (Wolf et al. 1978). Kleiger
30
reported in his classic work in 1987 HRV to be an independent predictor of mortality in
patients with acute myocardial infarction. The relative risk of mortality was 5.3 times
higher if 24-hour SDNN was <50 ms than if it was >100 ms (Kleiger et al. 1987). The
importance of autonomic cardiovascular regulation in predicting mortality after acute
myocardial infarction has later been confirmed by many others, using both decreased
HRV assessed with various methods (Algra et al. 1993; Bigger et al. 1992; Farrell et al.
1991; Hartikainen et al. 1996; Hohnloser et al. 1994; La Rovere et al. 1988; La Rovere
et al. 1998) and decreased BRS (Farrell et al. 1991; La Rovere et al. 1988; La Rovere et
al. 1998). Low HRV, altered heart rate behavior and low BRS have predicted total
mortality (Bigger et al. 1992; La Rovere et al. 1988; La Rovere et al. 1998), arrhythmic
death (Bigger et al. 1992; Farrell et al. 1991; Hartikainen et al. 1996) and nonarrhythmic
death (Hartikainen et al. 1996).
There is evidence to support the potential usefulness of HRV and BRS in predicting
mortality in various other patient groups as well. Decreased HRV predicts mortality in
subjects with angina pectoris without acute myocardial infarction (Rich et al. 1988), and
reduced HRV and altered heart rate behavior predict mortality in elderly populations
(Huikuri et al. 1998; Tsuji et al. 1994; Tsuji et al. 1996) and in heart failure patients
(Brouwer et al. 1996; Ho et al. 1997).
3. Purpose of the present study
The aims of the present study were to examine autonomic cardiovascular regulation
assessed by HRV and BRS in relation to age, sex and cardiovascular risk factors in
subjects without heart disease. The specific goals were:
1. to evaluate the interindividual variation of HRV and its association with various
cardiovascular risk factors in healthy middle-aged subjects (I,V);
2. to study the possible gender-related differences in cardiac autonomic regulation in
healthy middle-aged subjects (II,V);
3. to test whether there are differences in cardiac autonomic regulation in hypertensive
subjects with and without metabolic features of IRS and normotensive controls (III);
4. to study the effects of aging from early childhood to advanced age on heart rate
dynamics in healthy subjects using both conventional measures of HRV and newly
derived measures based on complexity and fractal scaling (chaos theory) (IV), and
5. to study the intersubject variation and determinants of short-term correlation properties
and complexity of HRV in middle-aged subjects (V).
4. Subjects and methods
4.1. Study populations
The study cohort used in the Oulu Project Elucidating Risk of Atherosclerosis (OPERA)
served as the random population of this study (I-III,V). OPERA was a population-based
epidemiological study addressing the risk factors and disease end-points of
atherosclerotic diseases. The hypertensive cohort comprised 600 subjects (300 males,
300 females) living in the Oulu district. The subjects were randomly selected from the
register of the Social Insurance Institution from those receiving reimbursement for
antihypertensive medication. The subjects were 40 to 59 years old at the baseline of the
study (1990). The randomization was age-stratified, i.e. for each year of birth, 15 males
and 15 females were selected. Age- and sex-matched controls (300 males and 300
females) were randomly selected for the hypertensive subjects from the general
population of Oulu, excluding those entitled to a refund for antihypertensive medication.
The whole study protocol was completed by 471 (79%) hypertensive subjects and 488
(81%) controls. Informed consent was obtained from the subjects, and the protocol was
approved by the Ethics Committee of the University of Oulu.
The subjects with symptoms of angina pectoris, cardiac medication, or
electrocardiographic evidence of CAD based on the Minnesota codes (Prineas 1982) (35
healthy, 60 hypertensive subjects) were excluded from further HRV and BRS analyses.
Known diabetes or fasting blood glucose >6.7 mmol/l (I-III,V) or 2-hour blood glucose
>10 mmol/l (III)(16 healthy, 41 hypertensive subjects) also led to exclusion. The subjects
in the hypertensive cohort or with recently started antihypertensive or antiarrhythmic
medication (37) were excluded from the studies of healthy subjects (I-II,V). Technical
artifacts and rhythm abnormalities (atrial fibrillation or frequent ectopic beats) during the
ECG-recording invalidated the analyses of HRV in some subjects. Due to an inadequate
blood pressure rise after the Valsalva strain release, the BRS analyses of some subjects
not otherwise excluded were not possible. The demographic, lifestyle, laboratory and
echocardiographic data of the study populations are presented in Table 1 (III,V).
Interindividual variation and relation to cardiovascular risk factors of HRV using ECG
recordings with supine, sitting and walking periods were analyzed in 172 healthy middleaged males (I). Since most of the technical artifacts were seen during the walking period
33
or were related to changes in body posture, more subjects were included when HRV was
analyzed only during the lying and sitting periods (II,V). The time and frequency domain
measures of HRV and BRS were studied in 188 healthy men and 186 women (HRV) and
151 men and 152 women (BRS) (II), respectively. The correlation properties and the
complexity of R-R intervals were studied in 192 men and 202 women (V). Women had
lower blood pressure (p<0.001 for both systolic and diastolic blood pressure), higher
Framingham psychosocial scores (p<0.001) and included fewer drinkers (p<0.001).
Women had lower fasting blood glucose and insulin values (p<0.001 for both), serum
lipid values (p<0.001) and left ventricular mass (p<0.001) compared to men. No other
significant differences in the baseline variables were found between men and women
(Table 1).
Sixty-nine hypertensive subjects (50 males, 19 females) with IRS were found, when
classified according to the following criteria: (1) fasting serum insulin > 12 mU/l, (2)
serum triglycerides > 2.0 mmol/l and (3) either established long-standing hypertension
with antihypertensive medication or systolic blood pressure >160 mmHg or diastolic
blood pressure >90 mmHg (criteria modified from references (WHO 1959, Consensus
conference 1984, Rönnemaa 1991)) (III). Age-, sex- and medication-matched
hypertensive controls without IRS were selected from the original study subjects, who
met the hypertension criteria, but not the two metabolic criteria and age- and sex-matched
normotensive controls were selected from the subjects who met none of the three IRS
criteria. Blood pressure was significantly higher in both hypertensive groups compared to
normotensives (p<0.0001 for both systolic and diastolic blood pressure), while BMI,
waist/hip ratio, fasting and 2-hour blood glucose and insulin and the lipid values (total
cholesterol, low-density and very-low-density lipoprotein cholesterol and serum
triglycerides) were significantly higher and high-density lipoprotein cholesterol lower in
the hypertensive subjects with IRS than in the other two groups (p<0.0001 for all, <0.05
for LDL cholesterol) (Table 1). The antihypertensive medication used in the hypertensive
groups is listed in Table 2.
34
Table 1. Clinical, lifestyle and laboratory characteristics of the study populations.
Healthy
middle-aged
males
(n=192)
(V)
50 ± 6
26.4 ± 3.5
0.91 ± 0.06
147 ± 19
Age (years)
BMI (kg/m²)
Waist/hip ratio
Systolic blood pressure
(mmHg)
Diastolic blood pressure
89 ± 11
(mmHg)
Current smoking habits
No smoking
123 (64%)
Moderate smoking (<20/d) 57 (30%)
Heavy smoking (>20/day)
12 (6%)
Duration of smoking
16 ± 14
(years when smoked
Median 15
20 cigarettes/day)
(0-46 years)
Alcohol consumption
24 (13%)
No drinking
1-100g/week
96 (50%)
>100g/week
72 (37%)
Leisure time physical activity
6 (3%)
No activity
57 (30%)
Mild activity
68 (35%)
Moderate activity
61 (32%)
Heavy activity
Personality type
Framingham
26 ± 5
Bortener
22 ± 3
Hostility
7±3
Fasting blood glucose
4.4 ± 0.5
(mmol/l)
2-hour blood glucose (mmol/l) 5.1 ± 1.5
Fasting serum insulin (mU/l)
13 ± 9
2-hour serum insulin (mU/l)
55 ± 53
5.78 ± 1.11
Total serum cholesterol
(mmol/l)
HDL-cholesterol (mmol/l)
1.24 ± 0.30
LDL-cholesterol (mmol/l)
3.75 ± 0.95
VLDL-cholesterol (mmol/l)
0.43 ± 0.29
Serum triglycerides (mmol/l)
1.51 ± 0.76
Left ventricular mass #
221 ± 53
Left ventricular mass index #
112 ± 26
Fractional shortening (%) #
34 ± 6
Healthy
Normotensive Hypertensive
middle-aged subjects (n=69)
subjects
females
(III)
without IRS
(n=202)
(n=69)
(V)
(III)
49 ± 5
51 ± 6
51 ± 6
24.5 ± 2.8
25.8 ± 3.1
25.6 ± 4.0
0.84 ± 0.08
0.87 ± 0.07
0.78 ± 0.05
130 ± 12
161 ± 14
138 ± 20
Hypertensive
subjects with
IRS (n=69)
(III)
49 ± 6
30.4 ± 4.1
0.94 ± 0.07
161 ± 18
82 ± 12
78 ± 7
97 ± 8
99 ± 6
150 (74%)
49 (24%)
3 (2%)
5 ± 10
Median 0
(0-57 years)
49 (71%)
17 (25%)
3 (4%)
49 (71%)
16 (23%)
4 (6%)
39 (57%)
25 (36%)
5 (7%)
37 (18%)
162 (80%)
3 (2%)
10 (15%)
50 (72%)
9 (13%)
12 (17%)
36 (52%)
21 (30%)
8 (12%)
38 (55%)
23 (33%)
7 (3%)
44 (22%)
69 (34%)
82 (41%)
15 (22%)
27 (39%)
22 (32%)
5 (7%)
4 (6%)
14 (20%)
23 (33%)
28 (41%)
2 (3%)
35 (51%)
21 (30%)
11 (16%)
28 ± 5
23 ± 3
7±3
4.3 ± 0.4
4.1 ± 0.4
4.3 ± 0.5
4.8 ± 0.6
5.1 ± 1.3
9±6
54 ± 45
5.49 ± 1.00
4.6 ± 0.9
6.8 ± 2.3
31.2 ± 15.4
5.4 ± 1.1
5.0 ± 1.1
7.4 ± 2.1
41.3 ± 22.5
5.5 ± 0.9
6.1 ± 1.6
22.7 ± 13.9
116.9 ± 97.9
6.2 ± 1.1
1.57 ± 0.38
3.30 ± 0.90
0.28 ± 0.20
1.13 ± 0.60
154 ± 35
91 ± 19
35 ± 5
1.45 ± 0.44
3.38 ± 1.00
0.26 ± 0.15
1.01 ± 0.33
1.44 ± 0.35
3.50 ± 0.89
0.27 ± 0.13
1.07 ± 1.31
1.05 ± 0.22
3.84 ± 1.01
0.85 ± 0.32
2.73 ± 0.84
The values are means ± SD or the numbers of subjects in each group when appropriate.
# n=193 women, n=166 men. BMI = body mass index, HDL = high-density lipoprotein, IRS =
insulin resistance syndrome, LDL = low-density lipoprotein, VLDL = very-low-density
lipoprotein.
35
Table 2. Antihypertensive medication of hypertensive subjects with and without IRS (III).
Medication
Hypertensive
subjects
without IRS
(n=69)
23
Hypertensive
subjects with
IRS
(n=69)
23
No medication
Monotherapy
10
9
β blocker
Diuretic
1
1
Calcium antagonist
3
3
ACE inhibitor
4
4
Combination therapy
2
2
β blocker + calcium antagonist
4
4
β blocker + ACE inhibitor
7
8
β blocker + diuretic
3
4
β blocker + ACE inhibitor + diuretic
Calcium antagonist + ACE inhibitor
1
1
Calcium antagonist + diuretic
2
2
ACE inhibitor + diuretic
3
2
ACE inhibitor + calcium antagonist + diuretic
3
2
1
2
β blocker + calcium antagonist + ACE inhibitor
1
1
β blocker + calcium antagonist +
ACE inhibitor + diuretic
Other antihypertensive medication
1
1
Values are numbers of subjects in each group. ACE = angiotensin converting enzyme.
For other abbreviations, see Table 1.
One-hundred and fourteen healthy subjects ranging in age from 1 to 82 years were
included in the study of 24-hour heart rate dynamics (IV). The subjects were divided into
four groups: (1) children, age under 15 years (mean 8 ± 5 years), (2) young adults, age
15-40 years (mean 28 ± 6 years), (3) middle-aged, age 40-60 years (mean 50 ± 6 years)
and (4) elderly, age over 60 years (mean 71 ± 5 years). There were 15 males and 12
females in the group of children and 17 males and 12 females in all the other groups. The
children and young adults were apparently healthy with no history or symptoms of heart
disease, hypertension or diabetes, and with normal findings in the clinical examination.
The middle-aged subjects were selected from the OPERA population. The elderly
subjects were selected from a sample of 480 elderly people living in Turku, who had
been randomly selected from the register of the Social Insurance Institution (Räiha et al.
1994). All middle-aged and elderly subjects underwent a physical examination with
blood pressure measurements, a standard 12-lead ECG, a chest x-ray and laboratory tests,
including blood glucose measurements. No subjects with a previous history, symptoms or
clinical evidence (including the analysis of a 12-lead ECG with the Minnesota code) of
ischemic heart disease, diabetes, hypertension or any other medical disorder were
36
included. Fasting blood glucose was < 6.7 mmol/l and blood pressure <160/90 mmHg in
all subjects. None of the subjects included in this study were taking any medication.
4.2. Collection of clinical data
The study subjects (I-III,V) were interviewed by a physician, and a standardized health
questionnaire of past medical history, medication used, cardiac symptoms (angina
pectoris, dyspnea), smoking habits, alcohol consumption, physical activity and
personality type was filled in. Alcohol intake was converted into grams of absolute
ethanol per week. On the basis of leisure-time physical activity, four groups were formed,
using a modification of the method described by Grimby (Grimby 1986). The personality
type was scaled by three different methods: (1) Framingham type A behavior pattern
scale (Haynes et al. 1978), (2) Bortner´s short rating scale of behavior pattern (Bortner
1969), and (3) hostility scaling (Koskenvuo et al. 1988). Women were classified as
postmenopausal if at least six months had passed since their latest menstruation.
All the subjects went through a clinical examination, including height, weight and
waist and hip circumference measurements. Blood pressure was measured and a standard
12-lead ECG and chest x-ray were taken. ECGs were evaluated using the Minnesota
Code (Prineas 1982). A wide range of laboratory tests, echocardiographic measurements,
a 45-minute or a 24-hour ECG-recording and tests to evaluate BRS were performed.
4.3. Blood pressure measurements
Blood pressure was measured with an automatic oscillometric blood pressure recorder
(DinamapR model 18465X, Criticon Ltd., Inc., Tampa, FL, USA). The subject was seated
for at least 5 minutes, after which blood pressure was measured three times at one-minute
intervals from the right arm. The mean of the three measurements was used in the
analysis.
4.4. Laboratory methods
All the laboratory test samples were drawn after a 12-hour requested fast. A two-hour
oral glucose tolerance test (75 g) was performed by determining the 0- and 2-hour
glucose (a glucose-oxidase method, Diagnostica, Merck, Darmstadt, Germany) and
insulin (a two-site immunoenzymometric assay, Tosoh Corp., Tokyo, Japan)
concentrations.
To determine the lipoprotein fractions, the VLDL fraction was first separated from
plasma by ultracentrifugation. Then the HDL fraction was determined from the VLDLfree fraction and the plasma LDL cholesterol concentration was calculated by subtracting
the cholesterol concentration in HDL from that in the VLDL-free fraction (Kervinen et
37
al. 1994). The concentrations of total cholesterol and triglycerides were determined in
the plasma and lipoprotein fractions by enzymatic colorimetric methods (kits of
Boehringer Diagnostica, Mannheim GmbH, Germany) using a Kone Specific analyser
(Kone Spesific, Selective Chemistry Analyser, Kone Instruments, Espoo, Finland).
4.5. Echocardiographic measurements
A Hewlett-Packard 77020A ultrasound color system (Hewlett-Packard, Andover, MA,
USA) was used for M-mode, two-dimensional and Doppler-echocardiographic studies of
each subject (I-III,V). All studies were performed by the same experienced cardiologist
blind to the clinical data of the subjects. Standard techniques were used. The M-mode
measurements were made according to the American Society of Echocardiography
guidelines (Sahn et al. 1978), and left ventricular mass was calculated with the formula
of Troy (Troy et al. 1972). Fractional shortening was calculated by dividing the
difference between the left ventricular internal dimensions in diastole and systole by the
diastolic dimension and then multiplying it by 100. Measurements of the left ventricle
were technically inadequate in some subjects, mostly due to obesity. Thus, the
echocardiographic data of 166 middle-aged males and 193 females (V) were used.
4.6. ECG recordings for HRV analysis
Both short-term ECG recordings in controlled conditions and ambulatory 24-hour ECG
recordings were used in this study. The short-term ECG recordings were obtained for 45
minutes with an ambulatory ECG recorder (Dynacord Holter Recorder, Model 420, DM
Scientific, Irvine, CA, USA) with a sample frequency of 256 Hz, and each subject was
monitored for 15 minutes while quietly lying down and breathing normally, for 15
minutes in the sitting position and for 15 minutes while walking (I-III,V). The length of
the short-term ECG recordings was chosen after a pilot study on 37 normotensive and 40
hypertensive subjects to calculate the shortest period (including lying, sitting and
walking) that would give a reasonable correlation (r>0.6) with the 24-hour measures of
HRV (Huikuri et al. 1996). These short-term ECG recordings were performed between 7
a.m. and 3 p.m. The mean values of the three periods (supine, sitting, walking) were used
in the studies I and III. Most of the technical artifacts were found during the walking
period. Thus, the decision to use the ECG data of only the supine and sitting periods
made it possible to include more subjects (II,V). In this summary, the interindividual
variations and the relations of measures of HRV to sex and risk factors in healthy
subjects are presented based on the results of these studies with controlled supine and
sitting periods (II,V). Any subject with the ECG data of the analyzed period including
less than 85% sinus beats was excluded from further analysis.
The 24-hour ambulatory ECG recordings were made while the subjects were
performing usual everyday activities out of hospital (IV). All the subjects had at least 18
hours (mean 23 ± 1 hours) of ECG data, including at least 85% normal sinus beats.
38
4.7. Measurement of HRV
The digitized ECG data were transferred from the Del Mar Avionics ECG scanner
(Model 500, Del Mar Avionics, Irvine, CA, USA) to a microcomputer for the analysis of
HRV with the use of a custom-made program (Hearts, Heart Signal co, Kempele,
Finland). The program automatically detects and labels each QRS complex. Premature
beats and noise were excluded automatically and manually from the computer-generated
tachograms. The gaps were either refilled with an average value computed in the local
neighborhood (I-III) or deleted (IV,V). Segments with >85% qualified beats were
included in the final analysis.
4.7.1. Time and frequency domain measures
A linear detrend was used for segments of 512 consecutive beats to make the data more
stationary. A straight line was first fitted to a segment by the standard least squares
method and then subtracted from the sample value. An autoregressive model was used to
estimate the power spectral densities from short-term ECG recordings (I-III). The value
20 was used as the order of the model. Four frequency bands (total power <0.4 Hz, HF
power 0.15-0.4 Hz, LF power 0.04-0.15 Hz and VLF power 0.005-0.04 Hz) were
calculated from successive segments of 512 beats. Average values of these segments
giving the spectral power in absolute units (ms2) during the period analyzed were used.
The mean R-R interval and SDANN (I-III) were used as time domain measures.
Furthermore, the ratio between the LF and HF components (LFHF ratio) and the
normalized units of the LF and HF components (e.g. LF or HF spectral power in absolute
units divided by total power from which the VLF power had been subtracted) were
calculated (Huikuri et al. 1994). Time and frequency domain measures of HRV were
calculated with linear detrending from a segment of the middle 13 minutes of the 15minute supine and sitting periods, respectively (V), to avoid the possible confounding
effects of the nonstationarities and artifacts in the ECG data resulting from changes in
body posture, but by otherwise using the above mentioned methods.
The power spectrum densities were estimated using the fast Fourier method from the
24-hour recordings (IV). ULF (<0.0033 Hz) and VLF power (0.0033-0.04 Hz) were
calculated from the entire 24-hour segment. LF power (0.04-0.15 Hz), HF power (0.150.4 Hz) and nighttime and daytime VLF power were calculated from 1-hour segments of
the 24-hour recording. The average value of these segments was used. The mean R-R
interval and SDNN calculated from the entire 24-hour period were used as time domain
measures.
39
4.7.3. Fractal scaling and complexity measures
The measures of R-R interval dynamics were calculated from the entire 24-hour
recording and also separately from the hours representing the nighttime (0 - 6 am) and
the daytime (9 am - 6 pm), to assess the possible differences between day and night.
Power-law relationship analysis. The slope (β) of the linear inverse power-law
relationship of (log) power to (log) frequency was calculated from the frequency range of
10-4 to 10-2 Hz (period approximately 3 hours (10 000 seconds) to 1.5 minutes (100
seconds)) from the 24-hour ECG recordings (IV). The point power spectrum was
logarithmically smoothed in the frequency domain, and the power was integrated into
bins spaced 0.0167 log(Hz) apart. A robust line-fitting algorithm of log(power) on
log(frequency) was then applied to the power spectrum between 10-4 and 10-2 Hz, and the
slope of this line was calculated. This range was chosen because of the typically linear
relationship between log (power) and log (frequency) in this frequency band (Bigger et
al. 1996).
Detrended fluctuation analysis The DFA technique was used to quantify the fractallike scaling properties of R-R interval data. The root-mean-square fluctuation of the
integrated and detrended data was measured in observation windows of varying sizes and
then plotted against the size of the window on a log-log scale. The scaling exponent α
represents the slope of the line which relates (log) fluctuation to (log) window size. In
this study (IV), both α1, the short-term (4 to 11 beats) and α2, the intermediate-term
(>11 beats to minutes) scaling exponents were calculated. The scaling exponents were
calculated from segments of 8000 beats of the 24-hour ECG recording with the method
earlier described (Iyengar et al. 1996; Peng et al. 1995), and average values of these
segments were used. α1 and α2 were also calculated from segments of 4000 beats in 3hour periods (0-3 am, 3-6 am, 9-12 am, 12 am -3 pm, 3-6 pm), to obtain the nighttime (06 am) and daytime (9 am-6 pm) average values. The short-term scaling exponent α1 was
also calculated from a block of the middle 13 minutes of the 15-minute supine and sitting
periods, respectively, in the short-term ECG recordings (V), using otherwise the
abovementioned methods.
Approximate entropy was used to measure the complexity of the time series data.
ApEn was calculated from segments of 8000 beats of the 24-hour ECG recording (IV)
with fixed input variables m=2 and r=20% with a method described earlier (Pincus et al.
1992; Pincus et al. 1994) and using the average values of these segments. ApEn was also
calculated from segments of 4000 beats in 3 hour periods (0-3 am, 3-6 am, 9-12 am, 12
am -3 pm, 3-6 pm), to obtain the nighttime (0-6 am) and daytime (9 am-6 pm) average
values. ApEn was also calculated from a block of the middle 13 minutes of the 15-minute
supine and sitting periods, respectively, in short-term ECG recordings (V), using
otherwise the abovementioned methods.
4.8. Measurement of BRS
BRS was measured using the Valsalva method (Airaksinen et al. 1993). The study
subjects performed the Valsalva maneuver in the sitting position by blowing into a rubber
40
tube connected to an aneroid manometer and maintaining a pressure of 40 mmHg for 15
seconds. The R-R intervals were obtained from the surface ECG and fed into the analogto-digital converter (DAP 1200, Microstar Laboratories, Inc., USA) of a microcomputer.
Non-invasive arterial pressure was measured on a beat-to-beat basis using the Finapres
finger-cuff method. The analysis of BRS was accomplished with a menu-driven software
package (CAFTS, Medikro Oy, Kuopio, Finland). The tachograms of beat-to-beat R-R
intervals and blood pressure values were viewed and optionally edited on the screen. The
time window of the rapid blood pressure rise after the strain release was defined for the
calculation of the BRS. This time window ranged from the beat on which the systolic
blood pressure exceeded the systolic blood pressure level at the end of the Valsalva strain
to the beat following the maximum systolic blood pressure overshoot. BRS was
calculated as the slope of the linear relationship between the length of the R-R interval
(in ms) and the preceding systolic blood pressure value (in mmHg) within the defined
time window using linear least-mean-squares fitting.
Only regression lines with a correlation coefficient greater than 0.8 or with a
significant p-value (<0.05) and with a blood pressure change = 15 mmHg were accepted
for analysis. The test was repeated three times. The BRS value used in this study was
calculated as the mean value of two accepted tests with the best correlation coefficients
of the regression lines.
BRS analyses could not be performed if the blood pressure rise after strain release was
inadequate or if frequent ectopies or artifacts were present during the ECG recording.
4.9. Statistics
Continuous variables are expressed as mean ± standard deviation, unless specified
otherwise. p-values <0.05 (I-IV) and <0.01 (V) were considered statistically significant.
A logarithmic transformation to the natural base was made on the variables with highly
skewed distributions. Parametric statistical methods were chosen based on the results of
the Kolmogorov-Smirnov test (Z < 1.0) of normal distribution (IV).
The associations between different variables were analyzed with univariate techniques
(Pearson’s bivariate correlation, one-way ANOVA). A stepwise linear multiple
regression analysis was performed by including all the variables which had a significant
univariate relation to the dependent variable in question (I,III). The results of the
stepwise linear multiple regression analysis are given as standardized multiple regression
correlates (β) coefficients and multiple R and R2 of the model.
The comparisons between two samples or groups were performed using the MannWhitney two-sample test (II), the χ2 test (II) and the independent samples t-test (III,
IV,V) and those between more than two groups using the one-way ANOVA followed by
the Bonferroni correction for multiple comparisons (III,IV). The differences between
supine vs. upright positions were tested with paired samples t-test (V). The significant
differences in the baseline variables between the groups were taken into account by
means of analysis of covariance (ANCOVA ) (II, IV,V).
5. Results
5.1. Time and frequency domain measures of HRV in healthy middleaged subjects
5.1.1. Interindividual variation in measures of HRV
Wide interindividual variation compared to the variation in the mean R-R interval was
observed in SDANN and the spectral measures of HRV as analyzed in both absolute and
normalized units (n=394, Table 3) (V). The interindividual variations were larger in the
controlled supine compared to the upright (sitting) position.
5.1.2. Sex-related differences in measures of HRV
The measures of R-R interval dynamics in men (n=192) and women (n=202) are shown
in Table 4. (V). Women had lower SDANN, higher ApEn and lower α1 compared to
men (p<0.001 for all), while the mean R-R interval, LF or HF power in absolute units did
not differ. The LFHF ratio (p<0.001) and normalized LF power (p<0.01) were lower in
women, while the normalized HF power was higher (p<0.001). When the observed
differences in the baseline variables (blood pressure, serum lipid values, fasting blood
glucose and insulin values, left ventricular mass, alcohol consumption and the
Framingham psychosocial score, Table 1.) between men and women were taken into
account in ANCOVA by using clinical and laboratory variables as covariates, α1 was
still lower (F=7.9, p<0.01), ApEn higher (F=8.8, p<0.01) and SDANN lower (F=28.7,
p<0.001) in women compared to men. The significant sex-related differences in the
LFHF ratio (p<0.01) and normalized HF power (p<0.001) also remained.
SDANN (62 ± 24 vs. 48 ± 18 ms, p<0.01), HF power (350 ± 290 vs. 260 ± 310 ms2,
p<0.05) and LF power (710 ± 110 vs. 410 ± 340 ms2, p<0.01) in absolute units were
significantly higher in postmenopausal women on estrogen replacement therapy (n=46)
than in age-matched postmenopausal women without hormone therapy (II).
43
Table 3. Interindividual variation in the measures of R-R interval dynamics in healthy
middle-aged subjects (n=394).
Mean R-R interval (ms)
supine+sitting
supine
sitting
SDANN (ms)
supine+sitting
supine
sitting
LF power (0.04-0.15 Hz)
(ms2) supine+sitting
supine
sitting
(nu) supine+sitting
supine
sitting
HF power (0.15-0.4 Hz)
(ms2) supine+sitting
supine
sitting
(nu) supine+sitting
supine
sitting
LFHF-ratio
supine+sitting
supine
sitting
Mean ± SD
Range
Coefficient
of variation
865 ± 122
891 ± 131
839 ± 121
535 - 1324
573 - 1402
428 - 1260
14 %
15 %
14 %
54 ± 21
49 ± 21
59 ± 24
18 - 174
13 - 168
16 - 180
39 %
43 %
41 %
736 ± 630
655 ± 655
817 ± 729
66 ± 15
63 ± 18
69 ± 17
36 - 4331
35 - 5941
26 - 5961
14 - 98
11 - 98
8 - 98
86 %
100 %
89 %
23%
28%
25%
354 ± 450
381 ± 585
326 ± 385
28 ± 11
31 ± 14
25 ± 11
20 - 5074
10 - 7231
10 - 3870
8 - 63
3 - 72
2 - 59
127 %
154 %
118 %
38 %
44 %
43 %
2.93 ± 1.78
2.64 ± 2.04
3.62 ± 3.31
0.50 -11.9
0.24 - 17.10
0.45 - 13.24
61%
77%
91%
1.21 ± 0.17
1.15 ± 0.20
1.26 ± 0.18
0.71 - 1.61
0.50 - 1.68
0.61 - 1.67
14 %
17 %
14 %
1.13 ± 0.14
1.14 ± 0.18
1.11 ± 0.19
0.56 - 1.45
0.31 - 1.49
0.48 - 1.48
12 %
16 %
17 %
α1
supine+sitting
supine
sitting
ApEn
supine+sitting
supine
sitting
α1 = short-term scaling exponent, ApEn = approximate entropy, HF = high frequency,
LF = low frequency, nu = normalized units, SDANN = standard deviation of R-R
intervals
44
Table 4. Measures of R-R interval dynamics in healthy middle-aged men and women.
Men (n=192)
Mean R-R Interval
864 ± 136
ApEn
1.10 ± 0.14
α1
1.25 ± 0.17
SDANN (ms)
58 ± 24
LF
(ms2)
805 ± 714
(nu)
68 ± 17
HF
(ms2)
357 ± 541
(nu)
25 ± 11
LFHF-ratio
3.4 ± 2.0
For abbreviations, see Table 3.
Women (n=202)
866 ± 107
1.15 ± 0.14
1.17 ± 0.16
50 ± 16
p-value
ns
<0.001
<0.001
<0.001
671 ± 532
64 ± 14
ns
<0.01
351 ± 344
31 ± 10
2.5 ± 1.5
ns
<0.001
<0.001
5.1.3. Relation of HRV to demographic, lifestyle and cardiovascular risk
factors
Tables 5 and 6 list the significant bivariate correlations between the time and frequency
domain measures of HRV and the demographic and laboratory variables in men (n=192)
and women (n=202) (V). The time and frequency domain measures had significant
inverse correlations with age and serum insulin levels in both men and women and
additionally with systolic blood pressure and serum triglyceride levels in men. They also
correlated directly with HDL cholesterol in men. The LFHF ratio and normalized
spectral HF power and LF power were not related to any measure in either men or
women. No significant relations were found between any measure of HRV and current
smoking habits, alcohol consumption, leisuretime physical activity, personality type or
echocardiographic parameters in either men or women. When the measures of HRV were
analyzed in supine or upright positions separately, the relations to the abovementioned
factors were similar.
Table 5. Significant relationships of measures of R-R interval dynamics to demographic
and laboratory parameters in healthy middle-aged men (n=192).
ApEn
SDANN
LF
HF
α1
Age
ns
ns
-0.19 *
-0.29 **
-0.19 *
Systolic blood pressure
ns
ns
ns
ns
-0.21 *
2-hour serum insulin
ns
ns
-0.23 **
-0.23 *
-0.29 **
HDL cholesterol
ns
ns
0.19 *
0.20 *
ns
Serum triglycerides
ns
ns
-0.25 **
-0.21 *
-0.23 **
* p<0.01,** p<0.001, ns = non-significant. For other abbreviations, see Tables 1 and 3.
45
Table 6. Significant relationships of measures of R-R interval dynamics to demographic
and laboratory parameters in healthy middle-aged women (n=202).
ApEn
α1
Age
ns
ns
Blood glucose
ns
ns
Fasting serum insulin
ns
ns
2-hour serum insulin
ns
ns
For abbreviations, see Tables 1,3 and 4.
SDANN
ns
ns
-0.23 **
ns
LF
-0.24 **
ns
-0.18 *
ns
HF
-0.21 *
-0.18 *
-0.19 *
-0.18 *
5.2. BRS in healthy middle-aged subjects
5.2.1. Interindividual variation and relations to demographic, lifestyle
and cardiovascular risk factors of BRS
Large interindividual variation in BRS was observed in both men (n=151) (10.5 ± 4.6
ms/mmHg, coefficient of variation 58%) and women (n=152) (8.0 ± 4.6 ms/mmHg,
coefficient of variation 44%) (II). BRS was significantly related to systolic blood
pressure (r= -0.31 and r= -0.30, in men and women, respectively, p<0.001 for both)
(unpublished). In addition BRS was related to blood glucose (r=-0.25, p<0.01) and 2hour serum insulin (r= -0.34,p<0.001) in women (unpublished). No other significant
relations to demographic, laboratory, lifestyle or echocardiographic factors were found.
5.2.2. Sex-related differences in BRS
BRS was significantly lower in women (n=152) than in men (n=151) (8.0 ± 4.6
ms/mmHg vs. 10.5 ± 4.6 ms/mmHg, p<0.001) and the difference remained significant
after adjustment for the differences in baseline variables (Table 1) (F=22.8, p<0.001)
(II). BRS (10.4 ± 5.0 vs. 6.8 ± 3.2 ms/mmHg, p<0.01) was significantly higher in
postmenopausal women on estrogen replacement therapy (n=46) than in age-matched
postmenopausal women without hormone therapy (II).
46
5.3. HRV and BRS in hypertensive subjects with and without insulin
resistance syndrome
5.3.1. Overall HRV and BRS in hypertensive subjects with and without
metabolic features of insulin resistance syndrome and normotensive
controls
All the time and frequency domain measures of HRV analyzed in absolute units, with the
exception of the HF component, were significantly lower in hypertensive subjects with
IRS (n=69) compared to hypertensive subjects without IRS and normotensive subjects
(Table 7) (III). On the other hand, hypertensive subjects without IRS (n=69) did not
differ significantly from normotensive subjects with respect to total or LF power or
SDANN. The HF power of HRV and BRS were significantly higher in normotensive
subjects than hypertensive subjects, but did not differ between the two hypertensive
groups. The average heart rate, LFHF ratio and normalized LF and HF power were not
significantly different between any of the groups.
Hypertensive subjects without IRS and antihypertensive medication (n=23) did not
differ significantly from the normotensive group, but the LF component (p<0.05), HF,
total power and SDANN (p<0.001 for all) were significantly lower in the hypertensive
subjects with IRS and without medication (n=23) than in the normotensive controls. The
measures of HRV or BRS did not differ when the subjects with and without β-blockers,
ACE inhibitors, calcium antagonists or diuretics were compared in each group.
5.3.2. Relationships of HRV and BRS to features of insulin resistance
syndrome
In the group of hypertensive subjects with and without IRS (n=138), all measures of
HRV had significant negative correlations with systolic blood pressure (r between -0.20
and -0.28, p<0.05), fasting blood glucose (r between -0.17 and -0.31, p<0.05) and 2-hour
glucose levels (r between -0.20 and -0.31, p<0.05). All measures except HF power had
significant correlations with the fasting serum insulin (-0.23 and -0.24, p<0.01) and 2hour insulin levels (r between -0.18 and -0.23, p<0.05), serum triglyceride levels (-0.27
and -0.31, p<0.01) and serum VLDL cholesterol levels (-0.21 and -0.28, p<0.05). Total
power and SDANN also correlated with body mass index (r between -0.17 and -0.18,
p<0.05). No significant correlations emerged for total, low-density or high-density
lipoprotein cholesterol. BRS was only related to systolic blood pressure (r=-0.32,
p<0.01). In the multiple regression analysis, all measures of HRV and BRS were
significantly predicted by systolic blood pressure (Table 8). Total and LF power and
SDANN were also independently predicted by the serum triglyceride level, which was an
even more powerful predictor of total and LF power and SDANN than systolic blood
pressure.
47
Table 7. Mean R-R interval, HRV and BRS of hypertensive subjects with and without IRS
and normotensive controls.
Normotensive
subjects
(n=69)
848 ± 101
Hypertensive
subjects without
IRS (n=69)
836 ± 125
Hypertensive
subjects with
IRS (n=69)
815 ± 139
p-value
Mean R-R interval
ns
(ms)
SDANN (ms)
65 ± 18
59 ± 20
50 ± 19§
<0.0001
HF power
ms²
231 ± 202
208 ± 221
322 ± 320
ln
5.46 ± 0.76§
5.13 ± 0.80
5.01 ± 0.76
<0.01
nu
32 ± 11
29 ± 10
34 ± 11
ns
LF power
ms²
659 ± 499
573 ± 423
414 ± 354
ln
6.25 ± 0.69
6.05 ± 0.86
5.72 ± 0.81§
<0.001
nu
68 ± 11
71 ± 10
66 ± 11
ns
Total power
ms²
2885 ± 1943
2386 ± 1628
1834 ± 1613
ln
7.78 ± 0.61
7.53 ± 0.77
7.23 ± 0.74§
<0.0001
LFHF ratio
2.57 ± 1.53
2.82 ± 1.38
2.33 ± 1.25
ns
BRS (ms/mmHg) #
11.1 ± 4.8¤
9.2 ± 6.0
7.7 ± 4.0
<0.05
Values are mean ± SD. # n=46, 40 and 30 subjects in the groups. BRS = baroreflex
sensitivity, HRV = heart rate variability, ln = natural logarithm. p-value from one-way
ANOVA, § indicates that the group differed in the Bonferroni post hoc analysis from the
other two groups, ¤ indicates difference between normotensive subjects and hypertensive
subjects with IRS. For other abbreviations, see Tables 1, 3 and 4.
Table 8. Results of the multiple regression analysis of the different measures of HRV and
BRS and the features of IRS in hypertensive subjects with and without IRS (n=138).
Systolic blood pressure
Serum triglycerides
Parameter
SDANN
β
p-value
β
p-value
Multiple R
R2
-0.22
<0.01
-0.28
<0.001
0.37
0.13
Total
power
-0.26
<0.01
-0.27
<0.001
0.39
0.15
LF
power
-0.27
<0.001
-0.30
<0.001
0.42
0.17
HF
power
-0.20
<0.05
ns
BRS #
0.20
0.04
0.32
0.10
-0.32
<0.01
ns
ns = factor did not enter the model, # n=70. For other abbreviations, see Tables 3 and 7.
48
5.4. Fractal correlation properties and complexity of HRV
5.4.1. Interindividual variation and relations to demographic, lifestyle
and cardiovascular risk factors in short-term scaling exponent and
approximate entropy
Both the short-term scaling exponent α1 and ApEn showed only slight interindividual
variation (coefficients of variations 14% and 12%, respectively) in controlled conditions
(n=394, Table 3)(V). Women (n=202) had higher ApEn (1.15 ± 0.14 vs. 1.10 ± 9.14,
p<0.001) and lower α1 (1.17 ± 0.16 vs. 1.25 ± 0.17, p<0.001) compared to men (n=192)
(Table 4)(V). α1 and ApEn were not significantly related to any other demographic,
lifestyle, laboratory or echocardiographic factors in middle-aged subjects (V).
5.4.2. Effects of age on 24-hour fractal correlation properties and
complexity of HRV
Representative examples of R-R interval time series, 24-hour power spectra, power-law
relationship analyses and detrended fluctuation analyses (DFA) of 7-year, 29-year and
76-year old healthy males are shown in Figures 1 and 2. Different measures of HR
dynamics as a function of age are plotted in the Figures 3-5 (IV).
Figure 1. Representative examples of R-R interval time series and 24-hour power spectra of healthy males aged 7 years (left), 29 years (middle)
and 76 years (right).
48
50
Figure 3. Time domain measures of heart rate variability in relation to age in 114 healthy subjects.
50
Figure 3. Time domain measures of heart rate variability in relation to age in 114 healthy subjects.
51
Figure 4. Frequency domain measures of heart rate variability in relation to age in 114 healthy
subjects.
52
Figure 5. Measures of complexity and fractal scaling of R-R interval dynamics in relation to
age in 114 healthy subjects.
54
The complexity (ApEn) and the short-term (α1) and longer-term temporal correlation
properties (α2 and β) of R-R intervals did not differ between children (n=27) and young
adults (n=29) (Table 9). However, the total variance and all the power spectral measures
were lower in children than in young adults. An age-related linear decrease of ApEn (r=0.69) and β (r=0.60) and an corresponding increase of α2 (r=0.63) took place in middle
age and old age (p<0.001 for all) (Figure 5, Table 9). Middle-aged (n=29) and elderly
subjects (n=29) had significantly lower values for β and ApEn and higher values for α2
than the two younger groups (Table 9). The short-term scaling exponent, α1, did not
differ between the three adult groups. A decrease of all time and frequency domain
measures also took place over age in adults (Table 9). The differences in various indices
between the age groups were similar during the daytime and the nighttime (Table 9) (IV).
To find out if a decrease in total HR variability explains the changes in the dynamical
measures of R-R interval variability with increasing age, an analysis of covariance
(ANCOVA) was performed using SDNN and age-group as explanatory variables and
each of the four dynamical measures of R-R interval variability as dependent variables.
The significant differences in ApEn, β, α1 and α2 between the groups remained after
adjustment for SDNN (p<0.001 for each).
In all age groups, ApEn was higher (p<0.001 in each group), α1 lower (p<0.001 in
children and young adults, p<0.05 in middle-aged and elderly) and all spectral measures
higher (p<0.01 for all in each group) during the sleeping times than in the daytime (Table
9).
α1 was significantly lower (1.10 ± 0.13 vs. 1.18 ± 0.16, p<0.01), β was slightly
steeper (-1.29 ± 0.21 vs. -1.21 ± 0.19, p<0.05) and VLF slightly lower (7.14 ± 0.70 vs.
7.45 ± 0.78, p<0.05) in females (n=48), while no other differences were observed
compared to males (n=66). Similar age-dependencies of the different measures of R-R
interval dynamics were observed in both males and females.
55
Table 9. Measures of 24-hour R-R interval dynamics in healthy children, young, middleaged and elderly subjects (n=114).
Age (years)
Mean R-R interval
(ms)
Children
<15 years
(n=27)
8±5
Young adults
15-39 years
(n=29)
28 ± 6
Middle-aged
40-60 years
(n=29)
50 ± 6
Elderly
>60 years
(n=29)
71 ± 5
678 ± 105 *
875 ± 121
876 ± 88
829 ± 96
24-hour
825 ± 161 *
1074 ± 117
1050 ± 127
954 ± 119
0 am – 6 am
614 ± 91 *
811 ± 146
796 ± 81
730 ± 168
9 am – 6 pm
SDNN (ms)
140 ± 46?
196 ± 39
169 ± 39
138 ± 32?
24-hour
0 am – 6 am
135 ± 36 *
92 ± 28
95 ± 41
73 ± 16
9 am – 6 pm
118 ± 33 *
98 ± 21
80 ± 23
76 ± 18
HF (ln)
6.83 ± 1.12
7.35 ± 0.94
6.10 ± 0.72 †
5.06 ± 0.61 *
24-hour
0 am – 6 am
7.27 ± 1.35
7.82 ± 0.97
6.51 ± 0.89 †
5.27 ± 0.73 *
9 am – 6 pm
6.20 ± 1.03
6.69 ± 1.04
5.55 ± 0.76 †
4.77 ± 0.59 *
LF (ln)
6.85 ± 0.97
7.74 ± 0.50 *
6.66 ± 0.74
5.73 ± 0.63 *
24-hour
0 am – 6 am
7.98 ± 0.55 *
6.90 ± 0.93
6.01 ± 0.73 *
7.03 ± 1.09
9 am – 6 pm
6.66 ± 0.98
7.49 ± 0.65 *
6.44 ± 0.66
5.39 ± 0.65 *
VLF (ln)
6.72 ± 0.45 *
7.30 ± 0.58
7.19 ± 0.75
8.07 ± 0.53 *
24-hour
0 am – 6 am
7.17 ± 0.48 *
7.84 ± 0.71
7.60 ± 0.76
8.52 ± 0.39 *
9 am – 6 pm
6.46 ± 0.51 *
7.05 ± 0.67
7.90 ± 0.62 *
7.22 ± 0.59
ULF (ln) 24-hour
9.39 ± 0.81
10.02 ± 0.66 *
9.54 ± 0.56
9.25 ± 0.43 *
-1.15 ± 0.18
-1.12 ± 0.19
-1.32 ± 0.14 †
-1.38 ± 0.17 ‡
β 24-hour
α1
1.19 ± 0.14
1.19 ± 0.16
1.15 ± 0.16
1.06 ± 0.11 §
24-hour
1.26 ± 0.20
1.13 ± 0.21
0 am – 6 am
0.91 ± 0.18 §
1.01 ± 0.21
9 am – 6 pm
1.13 ± 0.13
1.20 ± 0.16
1.24 ± 0.14
1.15 ± 0.16
α2
0.98 ± 0.06
1.00 ± 0.08
1.07 ± 0.07 †
1.14 ± 0.07 *
24-hour
0 am – 6 am
0.96 ± 0.10 §
0.99 ± 0.10
1.06 ± 0.10
1.12 ± 0.11
9 am – 6 pm
0.95 ± 0.10
0.98 ± 0.09
1.06 ± 0.08 †
1.12 ± 0.10 ‡
ApEn
1.26 ± 0.12
1.21 ± 0.14
1.01 ± 0.16 †
0.88 ± 0.16 *
24-hour
0 am – 6 am
1.37 ± 0.13
1.34 ± 0.19
1.27 ± 0.17
1.06 ± 0.15 *
9 am – 6 pm
1.18 ± 0.14
1.12 ± 0.12
0.98 ± 0.13 †
0.91 ± 0.17 ‡
Values are mean ± SD. β = slope of the power-law relationship of HRV; α2 = intermediate-term
scaling exponent; SDNN = standard deviation of R-R intervals, ULF = ultra low frequency power,
VLF = very low frequency. For other abbreviations, see Table 3. The symbols indicate the
difference between groups in one-way ANOVA followed by Bonferroni post hoc analysis with a
confidence level of p<0.05. * = group differed from the other three groups, † = middle-aged
subjects differed from children and young adults, ‡ = elderly subjects differed from children and
young adults, § = children differed from middle-aged and elderly subjects, ? = group differed from
young adults and middle-aged subjects.
56
Table 10. Correlation matrix of the relationship between heart rate and measures of 24hour R-R interval dynamics in healthy subjects (n=114).
Mean
R-R
Mean
R-R
interval
1.00
SDNN
HF
LF
VLF
ULF
β
α1
α2
ApEn
0.58 #
0.14
0.38 #
0.53 #
0.37 #
0.04
0.19 *
0.25 §
-0.17
1.00
0.48 #
1.00
0.70 #
0.72 #
1.00
0.70 #
0.52 #
0.81 #
1.00
0.80 #
0.50 #
0.65 #
0.71 #
1.00
0.24 *
0.36 #
0.46 #
0.47 #
0.24 §
1.00
0.13
-0.48 #
-0.15
-0.06
-0.04
-0.06
1.00
-0.25 §
-0.54 #
-0.56 #
-0.21 *
-0.26 §
-0.33 #
0.24 *
1.00
0.05
0.50 #
0.52 #
0.36 #
0.16
0.50 #
-0.55 #
-0.65 #
1.00
interval
SDNN
HF
LF
VLF
ULF
β
α1
α2
ApEn
* = p<0.05, § = p<0.01, # = p<0.001. For other abbreviations, see Tables 3, 7 and 9.
57
6. Discussion
6.1. Interindividual variation of HRV and BRS in healthy middleaged subjects
Large interindividual variation in HRV and BRS has been described in healthy subjects,
including relatively small samples of subjects without evidence of heart disease (AbdelRahman et al. 1994; Huikuri et al. 1990; Kupari et al. 1993; Molgaard et al. 1994). In
agreement with the results of the previous smaller studies, wide interindividual variation
was observed in the traditional nonspectral and spectral measures of HRV in this large
cohort of middle-aged subjects (II,V). The interindividual variation of HRV was more
than twofold compared to the variation in the mean heart rate. Large interindividual
variation was also observed in BRS and in the normalized spectral components of HRV.
These results show that both tonic autonomic regulation, expressed as HRV, and reflex
regulation, expressed as BRS, vary markedly between healthy subjects. The
interindividual variation in autonomic function in healthy subjects may indicate that some
subjects have a better capacity to adapt to changing conditions than some others. it is not
known whether these differences in autonomic function between healthy middle-aged
subjects have any prognostic significance.
6.2. Relations of HRV and BRS to demographic, lifestyle and
laboratory factors in healthy middle-aged subjects
The observed interindividual variation in the autonomic function (Abdel-Rahman et al.
1994; Huikuri et al. 1990; Kupari et al. 1993; Molgaard et al. 1994) has been suggested
to partly depend on the effect of various risk factors on HRV and BRS. Although HRV
has been found to be related to many clinical, lifestyle and laboratory factors (Dixon et
al. 1992; Goldsmith et al. 1992; Hayano et al. 1990; Jensen-Urstad et al. 1998; Kupari et
al. 1993; Molgaard et al. 1994; Tsuji et al. 1996), the effects of various factors on HRV
may nevertheless differ between the sexes (Jensen-Urstad et al. 1998; Kupari et al. 1993;
Stein et al. 1997). On the basis of the results of this study, most importantly age and
58
glucose metabolism in both men and women and additionally blood pressure and lipid
values in men, appear to have great relevance in relation to autonomic function. BRS was
related to blood pressure in both men and women and also to values of glucose
metabolism in women.
Advancing age during adult life is associated with a reduction in overall HRV (Bigger
et al. 1995; Hellman et al. 1976; Korkushko et al. 1991; Lipsitz et al. 1990; O'Brien et
al. 1986; Shannon et al. 1987) and a reduction in BRS (Gribbin et al. 1971; Laitinen et
al. 1998). Age has also remained an independent determinant of autonomic function in
studies with multivariate analysis (Molgaard et al. 1994, Tsuji et al. 1996). Here, too, the
time and frequency domain measures were related to age in both men and women.
HRV and BRS have been suggested to be related to certain personality features,
lifestyle and physical activity in healthy subjects (Barney et al. 1988; Dixon et al. 1992;
Goldsmith et al. 1992; Molgaard et al. 1991; Molgaard et al. 1994). In this large sample
of middle-aged subjects, lifestyle factors did not correlate with autonomic function. This
is contradictory to the previous smaller studies (Dixon et al. 1992; Goldsmith et al. 1992;
Hayano et al. 1990; Jensen-Urstad et al. 1998; Malpas et al. 1991; Molgaard et al. 1991;
Molgaard et al. 1994; Kupari et al. 1993). Few of the present subjects were either
physically inactive or well trained, which may partly explain the differences in the effects
of physical activity (Dixon et al. 1992; Goldsmith et al. 1992; Molgaard et al. 1991). A
trend towards lower values of HRV was also observed in smokers, but it did not reach
statistical significance.
Both HRV and BRS showed inverse correlations with glucose and insulin levels.
These findings agree with several previous studies on the effects of glucose metabolism
on cardiovascular autonomic regulation.
There are various speculative mechanisms on how risk factors may affect autonomic
function. The autonomic function may be altered in any part of the reflex arch, i.e. the
sensory receptors, the afferent pathways, the central connections, the efferent pathways
and the effector organs. Since all cardiovascular reflexes interact in a complex manner, it
is difficult to study the exact mechanisms of altered autonomic function. The altered
autonomic function with aging has been suggested to be due to both structural factors
(e.g., loss of sinoatrial pacemaker cells) and functional changes (e.g., altered coupling
between regulatory components) (Goldberger 1996). Arterial distensibility and thus
altered baroreceptor function may be associated with aging, elevated blood pressure and
atherogenic lipid metabolism (Bonyhay et al 1996; Kingwell et al. 1995; Kupari et al.
1993; Zanchetti et al. 1991). Altered glucose metabolism and extensive alcohol
consumption may have effects on the neural pathways, as in diabetic or alcoholic
neuropathy (Ewing et al. 1981; Malpas et al. 1991), while smoking may affect various
parts of the autonomic system (Hayano et al. 1990b). Various factors may also alter heart
rate more directly, which in turn may have effects on HRV measures.
The normalized units of the HF and LF spectral power of HRV were not related to
age, lifestyle or risk factors. Recent studies using direct muscle sympathetic nerve
activity as a reference index have suggested that the LF and HF spectral components
analyzed in normalized units may provide information on sympathetic and vagal outflow,
respectively, in subjects without structural heart disease (Montano et al 1998; Pagani et
al. 1997). Although the interpretation of the physiologic background of normalized
spectral units has been challenged (Eckberg 1997), the present data may suggest that
59
although the magnitude of overall HRV is related to age and laboratory variables, the
specific autonomic input, which may be mediated by central mechanisms (Pagani et al.
1997), is not related to these variables. Thus, it may be speculated that the normalized
units of HF and LF power reflect the intrinsic autonomic modulation of heart rate
independently of cardiovascular risk factors.
In summary, relatively weak relations were observed here between the traditional
measures of HRV or BRS and demographic variables, lifestyle, personality type or
laboratory variables. The differences in various clinical and laboratory factors have also
explained only a small proportion of the large interindividual variation in HRV in the
previous smaller studies (Molgaard et al. 1994). Thus, genetic or some other unmeasured
factors may explain most of the large interindividual variation in overall HRV and BRS
in healthy subjects. The large interindividual variation among healthy children (IV)
similarily suggests that genetic factors may largely determine autonomic function.
6.3. Sex-related differences in HRV and BRS
Previous studies have revealed sex-related differences in HRV and BRS (Abdel-Rahman
et al. 1994; Cowan et al. 1998; Jensen-Urstad et al. 1998; Laitinen et al. 1998; Liao et al.
1995; Molgaard et al. 1994). In these studies men have usually had higher BRS and
higher overall HRV, particularly at lower frequencies, compared to women, but women
have shown lower LFHF ratios. The results of this study (II) also indicate that there are
sex-related differences in cardiovascular autonomic regulation. The heart rate response to
an abrupt rise in blood pressure and the LF modulation of heart rate analyzed in
normalized units are lower in women than in men, whereas the HF modulation of heart
rate is higher in women. The heart rate response to the overshoot phase of the Valsalva
maneuver reflects the reflex vagal activity in response to a rapid rise of arterial blood
pressure, and the LF component of HRV is determined by spontaneous oscillation of
blood pressure, also reflecting the baroreflex-mediated control of heart rate. Consistently
with the previous studies (Abdel-Rahman et al. 1994; Laitinen et al. 1998) these
observations show that the baroreflex responsiveness is attenuated, whereas the tonic
vagal activity is augmented in women compared to men. The mechanisms of the sexrelated differences in R-R interval dynamics are not known. Possible effects of sex
hormones (Ryan et al. 1994) and differences in baseline variables, such as blood pressure
(Ryan et al. 1994) have been speculated. After adjustment for differences in several
baseline variables, the observed sex-related differences in HRV remained unchanged,
suggesting that the mechanisms behind gender-related differences are probably more
closely related to hormonal or genetic factors than to differences in lifestyle or laboratory
values. BRS and total HRV were higher in the postmenopausal women who were on
estrogen replacement therapy compared to those without hormone therapy, suggesting
that hormonal factors may partly explain the sex-related differences in autonomic
modulation of heart rate.
60
6.4. HRV and BRS in hypertensive subjects with and without insulin
resistance syndrome
Previous studies have shown that both HRV and BRS are impaired in patients with
systemic hypertension (Bristow et al. 1969; Chakko et al. 1993; Dassi et al. 1991;
Eckberg 1979; Guzzetti et al. 1988; Huikuri et al. 1996; Langewitz et al. 1994; Mancia
et al. 1983; Petretta et al. 1995; Radaelli et al. 1994; Sleight 1979; Ylitalo et al. 1997).
The possible effects of metabolic features have not been studied in patients with
hypertension, although a large proportion of hypertensive subjects are known to have IRS
(hypertension with insulin resistance, hyperinsulinemia, atherogenic lipid abnormalities
and central body obesity) (Reaven 1988).
The results of this study (III) show that the impairment of overall HRV is confined to
the hypertensive subjects with metabolic features of IRS (hyperinsulinemia and
hypertriglyceridemia), while BRS and respiratory modulation of heart rate are impaired
even in hypertensive subjects without IRS. In multiple regression analysis, both
hypertension and metabolic features of IRS predicted the HRV values, but BRS only
inversely related only to systolic blood pressure. Thus, metabolic aspects and elevated
blood pressure have independent roles in the impairment of total HRV. The observed
abnormalities in autonomic function might be related to the increased cardiovascular risk
of subjects with IRS (Casassus et al. 1992; Despres et al. 1994; Pyörälä et al. 1985).
The HF component of HRV and BRS mainly reflect the tonic and reflex vagal
outflow, respectively. Thus, cardiac vagal function seems to be impaired in systemic
hypertension even in the subjects without features of IRS. There are some disagreements
as to whether the LF component represents either sympathetic or vagal outflow in
humans (Eckberg 1997; Liao et al. 1995; Malik et al. 1993). Studies on patients with
heart failure (van de Borne et al. 1997) and on healthy subjects during exercise (Arai et
al. 1989; Breuer et al. 1993) have consistently revealed a reduction in LF power, which
has been attributed to increased sympathetic activity. This is speculated to be due to
saturation of the LF oscillatory systems caused by the high sympathetic drive or the
mechanism that includes the central effect of neurohumoral excitation (van de Borne et
al. 1997). Insulin resistance is assumed to be associated with altered sympathetic nerve
discharge (Arauz-Pacheco et al. 1996; Facchini et al. 1996; Reaven et al. 1996; Stern et
al. 1992; Troisi et al. 1991). Insulin-resistant subjects have increased plasma and urinary
concentrations of norepinephrine (Arauz-Pacheco et al. 1996; Troisi et al. 1991) and an
impaired ability of insulin to stimulate sympathetic activity and muscle blood flow
(Laakso et al. 1990; Vollenweider et al. 1994; Vollenweider et al. 1995). Although the
underlying mechanisms of this impairment are unknown, these factors may also be linked
to the impairment of LF modulation of heart rate.
61
6.5. Dynamical measures of HRV in healthy subjects
6.5.1. Dynamical analysis of R-R intervals
New dynamical methods of R-R interval variability based on fractals (“chaos theory”)
and complexity have been used in conjunction with the traditional measures, since they
may give complementary information of HR dynamics by revealing “hidden”
abnormalities or alterations in time series data that are not otherwise apparent (Ho et al
1997; Hogue et al 1998, Lipsitz & Goldberger 1992; Mäkikallio et al 1996, 1997, 1998).
Recently, dynamical analysis of HR variability with the detrended fluctuation method has
provided better prognostic information than the traditional methods of HR variability in
various populations (Mäkikallio et al 1999, in press), and the complexity measure,
approximate entropy (ApEn), has been shown to predict the onset of atrial fibrillation
(Vikman et al). The mathematical background of the new dynamical measures of R-R
interval variability used in this study has been described in detail previously (Bigger et
al. 1996; Huikuri et al. 1998; Iyengar et al. 1996; Mäkikallio et al. 1996; Peng et al.
1995; Pincus et al. 1992; Pincus et al. 1994; Saul et al 1987). Briefly, ApEn indicates the
predictability or complexity of time series data (Mäkikallio et al. 1996; Pincus et al.
1992; Pincus et al. 1994), while the slope of the power-law relationship represents the
fractal-like correlation properties of R-R interval data over very-low and ultra-low
frequency bands (10-4 to 10-2 Hz)(Bigger et al. 1996; Huikuri et al. 1998; Saul et al
1987), and α1 and α2 obtained by detrended fluctuation analysis (DFA) stand for the
correlation properties of short-term and intermediate-term R-R interval fluctuations,
respectively (Iyengar et al. 1996; Peng et al. 1995). The fluctuations of a time series can
also be described by comparing their behavior to various types of “noise” with different
dynamics. White noise represents time series where no correlations are found, and the
data points are completely random. The frequency spectrum is flat, because all
frequencies are represented in equal densities. The slope of the power-law relationship
(β) is 0 and the fractal scaling exponent (α) is 0.5. Brownian noise (random walk or 1/f 2
noise), can be observed upon integration of a white noise signal, and it is characterized
by the frequency spectrum of a curve rapidly decaying (power inversely proportional to
frequency squared) with β=-2 and α=1.5. For 1/f noise, the frequency curve is smooth
with fluctuations inversely proportional to frequency (1/f) and with β=-1 and α=1.
Values of ApEn are higher for 1/f noise compared to 1/f 2 noise. 1/f noise represents an
example of a fractal-like process characterized by fluctuations that display scale
invariance (self-similarity) and long-range correlations. Such processes generate irregular
and complex fluctuations over multiple time scales.
6.5.2. Effect of aging from childhood to advanced age on R-R interval
dynamics
In this study, children had similar complexity (ApEn) and similar wide-range temporal
correlation properties of HR behavior (α1, α2 and β) as young adults. Both children and
62
young adults showed R-R interval dynamics resembling that of 1/f behavior (α1 and α2
∼1.0, β ∼ -1.0) consistent with a system with fractal-like, scale-invariant correlations. Our
findings confirm the reduced frequency domain measures in young children vs. young
adults obtained previously (Finley et al. 1995; Korkushko et al. 1991).
A steeper β (slope of power-law relationship), a decrease of ApEn (complexity), and
an increase of α2 (intermediate-term scaling exponent) were observed upon increasing
age, suggesting that the longer-term R-R interval dynamics change from 1/f behavior
towards 1/f2 behavior. These findings are consistent with the lower complexity (higher
regularity and predictability) of R-R interval dynamics seen with increasing age. All time
and frequency domain measures of HRV also decrease during adult life with increasing
age, as evidenced by the lower total variance and smaller power seen at all frequencies.
These observations are consistent with the previous findings showing decreased total
variance (Hellman et al. 1976; Lipsitz et al. 1990; O'Brien et al. 1986; Shannon et al.
1987), decreased spectral power of VLF, LF and HF (Bigger et al. 1995), steeper slopes
of the power-law relationship (Lipsitz et al. 1990) and reduced ApEn values (Kaplan et
al. 1991; Ryan et al. 1994) in old age. The values of the short-term scaling exponent, α1,
did not change significantly during healthy adulthood in this study, although children had
somewhat lower values than elderly subjects.
The findings of increased ApEn, decreased α1 and increased spectral components in
the nighttime indicate increased variance and complexity of HR dynamics at night. The
age-dependence of different measures of R-R interval dynamics was similar regardless of
whether they were analyzed from 24-hour or nighttime recordings. Thus, the differences
in physical activity between the different age-groups (children vs. elderly) during the
daytime do not explain the observed age-related changes in R-R interval dynamics, since
the nighttime (0 - 6 am) can be considered a time period when physical activity is likely
to be most comparable between subjects.
6.5.3. Determinants of correlation properties and complexity of R-R
intervals in healthy middle-aged subjects
According to the results of this study (V), the fractal and complexity measures of
HRV show smaller interindividual variation compared to the traditional measures of
HRV among healthy subjects and are thus within a narrow range in the subjects without
evidence of structural heart disease. The dynamical measures were not related to any
demographic, lifestyle or laboratory values. However, a sex-difference was observed.
Women had a lower short-term scaling exponent, α1, and a higher ApEn compared to
men, despite the decreased overall variance, confirming that the R-R interval dynamics of
healthy middle-aged women are characterized by lower total variance but more complex,
fractal-like behavior compared to men. The sex-difference persisted even after
adjustment for differences in several baseline variables, suggesting that the mechanisms
of the observed difference propably involve hormonal or genetic factors. These measures
may quantify the “intrinsic” capacity of the healthy cardiovascular control system, which
only breaks down when disease states develop.
63
6.5.4. Physiological interpretation of R-R interval dynamics
It has been suggested that scale invariance may be a central organizing principle of
physiological structure and function. The breakdown of this scale-invariant fractal
organization could lead to either totally uncorrelated randomness or highly predictable
(single-scale) behavior, both of which may result in a less adaptable system (Goldberger
1996). Thus, changes from 1/f scale-invariant behavior towards behavior resembling
either random fluctuations (white noise) or 1/f2 behavior with less complexity might be
physiologically deleterious. Such changes seem to occur upon physiologic aging. In
contrast, children already show a “mature” pattern of R-R interval dynamics comparable
to that of healthy young adults, with complex fractal dynamics suggesting a highly
adaptive cardiovascular regulatory system.
The age-related changes in different measures of R-R interval dynamics are probably
markers of the various physiological mechanisms affecting these measures, especially
neuroautonomic inputs (Akselrod et al. 1985; Bigger et al. 1996). The finding that
children showed a similar slope of the power-law relationship of R-R interval dynamics
compared to young adults despite the reduced power of various spectral components,
indicates that these indexes are differentially regulated and cannot be used as surrogates
for “complexity” measures. Further studies are needed to address the physiological
background of the dynamical measures of HRV.
6.6. Limitations of the study
Due to the primary selection criteria of the large original study population, which only
excluded the subjects with refund for antihypertensive medication from the control
group, some subjects were excluded from the studies on healthy subjects (I,II,V) because
of diabetes, CAD or chronic arrhythmias. Thus, our population may not represent a truly
random sample of healthy middle-aged subjects. Since exercise tolerance test was not
used to evaluate CAD, some asymptomatic subjects with CAD may have been included,
although those with symptoms, medication, or electrocardiographic evidence of CAD
were excluded.
Because of the large original study population (I-III,V), 24-hour ECG recording to
could not be used to assess HRV. According to our pilot study, reasonably good
correlations (r between 0.61 and 0.81) were found between 24-hour HRV and HRV
analyzed from our specific standard 45-minute recording (Huikuri et al. 1996). The
dynamical measures from short-term ECG recordings may show less correlation to the
24-hour values, however. Artifacts and nonstationarities in the ECG were mostly found
during the 15-minute walking period. Heart rate behavior was studied in more controlled
and stationary conditions, including supine and sitting periods in the studies II and V.
Thus, more subjects were included and the effects of nonstationarities on the results were
possibly better eliminated in these studies. However, the frequency and depth of
breathing were not controlled, which means that the conditions were not completely
standardized. 24-hour ECG recordings during normal activities have been recommended
by the Task Force (1996) for HRV testing for risk stratification because of the better
64
reproducibility of long-term vs. short-term recordings. We therefore tested the R-R
interval dynamics of 24-hour recordings of healthy subjects despite the potential
confounding effects of nonstationarities on R-R interval data (IV).
The “golden standard” method, i.e. the phenylephrine test, was not used in this study
of a large population to assess the BRS because of its invasive nature. The validity and
reproducibility of the Valsalva maneuver in the assessment of BRS have been previously
described (Airaksinen et al. 1993), but the inadequate increase in blood pressure after the
release of the Valsalva strain unfortunately limits its use in a certain proportion of
subjects, as also found in our study.
Our criteria for IRS were chosen not to underestimate the incidence of IRS in the
population sample (III). The criteria seemed to be able to distinguish between the groups
in terms of all major features of the syndrome, i.e. hypertensives with IRS were more
obese, had poorer glucose tolerance with higher insulin levels and had an atherogenic
lipid profile (Reaven 1988). However, the euglycemic clamp technique was not used, and
the actual insulin resistance of the subjects therefore remains open. Due to the large
original study population, the antihypertensive medication could not be discontinued
(III). We therefore matched the hypertensive subjects with and without IRS according to
the medication, to make the groups comparable. Medication may influence the results,
although theoretically, based on previous studies (Lucini et al. 1993; Niemelä et al.
1994), medication might rather increase than decrease the HRV and BRS of hypertensive
subjects.
This study included multiple analyses between various factors and autonomic function
and multiple comparisons between different groups. Although Bonferroni adjustments
and multivariate analysis were used, the problem of multiplicity exists. This crosssectional study was an observational one. Thus, although significant associations of age,
sex and cardiovascular risk factors with autonomic function were observed, the results do
not actually give information on causality. It remains to be ascertained whether
autonomic cardiovascular function can be modified with interventions concerning various
risk factors.
65
7. Conclusions
1. A large interindividual variation is observed in the measures of autonomic
cardiovascular function in healthy middle-aged subjects. The measures of overall HRV
are related to several cardiovascular risk factors, most importantly to age and factors of
glucose metabolism in both men and women and, in addition, to blood pressure and lipid
metabolism in men. BRS is related to blood pressure in both men and women. However,
demographic and metabolic factors only explain a small portion of the interindividual
variation in autonomic function.
2. There are sex-related differences in autonomic cardiovascular function. Overall HRV
and BRS are attenuated in middle-aged women compared to men, but the HF component
of HRV is augmented.
3. In hypertensive subjects, the impairment of overall HRV is confined to those with
metabolic features of IRS (hyperinsulinemia and hypertriglyceridemia), but the BRS and
respiratory modulation of heart rate are also impaired in hypertensive subjects without
IRS. Both hypertension and metabolic features of IRS predict the HRV values, but BRS
is only related to systolic blood pressure.
4. There are gradual significant changes in R-R interval dynamics from childhood to old
age in healthy subjects. Children show comparable complexity and fractal correlation
properties of R-R interval time series as young adults, suggesting a highly adaptive
cardiovascular regulatory system despite the lower overall HRV. Healthy aging results in
R-R interval dynamics with higher regularity and predictability and altered fractal scaling
consistent with a possibly harmful loss of complex variability.
5. Healthy middle-aged subjects show little interindividual variation in their fractal
correlation properties and complexity of heart rate behavior, suggesting that these
dynamical measures quantify the “intrinsic” capacity of a healthy cardiovascular control
system without a significant influence of life-style, metabolic or demographic variables.
However, there are sex-related differences in fractal correlation properties and
complexity of heart rate behavior.
66
8. References
Abdel-Rahman AR, Merrill RH & Wooles WR (1994) Gender-related differences in the
baroreceptor reflex control of heart rate in normotensive humans. J Appl Physiol 77: 606-613.
Airaksinen KEJ, Hartikainen JE, Niemelä MJ, Huikuri HV, Mussalo HM & Tahvanainen KU
(1993) Valsalva manoeuvre in the assessment of baroreflex sensitivity in patients with coronary
artery disease. Eur Heart J 14: 1519-1523.
Airaksinen KEJ, Ikäheimo MJ, Linnaluoto MK, Niemelä M & Takkunen JT (1987) Impaired vagal
heart rate control in coronary artery disease. Br Heart J 58: 592-597.
Airaksinen KEJ, Niemelä MJ & Huikuri HV (1994) Effect of beta-blockade on baroreflex
sensitivity and cardiovascular autonomic function tests in patients with coronary artery disease.
Eur Heart J 15: 1482-1485.
Akselrod S, Gordon D, Madwed JB, Snidman NC, Shannon DC & Cohen RJ (1985)
Hemodynamic regulation: investigation by spectral analysis. Am J Physiol 249: H867-75.
Akselrod S, Gordon D, Ubel FA, Shannon DC, Berger AC & Cohen RJ (1981) Power spectrum
analysis of heart rate fluctuation: a quantitative probe of beat-to-beat cardiovascular control.
Science 213: 220-222.
Algra A, Tijssen JG, Roelandt JR, Pool J & Lubsen J (1993) Heart rate variability from 24-hour
electrocardiography and the 2-year risk for sudden death. Circulation 88: 180-185.
Arai Y, Saul JP, Albrecht P, Hartley LH, Lilly LS, Cohen RJ & Colucci WS (1989) Modulation of
cardiac autonomic activity during and immediately after exercise. Am J Physiol 256: H132-41.
Arauz-Pacheco C, Lender D, Snell PG, Huet B, Ramirez LC, Breen L, Mora P & Raskin P (1996)
Relationship between insulin sensitivity, hyperinsulinemia, and insulin-mediated sympathetic
activation in normotensive and hypertensive subjects. Am J Hypertens 9: 1172-1178.
Bailey JR, Fitzgerald DM & Applegate RJ (1996) Effects of constant cardiac autonomic nerve
stimulation on heart rate variability. Am J Physiol 270: H2081-H2087.
Barney JA, Ebert TJ, Groban L, Farrell PA, Hughes CV & Smith JJ (1988) Carotid baroreflex
responsiveness in high-fit and sedentary young men. J Appl Physiol 65: 2190-2194.
Berne RM, Levy MN, Regulation of the heartbeat. In Physiology. (Eds. R. M. Berne and M. N.
Levy) pp. 417-437, Mosby Year Book, Inc. 1993.
Bigger JT, Jr., Fleiss JL, Rolnitzky LM & Steinman RC (1993) Frequency domain measures of
heart period variability to assess risk late after myocardial infarction J Am Coll Cardiol 21:
729-736.
Bigger JT, Jr., Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger RE & Rottman JN (1992)
Frequency domain measures of heart period variability and mortality after myocardial
infarction. Circulation 85: 164-171.
Bigger JT, Jr., Fleiss JL, Steinman RC, Rolnitzky LM, Schneider WJ & Stein PK (1995) RR
variability in healthy, middle-aged persons compared with patients with chronic coronary heart
disease or recent acute myocardial infarction. Circulation 91: 1936-1943.
67
Bigger JT, Jr., La Rovere MT, Steinman RC, Fleiss JL, Rottman JN, Rolnitzky LM & Schwartz PJ
(1989) Comparison of baroreflex sensitivity and heart period variability after myocardial
infarction. J Am Coll Cardiol 14: 1511-1518.
Bigger JT, Jr., Steinman RC, Rolnitzky LM, Fleiss JL, Albrecht P & Cohen RJ (1996) Power law
behavior of RR-interval variability in healthy middle-aged persons, patients with recent acute
myocardial infarction, and patients with heart transplants. Circulation 93: 2142-2151.
Bonaduce D, Marciano F, Petretta M, Migaux ML, Morgano G, Bianchi V, Salemme L, Valva G
& Condorelli M (1994) Effects of converting enzyme inhibition on heart period variability in
patients with acute myocardial infarction. Circulation 90: 108-113.
Bonyhay I, Jokkel G and Kollai M (1996) Relation between baroreflex sensitivity and carotid
artery elasticity in healthy subjects. Am. J. Physiol. 271, H1139-H1144.
Bortner RW (1969) A short rating scale as a potential measure of pattern A behavior. J Chronic
Dis 22: 87-91.
Breuer HW, Skyschally A, Schulz R, Martin C, Wehr M & Heusch G (1993) Heart rate variability
and circulating catecholamine concentrations during steady state exercise in healthy volunteers.
Br Heart J 70: 144-149.
Bristow JD, Honour AJ, Pickering GW, Sleight P & Smyth HS (1969) Diminished baroreflex
sensitivity in high blood pressure. Circulation 39: 48-54.
Brouwer J, vanVeldhuisen DJ, Man int Veld AJ, Haaksma J, Dijk WA, Visser KR, Boomsma F,
Dunselman PHJM & Lie KI (1996) Prognostic value of heart rate variability during long-term
follow-up in patients with mild to moderate heart failure. J Am Coll Cardiol 28: 1183-1189.
Casadei B, Cochrane S, Johnston J, Conway J & Sleight P (1995) Pitfalls in the interpretation of
spectral analysis of the heart rate variability during exercise in humans. Acta Physiol Scand
153: 125-131.
Casassus P, Fontbonne A, Thibult N, Ducimetiere P, Richard JL, Claude JR, Warnet JM, Rosselin
G & Eschwege E (1992) Upper-body fat distribution: a hyperinsulinemia-independent predictor
of coronary heart disease mortality. The Paris Prospective Study. Arterioscler Thromb 12:
1387-1392.
Casolo G, Balli E, Taddei T, Amuhasi J & Gori C (1989) Decreased spontaneous heart rate
variability in congestive heart failure. Am J Cardiol 64: 1162-1167.
Chakko S, Mulingtapang RF, Huikuri HV, Kessler KM, Materson BJ & Myerburg RJ (1993)
Alterations in heart rate variability and its circadian rhythm in hypertensive patients with left
ventricular hypertrophy free of coronary artery disease. Am Heart J 126: 1364-1372.
Chess GF, Tam RMK & Calaresu FR (1975) Influence of cardiac neural inputs on rhythmic
variations of heart period in the cat. Am J Physiol 228: 775-780.
Cloarec-Blanchard L, Girard A, Houhou S, Grunveld JP & Elghozi JL (1992) Spectral analysis of
short-term blood pressure and heart rate variability in uremic patients. Kidney International
Suppl 37: S14-S18.
Coker R, Koziell A & Smith SE (1984) Does the sympathetic nervous system influence sinus
arrhtyhmia in man? Evidence from combined autonomic blockade. J Physiol (Lond) 356: 459464.
Consensus conference: Treatment of hypertriglyceridemia. (1984) JAMA 251: 1196-1200.
Cook JR, Bigger JT, Jr., Kleiger RE, Fleiss JL, Steinman RC & Rolnitzky LM (1991) Effect of
atenolol and diltiazem on heart period variability in normal persons. J Am Coll Cardiol 17:
480-484.
Cooley RL, Montano N, Cogliati C, van de Borne P, Richenbecher W, Oren R & Somers VK
(1998) Evidence for a central origin of the low-frequency oscillation in RR-interval variability.
Circulation 98: 556-561.
Cowan MJ, Pike K & Burr R (1998) Effects of gender and age on heart rate variability in healthy
individuals and in persons after sudden cardiac arrest. Journal of electrocardiology 27 Suppl: 19.
Dassi S, Balsama M, Guzzetti S, Ponti GB, Magni L, Pagani M & Malliani A (1991) Twenty-fourhour power spectral analysis of heart rate variability and of arterial pressure values in
normotensive and hypertensive subjects. J Hypertens Suppl 9: S72-3.
68
Denton TA, Diamond GA, Helfant RH, Kahn S & Karaqueuzian H (1990) Fascinating rhythm: A
primer on chaos theory and its application to cardiology. Am Heart J 120: 1419-1440.
Despres JP & Marette A (1994) Relation of components of insulin resistance syndrome to
coronary disease risk. Curr Opin Lipidol 5: 274-289.
Dibner-Dunlap ME, Eckberg DL, Magid NM & Cintron-Trevino NM (1985) The long-term
increase of baseline and reflexly augmented levels of human vagal-cardiac nervous activity
induced by scopolamine. Circulation 71: 797-804.
Dixon EM, Kamath MV, McCartney N & Fallen EL (1992) Neural regulation of heart rate
variability in endurance athletes and sedentary controls. Cardiovasc Res 26: 713-719.
Eckberg DL (1979) Carotid baroreflex function in young men with borderline blood pressure
elevation. Circulation 59: 632-636.
Eckberg DL (1980) Parasympathetic cardiovascular control in human disease: a critical review of
methods and results. Am J Physiol 239: H581-93.
Eckberg DL (1983) Human sinus arrhythmia as an index of vagal cardiac outflow. Journal of
Applied Physiology: Respiratory, Environmental & Exercise Physiology 54: 961-966.
Eckberg DL (1997) Sympathovagal balance: a critical appraisal. Circulation 96: 3224-3232.
Eckberg DL, Drabinsky M & Braunwald E (1971) Defective cardiac parasympathetic control in
patients with heart disease. N Engl J Med 285: 877-883.
Ewing DJ, Borsey DQ, Bellavere F & Clarke BF (1981) Cardiac autonomic neuropathy in
diabetes: comparison of measures of R-R interval variation. Diabetologia 21: 18-24.
Facchini FS, Stoohs RA & Reaven GM (1996) Enhanced sympathetic nervous system activity. The
linchpin between insulin resistance, hyperinsulinemia, and heart rate. Am J Hypertens 9: 10131017.
Farrell TG, Paul V, Cripps TR, Malik M, Bennett ED, Ward D & Camm AJ (1991) Baroreflex
sensitivity and electrophysiological correlates in patients after acute myocardial infarction.
Circulation 83: 945-952.
Finley JP & Nugent ST (1995) Heart rate variability in infants, children and young adults. J Auton
Nerv Syst 51: 103-108.
Goldberger AL (1996) Non-linear dynamics for clinicians: chaos theory, fractals, and complexity
at the bedside. Lancet 347: 1312-1314.
Goldsmith RL, Bigger JT, Jr., Steinman RC & Fleiss JL (1992) Comparison of 24-hour
parasympathetic activity in endurance-trained and untrained young men. J Am Coll Cardiol 20:
552-558.
Goldstein DS, Horwitz D & Keiser HR (1982) Comparison of techniques for measuring baroreflex
sensitivity in man. Circulation 66: 432-439.
Gribbin B, Pickering TG, Sleight P & Peto R (1971) Effect of age and high blood pressure on
baroreflex sensitivity in man. Circ Res 29: 424-431.
Grimby G (1986) Physical activity and muscle training in the elderly. Acta Med Scand Suppl 711:
233-237.
Guzzetti S, Piccaluga E, Casati R, Cerutti S, Lombardi F, Pagani M & Malliani A (1988)
Sympathetic predominance in essential hypertension: a study employing spectral analysis of
heart rate variability. J Hypertens 6: 711-717.
Hainsworth R (1998) The physiological approach to cardiovascular reflexes. Clin Sci (Colch) 4349.
Hartikainen J, Fyrquist F, Tahvanainen K, Länsimies E & Pyörälä K (1995) Baroreflex sensitivity
and neurohormonal activation in patients with acute myocardial infarction. Br Heart J 74:2126.
Hartikainen JE, Malik M, Staunton A, Poloniecki J & Camm AJ (1996) Distinction between
arrhythmic and nonarrhythmic death after acute myocardial infarction based on heart rate
variability, signal-averaged electrocardiogram, ventricular arrhythmias and left ventricular
ejection fraction. J Am Coll Cardiol 28: 296-304.
Hartikainen J, Tarkiainen I, Tahvanainen K, Mäntysaari M, Länsimies E & Pyörälä K (1993)
Circadian variation of cardiac autonomic regulation during 24-h bed rest. Clin Physiol 13:185196.
69
Hayano J, Sakakibara Y, Yamada A, Yamada M, Mukai S, Fujinami T, Yokoyama K, Watanabe Y
& Takata K (1991a) Accuracy of assessment of cardiac vagal tone by heart rate variability in
normal subjects. Am J Cardiol 67: 199-204.
Hayano J, Sakakibara Y, Yamada M, Ohte N, Fujinami T, Yokoyama K, Watanabe Y & Takata K
(1990a) Decreased magnitude of heart rate spectral components in coronary artery disease. Its
relation to angiographic severity. Circulation 81: 1217-1224.
Hayano J, Yamada A, Mukai S, Sakakibara Y, Yamada M, Ohte N, Hashimoto T, Fujinami T &
Takata K (1991b) Severity of coronary atherosclerosis correlates with the respiratory
component of heart rate variability. Am Heart J 121: 1070-1079.
Hayano J, Yamada M, Sakakibara Y, Fujinami T, Yokoyama K, Watanabe Y & Takata K (1990b)
Short- and long-term effects of cigarette smoking on heart rate variability. Am J Cardiol 65: 8488.
Haynes SG, Levine S, Scotch N, Feinleib M & Kannel WB (1978) The relationship of
psychosocial factors to coronary heart disease in the Framingham study. I. Methods and risk
factors. Am J Epidemiol 107: 362-383.
Hellman JB & Stacy RW (1976) Variation of respiratory sinus arrhythmia with age. J Appl Physiol
41: 734-738.
Hirsch JA & Bishop B (1981) Respiratory sinus arrhythmia in humans: how breathing pattern
modulates heart rate. Am J Physiol 241: H620-9.
Ho KK, Moody GB, Peng CK, Mietus JE, Larson MG, Levy D & Goldberger AL (1997)
Predicting survival in heart failure case and control subjects by use of fully automated methods
for deriving nonlinear and conventional indices of heart rate dynamics. Circulation 96: 842848.
Hogue CWJ, Domitrovich PP, SteinPK, Despotis GD, Re L & Shuessler RB (1998) RR interval
dynamics before atrial fibrillation in patients after coronary bypass graft surgery. Circulation
98: 429-434.
Hohnloser SH, Klingenheben T, van de Loo A, Hablawetz E, Just H & Schwartz PJ (1994) Reflex
versus tonic vagal activity as a prognostic parameter in patients with sustained ventricular
tachycardia or ventricular fibrillation. Circulation 89: 1068-1073.
Hohnloser SH, Klingenheben T, Zabel M, Schröder F & Just H (1992) Intraindividual
reproducibility of heart rate variability. PACE 15: 2211-2214.
Huikuri HV, Kessler KM, Terracall E, Castellanos AX, Linnaluoto M & Myerburg RJ (1990)
Reproducibility and circadian rhythm of heart rate variability in healthy subjects. Am J Cardiol
65: 391-393.
Huikuri HV, Linnaluoto MK, Seppänen T, Airaksinen KEJ, Kessler KM & Myerburg RJ (1992)
Circadian rhythm of heart rate variability in survivors of cardiac arrest. Am J Cardiol 70: 610615.
Huikuri HV, Mäkikallio TH, Airaksinen KEJ, Seppänen T, Puukka P, Räiha I & Sourander LB
(1998) Power-law relationship of heart rate variability as a predictor of mortality in the elderly.
Circulation 97: 2031-2036.
Huikuri HV, Niemelä MJ, Ojala S, Rantala A, Ikäheimo MJ & Airaksinen KEJ (1994) Circadian
rhythms of frequency domain measures of heart rate variability in healthy subjects and patients
with coronary artery disease. Effects of arousal and upright posture. Circulation 90: 121-126.
Huikuri HV, Ylitalo A, Pikkujämsä SM, Ikäheimo MJ, Airaksinen KEJ, Rantala AO, Lilja M &
Kesäniemi YA (1996) Heart rate variability in systemic hypertension. Am J Cardiol 77: 10731077.
Hyndman BW, Kitney RI & Sayers BM (1971) Spontaneous rhythms in physiological control
systems. Nature 233: 339-341.
Iyengar N, Peng CK, Morin R, Goldberger AL & Lipsitz LA (1996) Age-related alterations in the
fractal scaling of cardiac interbeat interval dynamics. Am J Physiol 271: R1078-84.
Jensen-Urstad M, Jensen-Urstad K, Ericson M & Johansson J (1998) Heart rate variability is
related to leucocyte count in men and to blood lipoproteins in women in a healthy population
of 35-year-old subjects. Journal of Internal Medicine 243: 33-40.
70
Jensen-Urstad K, Storck N, Bouvier F, Ericson M, Lindblad LE & Jensen-Urstad M (1997) Heart
rate variaibility in healthy subjects is related to age and gender. Acta Physiol Scand 160: 235241.
Kamada T, Miyake S, Kumashiro M, Monou H & Inoue K (1992) Power spectral analysis of heart
rate variability in type As and Type Bs during mental workload. Psychosom Med 54: 462-470.
Kaplan DT, Furman MI, Pincus SM, Ryan SM, Lipsitz LA & Goldberger AL (1991) Aging and
the complexity of cardiovascular dynamics. Biophys J 59: 945-949.
Katona PG & Jih F (1975) Respiratory sinus arrhythmia: noninvasive measure of parasympathetic
cardiac control. J Appl Physiol 39: 801-805.
Kaufman ES, Bosner MS, Bigger JT, Jr., Stein PK, Kleiger RE, Rolnitzky LM, Steinman RC &
Fleiss JL (1993) Effects of digoxin and enalapril on heart period variability and response to
head-up tilt in normal subjects. Am J Cardiol 72: 95-99.
Kervinen K, Savolainen MJ, Salokannel J, Hynninen A, Heikkinen J, Ehnholm C, Koistinen MJ &
Kesäniemi YA (1994) Apolipoprotein E and B polymorphisms--longevity factors assessed in
nonagenarians. Atherosclerosis 105: 89-95.
Kingwell BA, Cameron JD, Gillies KJ, Jennings GL & Dart AM (1995) Arterial compliance may
influence baroreflex function in athletes and hypertensives. Am J Physiol 268: H411-H418.
Kleiger RE, Bigger JT, Bosner MS, Chung MK, Cook JR, Rolnitzky LM, Steinman R & Fleiss JL
(1991) Stability over time of variables measuring heart rate variability in normal subjects. Am J
Cardiol 68: 626-630.
Kleiger RE, Miller JP, Bigger JT, Jr. & Moss AJ (1987) Decreased heart rate variability and its
association with increased mortality after acute myocardial infarction. Am J Cardiol 59: 256262.
Kobayashi M & Musha T (1982) 1/f fluctuation of heartbeat period. IEEE Trans Biomed Eng 29:
456-457.
Korkushko OV, Shatilo VB, Plachinda YuI & Shatilo TV (1991) Autonomic control of cardiac
chronotropic function in man as a function of age: assessment by power spectral analysis of
heart rate variability. J Auton Nerv Syst 32: 191-198.
Korpelainen JT, Sotaniemi KA, Huikuri HV & Myllylä VV (1996) Abnormal heart rate variability
as a manifestation of autonomic dysfunction in hemispheric brain infarction. Stroke 27: 10591063.
Koskenvuo M, Kaprio J, Rose RJ, Kesäniemi A, Sarna S, Heikkilä K & Langinvainio H (1988)
Hostility as a risk factor for mortality and ischemic heart disease in men. Psychosom Med 50:
330-340.
Kupari M, Virolainen J, Koskinen P & Tikkanen MJ (1993) Short-term heart rate variability and
factors modifying the risk of coronary artery disease in a population sample. Am J Cardiol 72:
897-903.
La Rovere MT, Bigger JT, Jr., Marcus FI, Mortara A & Schwartz PJ (1998) Baroreflex sensitivity
and heart-rate variability in prediction of total cardiac mortality after myocardial infarction.
ATRAMI (Autonomic Tone and Reflexes After Myocardial Infarction) Investigators. Lancet
351: 478-484.
La Rovere MT, Specchia G, Mortara A & Schwartz PJ (1988) Baroreflex sensitivity, clinical
correlates, and cardiovascular mortality among patients with a first myocardial infarction. A
prospective study. Circulation 78: 816-824.
Laakso M, Edelman SV, Brechtel G & Baron AD (1990) Decreased effect of insulin to stimulate
skeletal muscle blood flow in obese man. A novel mechanism for insulin resistance. J Clin
Invest 85: 1844-1852.
Laitinen T, Hartikainen J, Vanninen E, Niskanen L, Geelen G & Länsimies E (1998) Age and
gender dependency of baroreflex sensitivity in healthy subjects. J Appl Physiol 84: 576-583.
Langewitz W, Ruddel H & Schachinger H (1994) Reduced parasympathetic cardiac control in
patients with hypertension at rest and under mental stress. Am Heart J 127: 122-128.
Levy MN (1971) Sympathetic-parasympathetic interactions in the heart. Circ Res 29: 437-445.
Liao D, Barnes RW, Chambless LE, Simpson RJ, Jr., Sorlie P & Heiss G (1995) Age, race, and
sex differences in autonomic cardiac function measured by spectral analysis of heart rate
variability--the ARIC study. Atherosclerosis Risk in Communities. Am J Cardiol 76: 906-912.
71
Lindqvist A, Parviainen P, Kolari P, Tuominen J, Välimäki I, Antila K & Laitinen LA (1989) A
non-invasive method for testing neural circulatory control in man. Cardiovasc Res 23: 262-272.
Lipsitz LA & Goldberger AL (1992) Loss of 'complexity' and aging. Potential applications of
fractals and chaos theory to senescence. JAMA 267: 1806-1809.
Lipsitz LA, Mietus J, Moody GB & Goldberger AL (1990) Spectral characteristics of heart rate
variability before and during postural tilt. Relations to aging and risk of syncope. Circulation
81: 1803-1810.
Lowensohn RI, Weiss M & Hon EH (1977) Heart rate variability in brain-damaged adults. Lancet
1: 626-628.
Lucini D, Pagani M & Malliani A (1993) Improved baroreflex control of the heart rate with
chronic beta-adrenergic blockade in mild hypertension. J Hypertens Suppl 11 Suppl 5: S156-7.
Madwed JB, Albrecht P, Mark RG & Cohen RJ (1989) Low-frequency oscillation in arterial
pressure on heart rate: a simple computer model. Am J Physiol 256: H1573-H1579.
Malik M & Camm AJ (1993) Components of heart rate variability--what they really mean and
what we really measure. Am J Cardiol 72: 821-822.
Malik M, Farrell T, Cripps TR & Camm AJ (1989) Heart rate variability in relation to prognosis
after myocardial infarction: selection of optimal processing techniques. Eur Heart J 10: 10601074.
Malliani A, Pagani M, Lombardi F & Cerutti S (1991) Cardiovascular neural regulation explored
in the frequency domain. Circulation 84: 482-492.
Malpas SC, Whiteside EA & Maling TJB (1991) Heart rate variability and cardiac autonomic
function in men with chronic alcohol dependence. Br Heart J 65: 84-88.
Mancia G, Ferrari A, Gregorini L, Parati G, Pomidossi G, Bertinieri G, Grassi G, di Rienzo M,
Pedotti A & Zanchetti A (1983) Blood pressure and heart rate variabilities in normotensive and
hypertensive human beings. Circ Res 53: 96-104.
Martin GJ, Magid NM, Myers G, Barnett PS, Schaad JW, Weiss JS, Lesch M & Singer DH (1987)
Heart rate variability and sudden death secondary to coronary artery disease during ambulatory
electrocardiographic monitoring. Am J Cardiol 60: 86-89.
Mäkikallio TH, Hoiber S, Kober L, Torp-Pedersen C, Peng CK, Goldberger AL, Huikuri HV and
the TRACE Investigators (1999) Fractal analysis of heart rate dynamics as a predictor of
mortality in patients with depressed left ventricular function after acute myocardial infarction.
Am J Cardiol 83:880-884.
Mäkikallio TH, Koistinen MJ, Tulppo MP, Jordaens L, Wood N, Peng CK, Goldberger AL &
Huikuri HV (1999) HR dynamics before spontaneous onset of ventricular fibrillation. Am J
Cardiol 83:836-839.
Mäkikallio TH, Ristimäe T, Airaksinen KEJ, Peng CK, Goldberger AL & Huikuri HV (1998)
Heart rate dynamics in patients with stable angina pectoris and utility of fractal and complexity
measures. Am J Cardiol 81: 27-31.
Mäkikallio TH, Seppänen T, Niemelä M, Airaksinen KEJ, Tulppo M & Huikuri HV (1996)
Abnormalities in beat to beat complexity of heart rate dynamics in patients with a previous
myocardial infarction. J Am Coll Cardiol 28: 1005-1011.
Molgaard H, Hermansen K & Bjerregaard P (1994) Spectral components of short-term RR interval
variability in healthy subjects and effects of risk factors. Eur Heart J 15: 1174-1183.
Molgaard H, Mickley H, Pless P, Bjerregaard P & Moller M (1993) Effects of metoprolol on heart
rate variability in survivors of acute myocardial infarction. Am J Cardiol 71: 1357-1359.
Molgaard H, Sorensen KE & Bjerregaard P (1991) Circadian variation and influence of risk
factors on heart rate variability in healthy subjects. Am J Cardiol 68: 777-784.
Montano N, Cogliati C, Porta A, Pagani M, Malliani A, Narkiewicz K, Abbound M, Birkett C and
Somers VK (1998) Central vagotonic effects of atropine modulate oscillations of sympathetic
nerve activity. Circulation 98, 1349-1399.
Montano N, Ruscone TG, Porta A, Lombardi F, Pagani M & Malliani A (1994) Power spectrum
analysis of heart rate variability to assess the changes in sympathovagal balance during graded
orthostatic tilt. Circulation 90: 1826-1831.
72
Myers GA, Martin GJ, Magid NM, Barnett PS, Schaad JW, Weiss JS, Lesch M & Singer DH
(1986) Power spectral analysis of heart rate variability in sudden cardiac death: comparison to
other methods. IEEE Trans Biomed Eng 33: 1149-1156.
Niemelä MJ, Airaksinen KEJ & Huikuri HV (1994) Effect of beta-blockade on heart rate
variability in patients with coronary artery disease. J Am Coll Cardiol 23: 1370-1377.
Nolan J, Flapan AD, Capewell S, MacDonald TM, Neilson JM & Ewing DJ (1992) Decreased
cardiac parasympathetic activity in chronic heart failure and its relation to left ventricular
function. Br Heart J 67: 482-485.
O'Brien IA, O'Hare P & Corrall RJ (1986) Heart rate variability in healthy subjects: effect of age
and the derivation of normal ranges for tests of autonomic function. Br Heart J 55: 348-354.
Opie LH, The Heart. Physiology, from Cell to Circulation. Lippincott-Raven Publishers,
Philadelphia, USA 1998.
Osterziel KJ, Dietz R, Schmid W, Mikulaschek K, Manthey J & Kubler W (1990) ACE inhibition
improves vagal reactivity in patients with heart failure. Am Heart J 120: 1120-1129.
Öri Z, Monir G, Weiss J, Sayhouni X & Singer DH (1992) Heart rate variability. Frequency
domain analysis. Cardiol Clin 10: 499-537.
Pagani M, Lombardi F, Guzzetti S, Rimoldi O, Furlan R, Pizzinelli P, Sandrone G, Malfatto G,
Dell'Orto S, Piccaluga E & et al. (1986) Power spectral analysis of heart rate and arterial
pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circ
Res 59: 178-193.
Pagani M, Mazzuero G, Ferrari A, Liberati D, Cerutti S, Vaitl D, Tavazzi L & Malliani A (1991)
Sympathovagal interaction during mental stress. A study using spectral analysis of heart rate
variability in healthy control subjects and patients with a prior myocardial infarction.
Circulation 83: II43-51.
Pagani M, Montano N, Porta A, Malliani A, Abboud FM, Birkett C & Somers VK (1997)
Relationship between spectral components of cardiovascular variabilities and direct measures
of muscle sympathetic nerve activity in humans. Circulation 95: 1441-1448.
Parati G, Frattola A, di Rienzo M, Castiglioni P, Pedotti A & Mancia G (1995) Effects of aging on
24-hour dynamic baroreceptor control of heart rate in ambulant subjects. Am J Physiol 268:
H1606-H1612.
Penaz J (1978) Mayer waves: history and methodology. Automedica 2: 135-141.
Peng CK, Havlin S, Stanley HE & Goldberger AL (1995) Quantification of scaling exponents and
crossover phenomena in nonstationary heartbeat time series. Chaos 5: 82-87.
Peterson HR, Rothschild M, Weinberg CR, Fell RD, McLeish KR & Pfeifer MA (1988) Body fat
and the activity of the autonomic nervous system. N Engl J Med 318: 1077-1083.
Petretta M, Marciano F, Bianchi V, Migaux ML, Valva G, De Luca N, Salemme L, Berardino S &
Bonaduce D (1995) Power spectral analysis of heart period variability in hypertensive patients
with left ventricular hypertrophy. Am J Hypertens 8: 1206-1213.
Pincus SM & Goldberger AL (1994) Physiological time-series analysis: what does regularity
quantify? Am J Physiol 266: H1643-56.
Pincus SM & Viscarello RR (1992) Approximate entropy: a regularity measure for fetal heart rate
analysis. Obstet Gynecol 79: 249-255.
Pomeranz B, Macaulay RJB, Caudill MA, Kutz I, Adam D, Gordon KM, Kilborn KM, Barger AC,
Shannon DC, Cohen RJ & Benson H (1985) Assessment of autonomic function in humans with
heart rate spectral analysis. Am J Physiol 248: H151-H153.
Prineas R, Crow R and Blackburn H, The Minnesota code manual of electrocardiographic
findings: Standards and procedures for measurement and classification. John Wright/PSG,
Littleton 1982.
Pyörälä K, Savolainen E, Kaukola S & Haapakoski J (1985) Plasma insulin as coronary heart
disease risk factor: relationship to other risk factors and predictive value during 9 1/2-year
follow-up of the Helsinki Policemen Study population. Acta Med Scand Suppl 701: 38-52.
Radaelli A, Bernardi L, Valle F, Leuzzi S, Salvucci F, Pedrotti L, Marchesi E, Finardi G & Sleight
P (1994) Cardiovascular autonomic modulation in essential hypertension. Effect of tilting.
Hypertension 24: 556-563.
73
Räiha IJ, Piha SJ, Seppänen A, Puukka P & Sourander LB (1994) Predictive value of continuous
ambulatory electrocardiographic monitoring in elderly people. BMJ 309: 1263-1267.
Reaven GM (1988) Banting lecture 1988. Role of insulin resistance in human disease. Diabetes
37: 1595-1607.
Reaven GM, Lithell H & Landsberg L (1996) Hypertension and associated metabolic
abnormalities--the role of insulin resistance and the sympathoadrenal system. N Engl J Med
334: 374-381.
Rich MW, Saini JS, Kleiger RE, Carney RM, teVelde A & Freedland KE (1988) Correlation of
heart rate variability with clinical and angiographic variables and late mortality after coronary
angiography. Am J Cardiol 62: 714-717.
Robinson BF, Epstein SE, Beiser GD & Braunwald E (1966) Control of heart rate by the
autonomic nervous system. Circ Res 19: 400-411.
Rönnemaa T, Laakso M, Pyörälä K, Kallio V & Puukka P (1991) High fasting plasma insulin is an
indicator of coronary heart disease in non-insulin-dependent diabetic patients and nondiabetic
subjects. Arterioscler Thromb 11: 80-90.
Ryan SM, Goldberger AL, Pincus SM, Mietus J & Lipsitz LA (1994) Gender- and age-related
differences in heart rate dynamics: are women more complex than men? J Am Coll Cardiol 24:
1700-1707.
Sahn DJ, DeMaria A, Kisslo J & Weyman A (1978) Recommendations regarding quantitation in
M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation
58: 1072-1083.
Sands KE, Appel ML, Lilly LS, Schoen FJ, Mudge GHJ & Cohen RJ (1989) Power spectrum
analysis of heart rate variability in human cardiac transplant recipients. Circulation 79: 76-82.
Saul JP, Albrecht P, Berger RD and Cohen RJ, Analysis of long term heart rate variability:
methods, 1/f scaling and implications. In Computers in cardiology 1988 pp. 419-422, Silver
Spring, 1987.
Saul JP, Arai Y, Berger RD, Lilly LS, Colucci WS & Cohen RJ (1988) Assessment of autonomic
regulation in chronic congestive heart failure by heart rate spectral analysis. Am J Cardiol 61:
1292-1299.
Sayers BM (1973) Analysis of heart rate variability. Ergonomics 16: 17-32.
Seals DR & Chase PB (1989) Influence of physical training on heart rate variability and baroreflex
circulatory control. J Appl Physiol 66: 1886-1895.
Shannon DC, Carley DW & Benson H (1987) Aging of modulation of heart rate. Am J Physiol
253: H874-7.
Sleight P (1979) Reflex control of the heart. Am J Cardiol 44: 889-894.
Smyth HS, Sleight P & Pickering GW (1969) Reflex regulation of arterial pressure during sleep in
man. Circ Res 24: 109-121.
Stein PK, Kleiger RE & Rottman JN (1997) Differing effects of age on heart rate variability in
men and women. Am J Cardiol 80: 302-305.
Stern MP, Morales PA, Haffner SM & Valdez RA (1992) Hyperdynamic circulation and the
insulin resistance syndrome ("syndrome X"). Hypertension 20: 802-808.
Task Force of the European Society of Cardiology and the North American Society of Pacing and
Electrophysiology (1996) Heart rate variability: standards of measurement, physiological
interpretation and clinical use. Circulation 93: 1043-1065.
Taylor JA, Carr DL, Myers CW & Eckberg DL (1998) Mechanisms underlying very-lowfrequency RR-interval oscillations in humans. Circulation 98: 547-555.
Töyry J, Mäntysaari M, Hartikainen J & Länsimies E (1995) Day-to-day variability of cardiac
autonomic regulation parameters in normal subjects. Clin Physiol 15:39-46.
Troisi RJ, Weiss ST, Parker DR, Sparrow D, Young JB & Landsberg L (1991) Relation of obesity
and diet to sympathetic nervous system activity. Hypertension 17: 669-677.
Troy BL, Pombo J & Rackley CE (1972) Measurement of left ventricular wall thickness and mass
by echocardiography. Circulation 45: 602-611.
Tsuji H, Larson MG, Venditti FJ, Jr., Manders ES, Evans JC, Feldman CL & Levy D (1996a)
Impact of reduced heart rate variability on risk for cardiac events. The Framingham Heart
Study. Circulation 94: 2850-2855.
74
Tsuji H, Venditti FJ, Jr., Manders ES, Evans JC, Larson MG, Feldman CL & Levy D (1996b)
Determinants of heart rate variability. J Am Coll Cardiol 28: 1539-1546.
Tsuji H, Venditti FJ, Jr., Manders ES, Evans JC, Larson MG, Feldman CL & Levy D (1994)
Reduced heart rate variability and mortality risk in an elderly cohort. The Framingham Heart
Study. Circulation 90: 878-883.
Tulppo MP, Mäkikallio TH, Takala TES, Seppänen T & Huikuri HV (1996) Quantitative beat-tobeat analysis of heart rate dynamics during exercise. Am J Physiol 271: H244-H252.
van de Borne P, Montano N, Pagani M, Oren R & Somers VK (1997) Absence of low-frequency
variability of sympathetic nerve activity in severe heart failure. Circulation 95: 1449-1454.
Van Hoogenhuyze D, Weinstein N, Martin GJ, Weiss JS, Schaad JW, Sahyouni XN, Fintel D,
Remme WJ & Singer DH (1991) Reproducibility and relation to mean heart rate of heart rate
variability in normal subjects and in patients with congestive heart failure secondary to
coronary artery disease. Am J Cardiol 68: 1668-1676.
Vikman S, Ylimäyry S, Niemelä K, Reinikainen P & Huikuri HV (1998) RR-interval dynamics
before paroxysmal atrial fibrillation. Eur Heart J (Abstr.Suppl): 663 (Abstract).
Vollenweider P, Randin D, Tappy L, Jequier E, Nicod P & Scherrer U (1994) Impaired insulininduced sympathetic neural activation and vasodilation in skeletal muscle in obese humans. J
Clin Invest 93: 2365-2371.
Vollenweider L, Tappy L, Owlya R, Jequier E, Nicod P & Scherrer U (1995) Insulin-induced
sympathetic activation and vasodilation in skeletal muscle. Effects of insulin resistance in lean
subjects. Diabetes 44: 641-645.
Vybiral T, Bryg J, Maddens M, Bhasin SS, Cronin S, Boden WE & Lehman MH (1990) Effects of
transdermal scopolamine on heart rate variability in normal subjects. Am J Cardiol 65: 604608.
Vybiral T, Bryg RJ, Maddens ME & Boden WE (1989) Effect of passive tilt on sympathetic and
parasympathetic components of heart rate variability in normal subjects. Am J Cardiol 63:
1117-1120.
Wolf MM, Varigos GA, Hunt D & Sloman JG (1978) Sinus arrhythmia in acute myocardial
infarction. Med J Aust 2:52-53.
Woo MA, Stevenson WG, Moser DK & Middlekauff HR (1994) Coplex heart rate variability and
serum norepinephrine levels in patients with advanced heart failure. J Am Coll Cardiol 23:
565-569.
World Health Organization: Hypertension and Coronary Heart Disease: Classification and Criteria
for Epidemiological Studies. World Health Organization, Geneva 1959.
Yamamoto Y, Hughson RL & Peterson JC (1991) Autonomic control of heart rate during exercise
studied by heart rate variability spectral analysis. J Appl Physiol 71: 1136-1142.
Ylitalo A, Airaksinen KEJ, Tahvanainen KUO, Kuusela TA, Ikäheimo MJ, Rantala A, Lilja M &
Huikuri HV (1997) Baroreflex sensitivity in drug-treated systemic hypertension. Am J Cardiol
80: 1369-1372.
Zanchetti A & Mancia G (1991) Cardiovascular reflexes and hypertension. Hypertension
18(SupplII): III-13-III-21.
Zipes DP (1990) Influence of myocardial ischemia and infarction on autonomic innervation of
heart. Circulation 82: 1095-1105.
Zuanetti G, Latini R, Neilson JM, Schwartz PJ & Ewing DJ (1991) Heart rate variability in
patients with ventricular arrhythmias: effect of antiarrhythmic drugs. Antiarrhythmic Drug
Evaluation Group (ADEG). J Am Coll Cardiol 17: 604-612.