Guidelines heart rate_variability_ft_1996[1]

European Heart Journal (1996) 17, 354–381

Guidelines

Heart rate variability
Standards of measurement, physiological interpretation, and
clinical use
Task Force of The European Society of Cardiology and The North American
Society of Pacing and Electrophysiology (Membership of the Task Force listed in
the Appendix)

Introduction of Pacing and Electrophysiology to constitute a Task
Force charged with the responsibility of developing
The last two decades have witnessed the recognition of a appropriate standards. The specific goals of this Task
significant relationship between the autonomic nervous Force were to: standardize nomenclature and develop
system and cardiovascular mortality, including sudden definitions of terms; specify standard methods of
cardiac death[1–4]. Experimental evidence for an associ- measurement; define physiological and pathophysio-
ation between a propensity for lethal arrhythmias and logical correlates; describe currently appropriate clinical
signs of either increased sympathetic or reduced vagal applications, and identify areas for future research.
activity has encouraged the development of quantitative In order to achieve these goals, the members of
markers of autonomic activity. the Task Force were drawn from the fields of mathemat-
Heart rate variability (HRV) represents one of ics, engineering, physiology, and clinical medicine. The
the most promising such markers. The apparently easy standards and proposals offered in this text should
derivation of this measure has popularized its use. As not limit further development but, rather, should
many commercial devices now provide automated allow appropriate comparisons, promote circumspect
measurement of HRV, the cardiologist has been pro- interpretations, and lead to further progress in the field.
vided with a seemingly simple tool for both research and The phenomenon that is the focus of this report
clinical studies[5]. However, the significance and meaning is the oscillation in the interval between consecutive
of the many different measures of HRV are more heart beats as well as the oscillations between consecu-
complex than generally appreciated and there is a tive instantaneous heart rates. ‘Heart Rate Variability’
potential for incorrect conclusions and for excessive or has become the conventionally accepted term to describe
unfounded extrapolations. variations of both instantaneous heart rate and RR
Recognition of these problems led the European intervals. In order to describe oscillation in consecutive
Society of Cardiology and the North American Society cardiac cycles, other terms have been used in the litera-
ture, for example cycle length variability, heart period
Key Words: Heart rate, electrocardiography, computers, variability, RR variability and RR interval tachogram,
autonomic nervous system, risk factors. and they more appropriately emphasize the fact that it is
the interval between consecutive beats that is being
The Task Force was established by the Board of the European
Society of Cardiology and co-sponsored by the North American
analysed rather than the heart rate per se. However,
Society of Pacing and Electrophysiology. It was organised jointly these terms have not gained as wide acceptance as HRV,
by the Working Groups on Arrhythmias and on Computers of thus we will use the term HRV in this document.
Cardiology of the European Society of Cardiology. After ex-
changes of written views on the subject, the main meeting of a
writing core of the Task Force took place on May 8–10, 1994, on
Necker Island. Following external reviews, the text of this report Background
was approved by the Board of the European Society of Cardiology
on August 19, 1995, and by the Board of the North American The clinical relevance of heart rate variability was first
Society of Pacing and Electrophysiology on October 3, 1995. appreciated in 1965 when Hon and Lee[6] noted that fetal
Published simultaneously in Circulation. distress was preceded by alterations in interbeat intervals
before any appreciable change occurred in the heart rate
Correspondence: Marek Malik, PhD, MD, Chairman, Writing
Committee of the Task Force, Department of Cardiological
itself. Twenty years ago, Sayers and others focused
Sciences, St. George’s Hospital Medical School, Cranmer Terrace, attention on the existence of physiological rhythms
London SW17 0RE, U.K. imbedded in the beat-to-beat heart rate signal[7–10].

0195-668X/96/030354+28 $18.00/0 1996 American Heart Association Inc.; European Society of Cardiology

Standards of heart rate variability 355

During the 1970s, Ewing et al.[11] devised a number of The simplest variable to calculate is the standard
simple bedside tests of short-term RR differences to deviation of the NN interval (SDNN), i.e. the square root
detect autonomic neuropathy in diabetic patients. The of variance. Since variance is mathematically equal to
association of higher risk of post-infarction mortality total power of spectral analysis, SDNN reflects all the
with reduced HRV was first shown by Wolf et al. in cyclic components responsible for variability in the
1977[12]. In 1981, Akselrod et al. introduced power period of recording. In many studies, SDNN is calcu-
spectral analysis of heart rate fluctuations to quantita- lated over a 24-h period and thus encompasses both
tively evaluate beat-to-beat cardiovascular control[13]. short-term high frequency variations, as well as the
These frequency–domain analyses contributed to lowest frequency components seen in a 24-h period. As
the understanding of the autonomic background of RR the period of monitoring decreases, SDNN estimates
interval fluctuations in the heart rate record[14,15]. The shorter and shorter cycle lengths. It should also be noted
clinical importance of HRV became apparent in the late that the total variance of HRV increases with the length
1980s when it was confirmed that HRV was a strong and of analysed recording[19]. Thus, on arbitrarily selected
independent predictor of mortality following an acute ECGs, SDNN is not a well defined statistical quantity
myocardial infarction[16–18]. With the availability of new, because of its dependence on the length of recording
digital, high frequency, 24-h multi-channel electro- period. Thus, in practice, it is inappropriate to compare
cardiographic recorders, HRV has the potential to SDNN measures obtained from recordings of different
provide additional valuable insight into physiological durations. However, durations of the recordings used to
and pathological conditions and to enhance risk determine SDNN values (and similarly other HRV
stratification. measures) should be standardized. As discussed further
in this document, short-term 5-min recordings and
nominal 24 h long-term recordings seem to be
Measurement of heart rate variability appropriate options.
Other commonly used statistical variables calcu-
Time domain methods lated from segments of the total monitoring period
include SDANN, the standard deviation of the average
Variations in heart rate may be evaluated by a number NN interval calculated over short periods, usually 5 min,
of methods. Perhaps the simplest to perform are the time which is an estimate of the changes in heart rate due to
domain measures. With these methods either the heart cycles longer than 5 min, and the SDNN index, the mean
rate at any point in time or the intervals between of the 5-min standard deviation of the NN interval
successive normal complexes are determined. In a con- calculated over 24 h, which measures the variability due
tinuous electrocardiographic (ECG) record, each QRS to cycles shorter than 5 min.
complex is detected, and the so-called normal-to-normal The most commonly used measures derived from
(NN) intervals (that is all intervals between adjacent interval differences include RMSSD, the square root of
QRS complexes resulting from sinus node depolariza- the mean squared differences of successive NN intervals,
tions), or the instantaneous heart rate is determined. NN50, the number of interval differences of successive
Simple time–domain variables that can be calculated NN intervals greater than 50 ms, and pNN50 the pro-
include the mean NN interval, the mean heart rate, the portion derived by dividing NN50 by the total number
difference between the longest and shortest NN interval, of NN intervals. All these measurements of short-term
the difference between night and day heart rate, etc. variation estimate high frequency variations in heart
Other time–domain measurements that can be used are rate and thus are highly correlated (Fig. 1).
variations in instantaneous heart rate secondary to
respiration, tilt, Valsalva manoeuvre, or secondary to Geometrical methods
phenylephrine infusion. These differences can be de- The series of NN intervals can also be converted into a
scribed as either differences in heart rate or cycle length. geometric pattern, such as the sample density distribu-
tion of NN interval durations, sample density distribu-
Statistical methods tion of differences between adjacent NN intervals,
From a series of instantaneous heart rates or cycle Lorenz plot of NN or RR intervals, etc., and a simple
intervals, particularly those recorded over longer formula is used which judges the variability based on the
periods, traditionally 24 h, more complex statistical geometric and/or graphic properties of the resulting
time-domain measures can be calculated. These may be pattern. Three general approaches are used in geometric
divided into two classes, (a) those derived from direct methods: (a) a basic measurement of the geometric
measurements of the NN intervals or instantaneous pattern (e.g. the width of the distribution histogram at
heart rate, and (b) those derived from the differences the specified level) is converted into the measure of
between NN intervals. These variables may be derived HRV, (b) the geometric pattern is interpolated by a
from analysis of the total electrocardiographic recording mathematically defined shape (e.g. approximation of the
or may be calculated using smaller segments of the distribution histogram by a triangle, or approximation
recording period. The latter method allows comparison of the differential histogram by an exponential curve)
of HRV to be made during varying activities, e.g. rest, and then the parameters of this mathematical shape are
sleep, etc. used, and (c) the geometric shape is classified into several

Eur Heart J, Vol. 17, March 1996

356 Task Force

100
(a)

10

1
pNN50 (%)

0.1

0.01

0.001
0 20 40 60 80 100 120
RMSSD (ms)

100 000
(b)

10 000
NN50 (counts/24 h)

1000

100

10

1
0.001 0.01 0.1 1 10 100
pNN50 (%)
Figure 1 Relationship between the RMSSD and pNN50 (a), and pNN50 and
NN50 (b) measures of HRV assessed from 857 nominal 24-h Holter tapes recorded
in survivors of acute myocardial infarction prior to hospital discharge. The NN50
measure used in panel (b) was normalized in respect to the length of the recording
(Data of St. George’s Post-infarction Research Survey Programme.)

pattern-based categories which represent different NN intervals) divided by the maximum of the density
classes of HRV (e.g. elliptic, linear and triangular shapes distribution. Using a measurement of NN intervals on a
of Lorenz plots). Most geometric methods require the discrete scale, the measure is approximated by the value:
RR (or NN) interval sequence to be measured on or
(total number of NN intervals)/
converted to a discrete scale which is not too fine or too
(number of NN intervals in the modal bin)
coarse and which permits the construction of smoothed
histograms. Most experience has been obtained with which is dependent on the length of the bin, i.e. on the
bins approximately 8 ms long (precisely 7·8125 ms= precision of the discrete scale of measurement. Thus, if
1/128 s) which corresponds to the precision of current the discrete approximation of the measure is used with
commercial equipment. NN interval measurement on a scale different to the
The HRV triangular index measurement is the most frequent sampling of 128 Hz, the size of the bins
integral of the density distribution (i.e. the number of all should be quoted. The triangular interpolation of NN



Number of normal
RR intervals

Y

Sample
density
distribution
D

N X M

Duration of normal RR intervals
Figure 2 To perform geometrical measures on the NN interval
histogram, the sample density distribution D is constructed which
assigns the number of equally long NN intervals to each value of
their lengths. The most frequent NN interval length X is established,
that is Y=D(X) is the maximum of the sample density distribution
D. The HRV triangular index is the value obtained by dividing the
area integral of D by the maximum Y. When constructing the
distribution D with a discrete scale on the horizontal axis, the value
is obtained according to the formula
HRV index=(total number of all NN intervals)/Y.
For the computation of the TINN measure, the values N and M
are established on the time axis and a multilinear function q
constructed such that q(t)=0 for tcN and tdM and q(X)=Y, and
such that the integral
#0 (D(t)
+`
q(t))2dt
is the minimum among all selections of all values N and M. The
TINN measure is expressed in ms and given by the formula
TINN=M N.

interval histogram (TINN) is the baseline width of the HRV), and RMSSD (estimate of short-term compo-
distribution measured as a base of a triangle, approxi- nents of HRV). Two estimates of the overall HRV are
mating the NN interval distribution (the minimum recommended because the HRV triangular index
square difference is used to find such a triangle). Details permits only casual pre-processing of the ECG signal.
of computing the HRV triangular index and TINN are The RMSSD method is preferred to pNN50 and NN50
shown in Fig. 2. Both these measures express overall because it has better statistical properties.
HRV measured over 24 h and are more influenced by The methods expressing overall HRV and its
the lower than by the higher frequencies[17]. Other long- and short-term components cannot replace each
geometric methods are still in the phase of exploration other. The method selected should correspond to the
and explanation. aim of each study. Methods that might be recommended
The major advantage of geometric methods lies for clinical practices are summarized in the Section
in their relative insensitivity to the analytical quality of entitled Clinical use of heart rate variability.
the series of NN intervals[20]. The major disadvantage is Distinction should be made between measures
the need for a reasonable number of NN intervals to derived from direct measurements of NN intervals or
construct the geometric pattern. In practice, recordings instantaneous heart rate, and from the differences
of at least 20 min (but preferably 24 h) should be used between NN intervals.
to ensure the correct performance of the geometric It is inappropriate to compare time–domain
methods, i.e. the current geometric methods are in- measures, especially those expressing overall HRV,
appropriate to assess short-term changes in HRV. obtained from recordings of different durations.
Other practical recommendations are listed in
Summary and recommendations the Section on Recording requirements together with
The variety of time–domain measures of HRV is sum- suggestions related to the frequency analysis of HRV.
marized in Table 1. Since many of the measures correlate
closely with others, the following four are recommended Frequency domain methods
for time–domain HRV assessment: SDNN (estimate of
overall HRV); HRV triangular index (estimate of overall Various spectral methods[23] for the analysis of the
HRV); SDANN (estimate of long-term components of tachogram have been applied since the late 1960s. Power


358 Task Force

Table 1 Selected time-domain measures of HRV

Variable Units Description
Statistical measures

SDNN ms Standard deviation of all NN intervals.
SDANN ms Standard deviation of the averages of NN intervals in all 5 min segments of the entire recording.
RMSSD ms The square root of the mean of the sum of the squares of differences between adjacent NN
intervals.
SDNN index ms Mean of the standard deviations of all NN intervals for all 5 min segments of the entire recording.
SDSD ms Standard deviation of differences between adjacent NN intervals.
NN50 count Number of pairs of adjacent NN intervals differing by more than 50 ms in the entire recording.
Three variants are possible counting all such NN intervals pairs or only pairs in which the first or
the second interval is longer.
pNN50 % NN50 count divided by the total number of all NN intervals.

Geometric measures

HRV triangular index Total number of all NN intervals divided by the height of the histogram of all NN intervals
measured on a discrete scale with bins of 7·8125 ms (1/128 s). (Details in Fig. 2)
TINN ms Baseline width of the minimum square difference triangular interpolation of the highest peak of the
histogram of all NN intervals (Details in Fig. 2.)
Differential index ms Difference between the widths of the histogram of differences between adjacent NN intervals
measured at selected heights (e.g. at the levels of 1000 and 10 000 samples)[21].
Logarithmic index Coefficient of the negative exponential curve k · e t which is the best approximation of the
histogram of absolute differences between adjacent NN intervals[22].

spectral density (PSD) analysis provides the basic infor- the heart period[15,24,25]. The physiological explanation
mation of how power (i.e. variance) distributes as a of the VLF component is much less defined and the
function of frequency. Independent of the method existence of a specific physiological process attributable
employed, only an estimate of the true PSD of the to these heart period changes might even be questioned.
signals can be obtained by proper mathematical The non-harmonic component which does not have
algorithms. coherent properties and which is affected by algorithms
Methods for the calculation of PSD may be of baseline or trend removal is commonly accepted as a
generally classified as non-parametric and parametric. In major constituent of VLF. Thus VLF assessed from
most instances, both methods provide comparable short-term recordings (e.g. c5 min) is a dubious meas-
results. The advantages of the non-parametric methods ure and should be avoided when interpreting the PSD of
are: (a) the simplicity of the algorithm employed (Fast short-term ECGs.
Fourier Transform — FFT — in most of the cases) and Measurement of VLF, LF and HF power com-
(b) the high processing speed, whilst the advantages of ponents is usually made in absolute values of power
parametric methods are: (a) smoother spectral compo- (ms2), but LF and HF may also be measured in normal-
nents which can be distinguished independently of pre- ized units (n.u.)[15,24] which represent the relative value
selected frequency bands, (b) easy post-processing of the of each power component in proportion to the total
spectrum with an automatic calculation of low and high power minus the VLF component. The representation of
frequency power components and easy identification of LF and HF in n.u. emphasizes the controlled and
the central frequency of each component, and (c) an balanced behaviour of the two branches of the auto-
accurate estimation of PSD even on a small number of nomic nervous system. Moreover, normalization tends
samples on which the signal is supposed to maintain to minimize the effect on the values of LF and HF
stationarity. The basic disadvantage of parametric components of the changes in total power (Fig. 3).
methods is the need to verify the suitability of the Nevertheless, n.u. should always be quoted with abso-
chosen model and its complexity (i.e. the order of the lute values of LF and HF power in order to describe in
model). total the distribution of power in spectral components.

Spectral components Long-term recordings Spectral analysis may also be
Short-term recordings Three main spectral components used to analyse the sequence of NN intervals in the
are distinguished in a spectrum calculated from short- entire 24-h period. The result then includes an ultra-low
term recordings of 2 to 5 min[7,10,13,15,24]: very low fre- frequency component (ULF), in addition to VLF, LF
quency (VLF), low frequency (LF), and high frequency and HF components. The slope of the 24-h spectrum can
(HF) components. The distribution of the power and the also be assessed on a log–log scale by linear fitting the
central frequency of LF and HF are not fixed but may spectral values. Table 2 lists selected frequency–domain
vary in relation to changes in autonomic modulations of measures.



Rest Tilt 100
PSD (msec × 10 /Hz)

10
10
3

LF
1
2

Power (ms )
2
LF HF
HF 0.1

0.01
0 0.5 0 0.5
Hz Hz 0.001

0.0001 ULF VLF LF HF
LF
LF
0.00001
HF 0.0001 0.001 0.01 0.1 0.4 1

0.003

0.04

0.15
HF

Figure 3 Spectral analysis (autoregressive model, order Frequency (Hz)
12) of RR interval variability in a healthy subject at rest
and during 90 head-up tilt. At rest, two major components Figure 4 Example of an estimate of power spectral
of similar power are detectable at low and high frequen- density obtained from the entire 24-h interval of a long-
cies. During tilt, the LF component becomes dominant term Holter recording. Only the LF and HF components
but, as total variance is reduced, the absolute power of LF correspond to peaks of the spectrum while the VLF and
appears unchanged compared to rest. Normalization pro- ULF can be approximated by a line in this plot with
cedure leads to predominant LF and smaller HF compo- logarithmic scales on both axes. The slope of such a line
nents, which express the alteration of spectral components is the measure of HRV.
due to tilt. The pie charts show the relative distribution
together with the absolute power of the two components than of the level of autonomic tone[28] and averages of
represented by the area. During rest, the total variance of modulations do not represent an averaged level of tone.
the spectrum was 1201 ms2, and its VLF, LF, and HF
components were 586 ms2, 310 ms2, and 302 ms2, respec- Technical requirements and recommendations
tively. Expressed in normalized units, the LF and HF were Because of the important differences in the interpreta-
48·95 n.u. and 47·78 n.u., respectively. The LF/HF ratio tion of the results, the spectral analyses of short- and
was 1·02. During tilt, the total variance was 671 ms2, and long-term electrocardiograms should always be strictly
its VLF, LF, and HF components were 265 ms2, 308 ms2,
and 95 ms2, respectively. Expressed in normalized units, distinguished, as reported in Table 2.
the LF and HF were 75·96 n.u. and 23·48 n.u., respec- The analysed ECG signal should satisfy several
tively. The LF/HF ratio was 3·34. Thus note that, for requirements in order to obtain a reliable spectral esti-
instance, the absolute power of the LF component was mation. Any departure from the following requirements
slightly decreased during tilt whilst the normalized units of may lead to unreproducible results that are difficult to
LF were substantially increased interpret.
In order to attribute individual spectral compo-
nents to well defined physiological mechanisms, such
The problem of ‘stationarity’ is frequently dis- mechanisms modulating the heart rate should not
cussed with long-term recordings. If mechanisms change during the recording. Transient physiological
responsible for heart period modulations of a certain phenomena may perhaps be analysed by specific meth-
frequency remain unchanged during the whole period of ods which currently constitute a challenging research
recording, the corresponding frequency component of topic, but which are not yet ready to be used in applied
HRV may be used as a measure of these modulations. If research. To check the stability of the signal in terms of
the modulations are not stable, interpretation of the certain spectral components, traditional statistical tests
results of frequency analysis is less well defined. In may be employed[29].
particular, physiological mechanisms of heart period The sampling rate has to be properly chosen. A
modulations responsible for LF and HF power compo- low sampling rate may produce a jitter in the estimation
nents cannot be considered stationary during the 24-h of the R wave fiducial point which alters the spectrum
period[25]. Thus, spectral analysis performed in the entire considerably. The optimal range is 250–500 Mz or per-
24-h period as well as spectral results obtained from haps even higher[30], while a lower sampling rate (in any
shorter segments (e.g. 5 min) averaged over the entire case d100 Hz) may behave satisfactorily only if an
24-h period (the LF and HF results of these two algorithm of interpolation (e.g. parabolic) is used to
computations are not different[26,27]) provide averages of refine the R wave fiducial point[31,32].
the modulations attributable to the LF and HF compo- Baseline and trend removal (if used) may affect
nents (Fig. 4). Such averages obscure detailed informa- the lower components in the spectrum. It is advisable to
tion about autonomic modulation of RR intervals check the frequency response of the filter or the behav-
available in shorter recordings[25]. It should be remem- iour of the regression algorithm and to verify that the
bered that the components of HRV provide measure- spectral components of interest are not significantly
ments of the degree of autonomic modulations rather affected.


360 Task Force

Table 2 Selected frequency domain measures of HRV

Variable Units Description Frequency range
Analysis of short-term recordings (5 min)

5 min total power ms2 The variance of NN intervals over the approximately c0·4 Hz
temporal segment
VLF ms2 Power in very low frequency range c0·04 Hz
LF ms2 Power in low frequency range 0·04–0·15 Hz
LF norm n.u. LF power in normalised units
LF/(Total Power–VLF) 100
HF ms2 Power in high frequency range 0·15–0·4 Hz
HF norm n.u. HF power in normalised units
HF/(Total Power–VLF) 100
LF/HF Ratio LF [ms2]/HF [ms2]

Analysis of entire 24 h

Total power ms2 Variance of all NN intervals approximately c0·4 Hz
ULF ms2 Power in the ultra low frequency range c0·003 Hz
VLF ms2 Power in the very low frequency range 0·003–0·04 Hz
LF ms2 Power in the low frequency range 0·04–0·15 Hz
HF ms2 Power in the high frequency range 0·15–0·4 Hz
Slope of the linear interpolation of the approximately c0·04 Hz
spectrum in a log-log scale

The choice of QRS fiducial point may be critical. 5a,b) or by interpolating the DES, thus obtaining a
It is necessary to use a well tested algorithm (i.e. continuous signal as a function of time, or by calculating
derivative+threshold, template, correlation method, the spectrum of the counts — unitary pulses as a func-
etc.) in order to locate a stable and noise-independent tion of time corresponding to each recognised QRS
reference point[33]. A fiducial point localized far within complex[35]. Such a choice may have implications on the
the QRS complex may also be influenced by varying morphology, the measurement units of the spectra and
ventricular conduction disturbances. the measurement of the relevant spectral parameters. In
Ectopic beats, arrhythmic events, missing data order to standardize the methods, the use of the RR
and noise effects may alter the estimation of the PSD of interval tachogram with the parametric method, or the
HRV. Proper interpolation (or linear regression or simi- regularly sampled interpolation of DES with the non-
lar algorithms) on preceding/successive beats on the parametric method may be suggested; nevertheless,
HRV signals or on its autocorrelation function may regularly sampled interpolation of DES is also suitable
reduce this error. Preferentially, short-term recordings for parametric methods. The sampling frequency of
which are free of ectopy, missing data, and noise should interpolation of DES has to be sufficiently high that the
be used. In some circumstances, however, acceptance of Nyquist frequency of the spectrum is not within the
only ectopic-free short-term recordings may introduce frequency range of interest.
significant selection bias. In such cases, proper inter- Standards for non-parametric methods (based
polation should be used and the possibility of the results upon the FFT algorithm) should include the values
being influenced by ectopy should be considered[34]. The reported in Table 2, the formula of DES interpolation,
relative number and relative duration of RR intervals the frequency of sampling the DES interpolation, the
which were omitted and interpolated should also be number of samples used for the spectrum calculation,
quoted. and the spectral window employed (Hann, Hamming,
and triangular windows are most frequently used)[36].
Algorithmic standards and recommendations The method of calculating the power in respect of the
The series of data subjected to spectral analysis can be window should also be quoted. In addition to require-
obtained in different ways. A useful pictorial represen- ments described in other parts of this document, each
tation of the data is the discrete event series (DES), that study employing the non-parametric spectral analysis of
is the plot of Ri-Ri 1 interval vs time (indicated at Ri HRV should quote all these parameters.
occurrence) which is an irregularly time-sampled signal. Standards for parametric methods should
Nevertheless, spectral analysis of the sequence of include the values reported in Table 2, the type of the
instantaneous heart rates has also been used in many model used, the number of samples, the central fre-
studies[26]. quency for each spectral component (LF and HF) and
The spectrum of the HRV signal is generally the value of the model order (numbers of parameters).
calculated either from the RR interval tachogram (RR Furthermore, statistical figures have to be calculated in
durations vs number of progressive beats — see Fig. order to test the reliability of the model. The prediction



Rest Tilt
tachogram tachogram
1.0 1.0
(a) (b)
0.9 0.9
n = 256
0.8 0.8
Mean = 564.7 ms σ2 = 723 ms2
RR (s)

0.7 0.7

0.6 0.6
n = 256
0.5 2 2 0.5
Mean = 842.5 ms σ = 1784 ms
0.4 0.4
0 100 200 0 100 200
Beat # Beat #

0.015 0.015
Frequency Power Power Frequency Power Power
(c) Hz ms2 n.u. (d) Hz ms2 n.u.
VLF 0.00 785 VLF 0.00 192
LF 0.11 479 47.95 LF 0.09 413 77.78
HF 0.24 450 45.05 HF 0.24 107 20.15

0.010 LF/HF = 1.06 0.010 LF/HF = 3.66
PSD (s2/Hz)

VLF PEWT 3 PEWT 11
OOT = 9 OOT = 15
p=9 LF p = 15

VLF
0.005 0.005
LF
HF

HF

0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Frequency (Hz) Frequency (Hz)

0.015 0.015
Frequency Power Frequency Power
(e) Hz ms2 (f) Hz ms2
VLF 0.00 266 VLF 0.00 140
VLF LF 0.10 164 LF 0.10 312
HF 0.25 214 HF 0.25 59

0.010 window = Hann 0.010 LF window = Hann
PSD (s2/Hz)

VLF
0.005 HF 0.005
LF

HF

0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Frequency (Hz) Frequency (Hz)
Figure 5 Interval tachogram of 256 consecutive RR values in a normal subject at supine rest (a) and after
head-up tilt (b). The HRV spectra are shown, calculated by parametric autoregressive modelling (c and d),
and by a FFT based non-parametric algorithm (e and f). Mean values (m), variances (s2) and the number
(n) of samples are indicated. For (c) and (d), VLF, LF and HF central frequency, power in absolute value
and power in normalized units (n.u.) are also indicated together with the order p of the chosen model and
minimal values of PEWT and OOT which satisfy the tests. In (e) and (f), the peak frequency and the power
of VLF, LF and HF were calculated by integrating the PSD in the deﬁned frequency bands. The window
type is also speciﬁed. In panels (c) to (f), the LF component is indicated by dark shaded areas and the HF
component by light shaded areas.


362 Task Force

Table 3 Approximate correspondence of time domain However, many time- and frequency-domain
and frequency domain methods applied to 24-h ECG variables measured over the entire 24-h period are
recordings strongly correlated with each other (Table 3). These
strong correlations exist because of both mathematical
Approximate frequency and physiological relationships. In addition, the physio-
Time domain variable
domain correlate logical interpretation of the spectral components calcu-
lated over 24 h is difficult, for the reasons mentioned
SDNN Total power (section entitled Long-term recordings). Thus, unless
HRV triangular index Total power special investigations are performed which use the 24-h
TINN Total power
SDANN ULF
HRV signal to extract information other than the usual
SDNN index Mean of 5 min total power frequency components (e.g. the log–log slope of spectro-
RMSSD HF gram), the results of frequency–domain analysis are
SDSD HF equivalent to those of time–domain analysis, which is
NN50 count HF easier to perform.
pNN50 HF
Differential index HF
Logarithmic index HF
Rhythm pattern analysis
As illustrated in Fig. 6[39], the time–domain and spectral
methods share limitations imposed by the irregularity of
error whiteness test (PEWT) provides information about the RR series. Clearly different profiles analysed by these
the ‘goodness’ of the fitting model[37] while the optimal techniques may give identical results. Trends of decreas-
order test (OOT) checks the suitability of the order of ing or increasing cycle length are in reality not symmetri-
the model used[38]. There are different possibilities of cal[40,41] as heart rate accelerations are usually followed
performing OOT which include final prediction error by a faster decrease. In spectral results, this tends to
and Akaike information criteria. The following opera- reduce the peak at the fundamental frequency, and to
tive criterion for choosing the order p of an autoregres- enlarge its basis. This leads to the idea of measuring
sive model might be proposed: the order shall be in the blocks of RR intervals determined by properties of the
range 8–20, fulfilling the PEWT test and complying with rhythm and investigating the relationship of such blocks
the OOT test (p]min(OOT)). without considering the internal variability.
Approaches derived from the time–domain and
Correlation and differences between time and frequency the frequency–domain have been proposed in order to
domain measures reduce these difficulties. The interval spectrum and spec-
When analysing stationary short-term recordings, more trum of counts methods lead to equivalent results (d,
experience and theoretical knowledge exists on the Fig. 6) and are well suited to investigate the relationship
physiological interpretation of the frequency–domain between HRV and the variability of other physiological
measures compared to the time–domain measures measures. The interval spectrum is well adapted to link
derived from the same recordings. RR intervals to variables defined on a beat-to-beat basis

(a)

(b)

(c)

(d)

0 500 1000
Figure 6 Example of four synthesised time series with identical means,
standard deviations, and ranges. Series (c) and (d) also have identical
autocorrelation functions and therefore identical power spectra. Reprinted
with permission[39].



(e.g.blood pressure). The spectrum of counts is prefer- At present, the non-linear methods represent
able if RR intervals are related to a continuous signal potentially promising tools for HRV assessment, but
(e.g.respiration), or to the occurrence of special events standards are lacking and the full scope of these meth-
(e.g.arrhythmia). ods cannot be assessed. Advances in technology and the
The ‘peak-valley’ procedures are based either on interpretation of the results of non-linear methods are
the detection of the summit and the nadir of oscilla- needed before these methods are ready for physiological
tions[42,43] or on the detection of trends of heart rate[44]. and clinical studies.
The detection may be limited to short-term changes[42]
but it can be extended to longer variations: second and
third order peaks and troughs[43] or stepwise increase of Stability and reproducibility of HRV
a sequence of consecutive increasing or decreasing cycles
surrounded by opposite trends[44]. The various oscilla- measurement
tions can be characterized on the basis of the heart
Multiple studies have demonstrated that short-term
rate accelerating or slowing, the wavelength and/or the
measures of HRV rapidly return to baseline after tran-
amplitude. In a majority of short- to mid-term record-
sient perturbations induced by such manipulations as
ings, the results are correlated with frequency compo-
mild exercise, administration of short acting vasodila-
nents of HRV[45]. The correlations, however, tend to
tors, transient coronary occlusion, etc. More powerful
diminish as the wavelength of the oscillations and the
stimuli, such as maximum exercise or administration of
recording duration increase. Complex demodulation
long acting drugs may result in a much more prolonged
uses the techniques of interpolation and detrending[46]
interval before return to control values.
and provides the time resolution necessary to detect
There are far fewer data on the stability of
short-term heart rate changes, as well as to describe
long-term measures of HRV obtained from 24-h ambu-
the amplitude and phase of particular frequency
latory monitoring. Nonetheless, the same amount of
components as functions of time.
data available suggest great stability of HRV measures
derived from 24-h ambulatory monitoring in both
normal subjects[52,53] and in the post-infarction[54] and
Non-linear methods ventricular arrhythmia[55] populations. There also exists
some fragmentary data to suggest that stability of HRV
Non-linear phenomena are certainly involved in the measures may persist for months and years. Because
genesis of HRV. They are determined by complex inter- 24-h indices seem to be stable and free of placebo
actions of haemodynamic, electrophysiological and effect, they may be ideal variables with which to assess
humoral variables, as well as by autonomic and central intervention therapies.
nervous regulations. It has been speculated that analysis
of HRV based on the methods of non-linear dynamics
might elicit valuable information for the physiological
interpretation of HRV and for the assessment of the risk Recording requirements
of sudden death. The parameters which have been used
to measure non-linear properties of HRV include 1/f ECG signal
scaling of Fourier spectra[47,19], H scaling exponent, and The fiducial point recognised on the ECG tracing which
Coarse Graining Spectral Analysis (CGSA)[48]. For data identifies a QRS complex may be based on the maxi-
representation, Poincarè sections, low-dimension attrac- mum or baricentrum of the complex, on the determina-
tor plots, singular value decomposition, and attractor tion of the maximum of an interpolating curve, or found
trajectories have been used. For other quantitative by matching with a template or other event markers. In
descriptions, the D2 correlation dimension, Lyapunov order to localize the fiducial point, voluntary standards
exponents, and Kolmogorov entropy have been for diagnostic ECG equipment are satisfactory in
employed[49]. terms of signal/noise ratio, common mode rejection,
Although in principle these techniques have been bandwidth, etc.[56] An upper-band frequency cut-off
shown to be powerful tools for characterization of substantially lower than that established for diagnostic
various complex systems, no major breakthrough has equipment (200 Hz) may create a jitter in the recogni-
yet been achieved by their application to bio-medical tion of the QRS complex fiducial point, introducing
data including HRV analysis. It is possible that integral an error of measured RR intervals. Similarly, limited
complexity measures are not adequate to analyse bio- sampling rate induces an error in the HRV spectrum
logical systems and thus, are too insensitive to detect the which increases with frequency, thus affecting more high
non-linear perturbations of RR interval which would be frequency components[31]. An interpolation of the
of physiological or practical importance. More encour- undersampled ECG signal may decrease this error. With
aging results have been obtained using differential, proper interpolation, even a 100 Hz sampling rate can be
rather than integral complexity measures, e.g. the scaling sufficient[32].
index method[50,51]. However, no systematic study has When using solid-state storage recorders, data
been conducted to investigate large patient populations compression techniques have to be carefully considered
using these methods. in terms of both the effective sampling rate and the


364 Task Force

quality of reconstruction methods which may yield Editing of the RR interval sequence
amplitude and phase distortion[57]. The errors imposed by the imprecision of the NN
interval sequence are known to affect substantially the
Duration and circumstances of ECG recording results of statistical time–domain and all frequency–
In studies researching HRV, the duration of recording is domain methods. It is known that casual editing of the
dictated by the nature of each investigation. Standardiz- RR interval data is sufficient for the approximate assess-
ation is needed, particularly in studies investigating the ment of total HRV by the geometric methods, but it is
physiological and clinical potential of HRV. not known how precise the editing should be to ensure
Frequency–domain methods should be preferred correct results from other methods. Thus when using the
to the time–domain methods when investigating short- statistical time–domain and/or frequency–domain meth-
term recordings. The recording should last for at least 10 ods, the manual editing of the RR data should be
times the wavelength of the lower frequency bound of performed to a very high standard ensuring correct
the investigated component, and, in order to ensure identification and classification of every QRS complex.
the stability of the signal, should not be substantially Automatic ‘filters’ which exclude some intervals from
extended. Thus, recording of approximately 1 min is the original RR sequence (e.g. those differing by more
needed to assess the HF components of HRV while than 20% from the previous interval) should not replace
approximately 2 min are needed to address the LF manual editing as they are known to behave unsatisfac-
component. In order to standardize different studies torily and to have undesirable effects leading potentially
investigating short-term HRV, 5 min recordings of a to errors[58].
stationary system are preferred unless the nature of the
study dictates another design. Suggestions for standardisation of commercial
Averaging of spectral components obtained from equipment
sequential periods of time is able to minimize the error Standard measurement of HRV Commercial equipment
imposed by the analysis of very short segments. Never- designed to analyse short-term HRV should incorporate
theless, if the nature and degree of physiological heart non-parametric and preferably also parametric spectral
period modulations changes from one short segment of analysis. In order to minimize the possible confusion
the recording to another, the physiological interpreta- imposed by reporting the components of the cardiac
tion of such averaged spectral components suffers from beat-based analysis in time–frequency components, the
the same intrinsic problems as that of the spectral analysis based on regular sampling of the tachograms
analysis of long-term recordings and warrants further should be offered in all cases. Non-parametric spectral
elucidation. A display of stacked series of sequential analysis should employ at least 512 but preferably 1024
power spectra (e.g. over 20 min) may help confirm points for 5 min recordings.
steady state conditions for a given physiological state. Equipment designed to analyse HRV in long-
Although the time–domain methods, especially term recordings should implement time–domain
the SDNN and RMSSD methods, can be used to methods including all four standard measures (SDNN,
investigate recordings of short durations, the frequency SDANN, RMSSD, and HRV triangular index). In
methods are usually able to provide more easily inter- addition to other options, the frequency analysis should
pretable results in terms of physiological regulations. In be performed in 5 min segments (using the same preci-
general, the time–domain methods are ideal for the sion as with the analysis of short-term ECGs). When
analysis of long-term recordings (the lower stability of performing the spectral analysis of the total nominal
heart rate modulations during long-term recordings 24-h record in order to compute the whole range of HF,
makes the results of frequency methods less easily inter- LF, VLF and ULF components, the analysis should
pretable). The experience shows that a substantial part be performed with a similar precision of periodogram
of the long-term HRV value is contributed by the sampling, as suggested for the short-term analysis, e.g.
day–night differences. Thus the long-term recording using 218 points.
analysed by the time–domain methods should contain at The strategy of obtaining the data for the HRV
least 18 h of analysable ECG data that includes the analysis should copy the design outlined in Fig. 7.
whole night.
Little is known about the effects of the environ- Precision and testing of commercial equipment In order
ment (e.g. type and nature of physical activity and to ensure the quality of different equipment involved
of emotional circumstances) during long-term ECG in HRV analysis and to find an appropriate balance
recordings. For some experimental designs, environ- between the precision essential to research and clinical
mental variables should be controlled and in each study, studies and the cost of the equipment required, indepen-
the character of the environment should always be dent testing of all equipment is needed. As the potential
described. The design of investigations should also errors of the HRV assessment include inaccuracies in the
ensure that the recording environment of individual identification of fiducial points of QRS complexes, the
subjects is similar. In physiological studies comparing testing should include all the recording, replay, and
HRV in different well-defined groups, the differences analysis phases. Thus, it seems ideal to test various
between underlying heart rate should also be properly equipment with signals (e.g. computer simulated) of
acknowledged. known HRV properties rather than with existing



RR intervals according to a certain prematurity thresh-
RR interval NN data Interpolation
rejection sequence + sampling old) should not be relied on when ensuring the quality of
the RR interval sequence.
RR data
editing
Physiological correlates of heart rate
Artefact variability
identification
TIME FREQUENCY Physiological correlates of HRV components
Microcomputer DOMAIN DOMAIN
digitising
HRV HRV
Autonomic influences of heart rate
Although cardiac automaticity is intrinsic to various
ECG
pacemaker tissues, heart rate and rhythm are largely
recording under the control of the autonomic nervous system[59].
The parasympathetic influence on heart rate is mediated
Figure 7 Flow chart summarizing individual steps used via release of acetylcholine by the vagus nerve.
when recording and processing the ECG signal in order to
Muscarinic acetylcholine receptors respond to this
obtain data for HRV analysis.
release mostly by an increase in cell membrane K+
conductance[60–62]. Acetylcholine also inhibits the
hyperpolarization-activated ‘pacemaker’ current If[63,64].
databases of already digitized ECGs. When employing The ‘Ik decay’ hypothesis[65] proposes that pacemaker
commercial equipment in studies investigating physi- depolarization results from slow deactivation of the
ological and clinical aspects of HRV, independent tests delayed rectifier current, Ik, which, due to a time-
of the equipment used should always be required. A independent background inward current, causes diasto-
possible strategy for testing of commercial equipment is lic depolarization[65,66]. Conversely, the ‘If activation’
proposed in Appendix B. Voluntary industrial standards hypothesis[67] suggest that following action potential
should be developed adopting this or similar strategy. termination, If provides a slowly activating inward cur-
rent predominating over decaying Ik, thus initiating
slow diastolic depolarization.
Summary and recommendations
The sympathetic influence on heart rate is medi-
In order to minimize the errors caused by improperly
ated by release of epinephrine and norepinephrine.
designed or incorrectly used techniques, the following
Activation of -adrenergic receptors results in cyclic
points are recommended:
AMP mediated phosphorilation of membrane proteins
The ECG equipment used should satisfy the
and increases in ICaL[68] and in If[69,70]. The end result is
current voluntary industrial standards in terms of signal/
an acceleration of the slow diastolic depolarization.
noise ratio, common mode rejection, bandwidth, etc.
Under resting conditions, vagal tone prevails[71]
Solid-state recorders used should allow signal
and variations in heart period are largely dependent on
reconstruction without amplitude and phase distortion;
vagal modulation[72]. The vagal and sympathetic activity
long-term ECG recorders using analogue magnetic
constantly interact. As the sinus node is rich in acetyl-
media should accompany the signal with phase-locked
cholinesterase, the effect of any vagal impulse is brief
time tracking.
because the acetylcholine is rapidly hydrolysed. Para-
Commercial equipment used to assess HRV
sympathetic influences exceed sympathetic effects
should satisfy the technical requirements listed in the
probably via two independent mechanisms: a cholin-
Section on Standard Measurement of HRV and its
ergically induced reduction of norepinephrine released
performance should be independently tested.
in response to sympathetic activity, and a cholinergic
In order to standardize physiological and clinical
attenuation of the response to a adrenergic stimulus.
studies, two types of recordings should be used whenever
possible: (a) short-term recordings of 5 min made
under physiologically stable conditions processed by Components of HRV
frequency–domain methods, and/or (b) nominal 24-h The RR interval variations present during resting con-
recordings processed by time–domain methods. ditions represent a fine tuning of the beat-to-beat control
When using long-term ECGs in clinical studies, mechanisms[73,74]. Vagal afferent stimulation leads to
individual, subjects should be recorded under fairly reflex excitation of vagal efferent activity and inhibition
similar conditions and in a fairly similar environment. of sympathetic efferent activity[75]. The opposite reflex
When using statistical time–domain or effects are mediated by the stimulation of sympathetic
frequency–domain methods, the complete signal should afferent activity[76]. Efferent vagal activity also appears
be carefully edited using visual checks and manual to be under ‘tonic’ restraint by cardiac afferent sympa-
corrections of individual RR intervals and QRS complex thetic activity[77]. Efferent sympathetic and vagal activi-
classifications. Automatic ‘filters’ based on hypotheses ties directed to the sinus node are characterized by
on the logic of RR interval sequence (e.g. exclusion of discharge largely synchronous with each cardiac cycle


366 Task Force

which can be modulated by central (e.g. vasomotor and coronary artery or common carotid arteries in conscious
respiratory centres) and peripheral (e.g. oscillation in dogs[24,79]. Conversely, an increase in HF is induced by
arterial pressure and respiratory movements) oscilla- controlled respiration, cold stimulation of the face and
tors[24]. These oscillators generate rhythmic fluctuations rotational stimuli[24,78].
in efferent neural discharge which manifest as short and
long-term oscillation in the heart period. Analysis of Summary and recommendations for interpretation of
these rhythms may permit inferences on the state and HRV components
function of (a) the central oscillators, (b) the sympa- Vagal activity is the major contributor to the HF
thetic and vagal efferent activity, (c) humoral factors, component.
and (d) the sinus node. Disagreement exists in respect of the LF compo-
An understanding of the modulatory effects of nent. Some studies suggest that LF, when expressed in
neural mechanisms on the sinus node has been enhanced normalized units, is a quantitative marker for sympa-
by spectral analysis of HRV. The efferent vagal activity thetic modulations, other studies view LF as reflecting
is a major contributor to the HF component, as seen in both sympathetic and vagal activity. Consequently, the
clinical and experimental observations of autonomic LF/HF ratio is considered by some investigators to
manoeuvres such as electrical vagal stimulation, mus- mirror sympatho/vagal balance or to reflect sympathetic
carinic receptor blockade, and vagotomy[13,14,24]. More modulations.
controversial is the interpretation of the LF component Physiological interpretation of lower frequency
which is considered by some[24,78–80] as a marker of components of HRV (that is of the VLF and ULF
sympathetic modulation (especially when expressing it in components) warrants further elucidation.
normalized units) and by others[13,81] as a parameter that It is important to note that HRV measures
includes both sympathetic and vagal influences. This fluctuations in autonomic inputs to the heart rather than
discrepancy is due to the fact that in some conditions, the mean level of autonomic inputs. Thus both auto-
associated with sympathetic excitation, a decrease in the nomic withdrawal and a saturatingly high level of
absolute power of the LF component is observed. It is sympathetic input leads to diminished HRV[28].
important to recall that during sympathetic activation
the resulting tachycardia is usually accompanied by a
marked reduction in total power, whereas the reverse
occurs during vagal activation. When the spectral com- Changes of HRV related to specific
ponents are expressed in absolute units (ms2), the pathologies
changes in total power influence LF and HF in the same
direction and prevent the appreciation of the fractional A reduction of HRV has been reported in several
distribution of the energy. This explains why in supine cardiological and non-cardiological diseases[24,78,81,83].
subjects under controlled respiration atropine reduces
both LF and HF[14] and why during exercise LF is Myocardial infarction
markedly reduced[24]. This concept is exemplified in Depressed HRV after MI may reflect a decrease in vagal
Fig. 3 showing the spectral analysis of HRV in a normal activity directed to the heart which leads to prevalence
subject during control supine conditions and 90 of sympathetic mechanisms and to cardiac electrical
head-up tilt. Due to the reduction in total power, instability. In the acute phase of MI, the reduction in
LF appears as unchanged if considered in absolute 24-h SDNN is significantly related to left ventricular
units. However, after normalization an increase in LF dysfunction, peak creatine kinase, and Killip class[84].
becomes evident. Similar results apply to the LF/HF The mechanism by which HRV is transiently
ratio[82]. reduced after MI and by which a depressed HRV is
Spectral analysis of 24-h recordings[24–25] shows predictive of the neural response to acute MI is not yet
that in normal subjects LF and HF expressed in normal- defined, but it is likely to involve derangements in the
ized units exhibit a circadian pattern and reciprocal neural activity of cardiac origin. One hypothesis[85]
fluctuations, with higher values of LF in the daytime and involves cardio-cardiac sympatho-sympathetic[86,87] and
of HF at night. These patterns become undetectable sympatho-vagal reflexes[75] and suggests that the changes
when a single spectrum of the entire 24-h period is used in the geometry of a beating heart due to necrotic and
or when spectra of subsequent shorter segments are non-contracting segments may abnormally increase the
averaged. In long-term recordings, the HF and LF firing of sympathetic afferent fibres by mechanical dis-
components account for approximately 5% of total tortion of the sensory ending[76,87,88]. This sympathetic
power. Although the ULF and VLF components excitation attenuates the activity of vagal fibres directed
account for the remaining 95% of total power, their to the sinus node. Another explanation, especially
physiological correlates are still unknown. applicable to marked reduction of HRV, is the
LF and HF can increase under different condi- reduced responsiveness of sinus nodal cells to neural
tions. An increased LF (expressed in normalized units) is modulations[82,85].
observed during 90 tilt, standing, mental stress and Spectral analysis of HRV in patients surviving an
moderate exercise in healthy subjects, and during mod- acute MI revealed a reduction in total and in the
erate hypotension, physical activity and occlusion of a individual power of spectral components[89]. However,



when the power of LF and HF was calculated in such as faster heart rates and high levels of circulating
normalized units, an increased LF and a diminished HF cathecolamines, a relationship between changes in HRV
were observed during both resting controlled conditions and the extent of left ventricular dysfunction was con-
and 24-h recordings analysed over multiple 5 min troversially reported[102,104]. In fact, whereas the reduc-
periods[90,91]. These changes may indicate a shift of tion in time domain measures of HRV seemed to parallel
sympatho-vagal balance towards sympathetic predomi- the severity of the disease, the relationship between
nance and reduced vagal tone. Similar conclusions were spectral components and indices of ventricular dysfunc-
obtained by considering the changes in the LF/HF ratio. tion appears to be more complex. In particular, in most
The presence of an alteration in neural control mecha- patients with a very advanced phase of the disease and
nisms was also reflected by the blunting of the day–night with a drastic reduction in HRV, a LF component could
variations of the RR interval[91] and LF and HF spectral not be detected despite clinical signs of sympathetic
components [91,92] present in a period ranging from days activation. Thus, in conditions characterized by marked
to a few weeks after the acute event. In post MI patients and unopposed persistent sympathetic excitation, the
with a very depressed HRV, most of the residual energy sinus node seems to drastically diminish its responsive-
is distributed in the VLF frequency range below 0·03 Hz, ness to neural inputs[104].
with only a small respiration-related HF[93]. These char-
acteristics of the spectral profile are similar to those Tetraplegia
observed in advanced cardiac failure or after cardiac Patients with chronic complete high cervical spinal cord
transplant, and are likely to reflect either diminished lesions have intact efferent vagal and sympathetic neural
responsiveness of the target organ to neural modulatory pathways directed to the sinus node. However, spinal
inputs[82] or a saturating influence on the sinus node of a sympathetic neurons are deprived of modulatory control
persistently high sympathetic tone[28]. and in particular of baroreflex supraspinal inhibitory
inputs. For this reason, these patients represent a unique
Diabetic neuropathy clinical model with which to evaluate the contribution of
In neuropathy associated with diabetes mellitus, charac- supraspinal mechanisms in determining the sympathetic
terized by alteration of small nerve fibres, a reduction in activity responsible for low frequency oscillations of
time–domain parameters of HRV seems not only to HRV. It has been reported[107] that no LF could be
carry negative prognostic value but also to precede the detected in tetraplegic patients, thus suggesting the criti-
clinical expression of autonomic neuropathy[94–97]. In cal role of supraspinal mechanisms in determining the
diabetic patients without evidence of autonomic neuro- 0·1 Hz rhythm. Two recent studies, however, have indi-
pathy, reduction of the absolute power of LF and HF cated that an LF component can also be detected in
during controlled conditions was also reported[96]. How- HRV and arterial pressure variabilities of some tetra-
ever, when the LF/HF ratio was considered or when LF plegic patients[108,109]. While Koh et al.[108] attributed the
and HF were analysed in normalized units, no signifi- LF component of HRV to vagal modulations, Guzzetti
cant difference in comparison to normals was present. et al.[109] attributed the same component to sympathetic
Thus, the initial manifestation of this neuropathy is activity because of the delay with which the LF com-
likely to involve both efferent limbs of the autonomic ponent appeared after spinal section, suggesting an
nervous system[96,98]. emerging spinal rhythmicity capable of modulating
sympathetic discharge.
Cardiac transplantation
A very reduced HRV with no definite spectral compo-
nents was reported in patients with a recent heart Modifications of HRV by specific
transplant[97,99,100]. The appearance of discrete spectral interventions
components in a few patients is considered to reflect
cardiac re-innervation[101]. This re-innervation may The rationale for trying to modify HRV after MI stems
occur as early as 1 to 2 years post transplantation from multiple observations indicating that cardiac
and is usually of sympathetic origin. Indeed, the corre- mortality is higher among post-MI patients who have a
lation between the respiratory rate and the HF compo- more depressed HRV[93,110]. The inference is that inter-
nent of HRV observed in some transplanted patients ventions that augment HRV may be protective against
indicates that a non-neural mechanism may also contrib- cardiac mortality and sudden cardiac death. Although
ute to generate respiration-related rhythmic oscilla- the rationale for changing HRV is sound, it contains
tion[100]. The initial observation of identifying patients also the inherent danger of leading to the unwarranted
developing an allograft rejection according to changes in assumption that modification of HRV translates directly
HRV could be of clinical interest but needs further into cardiac protection, which may not be the case[111].
confirmation. The target is the improvement of cardiac electrical
stability, and HRV is just a marker of autonomic
Myocardial dysfunction activity. Despite the growing consensus that increases in
A reduced HRV has been consistently observed in vagal activity can be beneficial[112], it is not yet known
patients with cardiac failure[24,78,81,102–106]. In this condi- how much vagal activity (or its markers) has to increase
tion characterized by signs of sympathetic activation, in order to provide adequate protection.


368 Task Force

Beta-adrenergic blockade and HRV dogs, documented to be at high risk by the previous
The data on the effect of -blockers on HRV in post- occurrence of ventricular fibrillation during acute myo-
MI patients are surprisingly scanty[113,114]. Despite cardial ischaemia, were randomly assigned to 6 weeks of
the observation of statistically significant increases, the either daily exercise training or cage rest followed by
actual changes are very modest. However, it is of note exercise training[129]. After training, HRV (SDNN) in-
that -blockade prevents the rise in the LF component creased by 74% and all animals survived a new ischaemic
observed in the morning hours[114]. In conscious post-MI test. Exercise training can also accelerate recovery of the
dogs, -blockers do not modify HRV[115]. The un- physiological sympatho-vagal interaction, as shown in
expected observation that, prior to MI, -blockade post-MI patients[130].
increases HRV only in the animals destined to be at low
risk for lethal arrhythmias post-MI[115] may suggest
novel approaches to post-MI risk stratification. Clinical use of heart rate variability
Antiarrhythmic drugs and HRV Although HRV has been the subject of numerous clini-
Data exist for several antiarrhythmic drugs. Flecainide cal studies investigating a wide spectrum of cardiological
and propafenone, but not amiodarone, were reported to and non-cardiological diseases and clinical conditions, a
decrease time–domain measures of HRV in patients with general consensus of the practical use of HRV in adult
chronic ventricular arrhythmia[116]. In another study[117] medicine has been reached only in two clinical scenarios.
propafenone reduced HRV and decreased LF much Depressed HRV can be used as a predictor of risk after
more than HF, resulting in a significantly smaller acute MI and as an early warning sign of diabetic
LF/HF ratio. A larger study[118] confirmed that flecain- neuropathy.
ide, and also encainide and moricizine, decreased HRV
in post-MI patients but found no correlation between
the change in HRV and mortality during follow-up. Assessment of risk after acute myocardial
Thus, some antiarrhythmic drugs associated with infarction
increased mortality can reduce HRV. However, it is not
known whether these changes in HRV have any direct The observation[12] that in patients with an acute MI the
prognostic significance. absence of respiratory sinus arrhythmias is associated
with an increase in ‘in-hospital’ mortality represents the
Scopolamine and HRV first of a large number of reports[16,93,131] which have
Low dose muscarinic receptor blockers, such as atropine demonstrated the prognostic value of assessing HRV to
and scopolamine, may produce a paradoxical increase in identify high risk patients.
vagal efferent activity, as suggested by a decrease in Depressed HRV is a powerful predictor of mor-
heart rate. Different studies examined the effects of tality and of arrhythmic complications (e.g. sympto-
transdermal scopolamine on indices of vagal activity in matic sustained ventricular tachycardia) in patients
patients with a recent MI[119–122] and with congestive following acute MI[16,131] (Fig. 8). The predictive value
heart failure[123]. Scopolamine markedly increases HRV, of HRV is independent of other factors established for
which indicates that pharmacological modulation of post-infarction risk stratification, such as depressed
neural activity with scopolamine may effectively increase left ventricular ejection fraction, increased ventricular
vagal activity. However, efficacy during long-term treat- ectopic activity, and presence of late potentials. For
ment has not been assessed. Furthermore, low dose prediction of all-cause mortality, the value of HRV is
scopolamine does not prevent ventricular fibrillation due similar to that of left ventricular ejection fraction, but
to acute myocardial ischaemia in post-MI dogs[124]. HRV is superior to left ventricular ejection fraction in
predicting arrhythmic events (sudden cardiac death and
Thrombolysis and HRV ventricular tachycardia)[131]. This permits speculation
The effect of thrombolysis on HRV (assessed by that HRV is a stronger predictor of arrhythmic mortal-
pNN50), was reported in 95 patients with acute MI[125]. ity rather than non-arrhythmic mortality. However,
HRV was higher 90 min after thrombolysis in the clear differences between HRV in patients suffering from
patients with patency of the infarct-related artery. How- sudden and non-sudden cardiac death after acute MI
ever, this difference was no longer evident when the have not been observed. Nevertheless, this might also be
entire 24 h were analysed. related to the nature of the presently used definition of
sudden cardiac death[132], which is bound to include not
Exercise training and HRV only patients suffering from arrhythmia-related death
Exercise training may decrease cardiovascular mortality but also fatal reinfarctions and other cardiovascular
and sudden cardiac death[126]. Regular exercise training events.
is also thought capable of modifying the autonomic The value of both conventional time–domain
balance[127,128]. A recent experimental study, designed to and frequency–domain parameters have been fully
assess the effects of exercise training on markers of vagal assessed in several independent prospective studies, but
activity, has simultaneously provided information on because of using optimized cut-off values defining nor-
changes in cardiac electrical stability[129]. Conscious mal and depressed HRV, these studies may slightly



1.0
Above 100 (a)

0.9

50–100
Survival

0.8

0.7

0.6 Below 50

0.5
0 10 20 30 40
Time after MI (years)

1.0
HRV index 20 (b)
0.95
Survival probability

0.9 HRV index 15

0.85

0.8 HRV index ≤ 20

0.75
HRV index ≤ 15
0.7

0.65
0 1 2 3 4
Follow-up (years)
Figure 8 Cumulative survival of patients after MI. (a) Shows
survival of patients stratified according to 24-h SDNN values in
three groups with cut-off points of 50 and 100 ms. (Reprinted with
permission[16]. (b) Shows similar survival curves of patients strati-
fied according to 24-h HRV triangular index values with cut-off
points of 15 and 20 units, respectively. (Data of St. George’s
Post-infarction Research Survey Programme.)

overestimate the predictive value of HRV. Nevertheless, tality or is merely a marker of poor prognosis. The data
the confidence intervals of such cut-off values are rather suggest that depressed HRV is not a simple reflection of
narrow because of the sizes of the investigated popula- the sympathetic overdrive and/or vagal withdrawal due
tions. Thus, the observed cut-off values of 24-h measures to poor ventricular performance but that it also reflects
of HRV, e.g. SDNN 50 ms and HRV triangular index depressed vagal activity which has a strong association
15 for highly depressed HRV, or SDNN 100 ms and with the pathogenesis of ventricular arrhythmias and
HRV triangular index 20 for moderately depressed sudden cardiac death[112].
HRV are likely to be broadly applicable.
It is not known whether different indices of HRV Assessment of HRV for risk stratification after acute
(e.g. assessments of the short- and long-term compo- myocardial infarction
nents) can be combined in a multivariate fashion in Traditionally, HRV used for risk stratification after MI
order to improve post-infarction risk stratification. has been assessed from 24-h recordings. HRV measured
There is a general consensus, however, that combination from short-term electrocardiogram recordings also
of other measures with the assessment of overall 24-h provides prognostic information for risk stratification
HRV is probably redundant. following MI but whether it is as powerful as that from
24-h recordings is uncertain[133–135]. HRV measured
Pathophysiological considerations from short-term recordings is depressed in patients at
It has not been established whether depressed HRV is high risk; the predictive value of depressed HRV in-
part of the mechanism of increased post-infarction mor- creases with increasing the length of recording. Thus, the


370 Task Force

70 Heart rate variability is decreased early after
(a)
acute MI and begins to recover within a few weeks; it is
60
maximally but not fully recovered by 6 to 12 months
50
after MI[91,137]. Assessment of heart rate variability at
40 both the early stage of MI (2 to 3 days after acute MI)[84]
Positive predictive accuracy (%)

30 and pre-discharge from hospital (1 to 3 weeks after acute
20 MI) offers important prognostic information. Heart rate
10 variability measured late (1 year) after acute MI also
0 predicts further mortality[138]. Data from animal models
20 30 40 50 60 70 80 suggest that the speed of HRV recovery after MI
correlates with subsequent risk[115].
70 (b)
60 HRV used for multivariate risk stratification
50 The predictive value of heart rate variability alone is
40 modest, but combination with other techniques substan-
30 tially improves its positive predictive accuracy over a
20 clinically important range of sensitivity (25% to 75%) for
10
cardiac mortality and arrhythmic events (Fig. 9).
Improvements in the positive predictive accuracy
0
20 30 40 50 60 70 80 over the range of sensitivities have been reported
Sensitivity (%) for combinations of HRV with mean heart rate, left
ventricular ejection fraction, frequency of ventricular
Figure 9 Comparison of positive predictive characteris- ectopic activity, parameters of high resolution electro-
tics of HRV (solid lines) and of combinations of HRV with
left ventricular ejection fraction (dashed lines) and of cardiograms (e.g. presence or absence of late potentials),
HRV with left ventricular ejection fraction and ectopic and clinical assessment[139]. However, it is not known
counts on 24-h ECGs (dotted lines) used for identification which other stratification factors are the most practical
of patients at risk of 1-year cardiac mortality (a) and and most feasible to be combined with HRV for
1-year arrhythmic events (sudden death and/or sympto- multifactorial risk stratification.
matic sustained ventricular tachycardia (b) after acute Systematic multivariate studies of post MI risk
myocardial infarction. (Data of St. George’s Post- stratification are needed before a consensus can be
infarction Research Survey Programme.) reached and before a combination of HRV with other
variables of proven prognostic importance can be rec-
ommended. Many aspects that are not relevant for
univariate risk stratification need to be examined: it is
use of nominal 24-h recordings may be recommended
not obvious whether the optimum cut-off values of
for risk stratification studies after MI. On the other
individual risk factors known from univariate studies are
hand, the assessment of HRV from short-term record-
appropriate in a multivariate setting. Different multi-
ings can be used for initial screening of survivors of
variate combinations are probably needed for optimiz-
acute MI[136]. Such an assessment has similar sensitivity
ing predictive accuracy at different ranges of sensitivity.
but lower specificity for predicting patients at high risk
Stepwise strategies should be examined to identify
compared to 24-h HRV.
optimum sequences of performing individual tests used
Spectral analysis of HRV in survivors of MI
in multivariate stratification.
suggested that the ULF and VLF components carry the
highest predictive value[93]. As the physiological correlate
of these components is unknown and as these compo- Summary and recommendations for interpreting
nents correspond to up to 95% of the total power which predictive value of depressed HRV after acute
can be easily assessed in the time-domain, the use of myocardial infarction
individual spectral components of HRV for risk stratifi- The following facts should be noted when exploiting
cation after MI is not more powerful than the use of HRV assessment in clinical studies and/or trials
those time-domain methods which assess overall HRV. involving survivors of acute myocardial infarction.
Depressed HRV is a predictor of mortality and
Development of HRV after acute myocardial infarction arrhythmic complications independent of other
The time after acute MI at which the depressed HRV recognised risk factors.
reaches the highest predictive value has not been inves- There is a general consensus that HRV should be
tigated comprehensively. Nevertheless, the general con- measured approximately 1 week after index infarction.
sensus is that HRV should be assessed shortly prior to Although HRV assessed from short-term record-
hospital discharge, i.e. approximately 1 week after index ings provides prognostic information, HRV measured in
infarction. Such a recommendation also fits well into the nominal 24-h recordings is a stronger risk predictor.
common practice of hospital management of survivors HRV assessed from short-term recordings may be used
of acute MI. for initial screening of all survivors of an acute MI.



No currently recognised HRV measure provides survivors of MI, patients with DAN are also predis-
better prognostic information than the time–domain posed to poor outcomes such as sudden death but it
HRV measures assessing overall HRV (e.g. SDNN or remains to be determined whether the HRV measures
HRV triangular index). Other measures, e.g. ULF of confer prognostic information among diabetics.
entire 24-h spectral analysis, perform equally well. A
high risk group may be selected by the dichotomy limits Frequency domain measures
of SDNN 50 ms or HRV triangular index 15. The following abnormalities in frequency HRV analy-
For clinically meaningful ranges of sensitivity, sis are associated with DAN: (a) reduced power in
the predictive value of HRV alone is modest, although it all spectral bands which is the most common find-
is higher than that of any other so far recognised risk ing[96,146–148], (b) failure to increase LF on standing,
factor. To improve the predictive value, HRV may be which is a reflection of impaired sympathetic response or
combined with other factors. However, optimum set of depressed baroreceptor sensitivity[96,147]; (c) abnormally
risk factors and corresponding dichotomy limits have reduced total power with unchanged LF/HF ratio[96],
not yet been established. and (d) a leftward shift in the LF central frequency,
the physiological meaning of which needs further
elucidation[147].
Assessment of diabetic neuropathy In advanced neuropathic states, the resting
supine power spectrum often reveals extremely low
As a complication of diabetes mellitus, autonomic amplitudes of all spectral components making it difficult
neuropathy is characterized by early and widespread to separate signal from noise[96,146,147]. It is therefore
neuronal degeneration of small nerve fibres of both recommended that an intervention such as standing or
sympathetic and parasympathetic tracts[140]. Its clinical tilt be included. Another method to overcome the low
manifestations are ubiquitous with functional impair- signal to noise ratio is to introduce a coherence function
ment and include postural hypotension, persistent tachy- which utilizes the total power coherent with one or the
cardia, gustatory sweating, gastroparesis, bladder atony other frequency band[146].
and nocturnal diarrhoea. Once clinical manifestations of
diabetic autonomic neuropathy (DAN) supervene, the
estimated 5-year mortality is approximately 50%[141]. Other clinical potential
Thus, early subclinical detection of autonomic dysfunc-
tion is important for risk stratification and subsequent Selected studies investigating HRV in other cardio-
management. Analyses of short-term and/or long-term logical diseases are listed in Table 4.
HRV have proven useful in detecting DAN[96,142–147].
For the patient presenting with a real or suspect
DAN there are three HRV methods from which to Future possibilities
choose: (a) simple bedside RR interval methods, (b)
long-term time–domain measures, which are more Development of HRV measurement
sensitive and more reproducible than the short-term
tests, and (c) frequency–domain analysis performed The currently available time–domain methods predomi-
under short-term steady state conditions and which are nantly used to assess the long-term profile of HRV are
useful in separating sympathetic from parasympathetic probably sufficient for this purpose. Improvements are
abnormalities. possible, especially in terms of numerical robustness.
The contemporary non-parametric and parametric
Long-term time domain measures spectral methods are probably sufficient to analyse
HRV computed from 24-h Holter records are more short-term ECGs without transient changes of heart
sensitive than simple bedside tests (e.g. Valsava period modulations.
manoeuvre, orthostatic test, and deep breathing[11]) for Apart from the need to develop numerically
detecting DAN. Most experience has been obtained with robust techniques suitable for fully automatic measure-
the NN50[144] and SDSD (see Table 1)[145] methods. ment (the geometrical methods are only one possibility
Using the NN50 count, where the lower 95% confidence in this direction), the following three areas deserve
interval for total counts range from 500 to 2000 depend- attention.
ing on the age, about half of diabetic patients will
demonstrate abnormally low counts per 24 h. Moreover, Dynamics and transient changes of HRV
there is a strong correlation between the percentage The present possibilities of characterizing the quantify-
of patients with abnormal counts and the extent of ing the dynamics of the RR interval sequence and
autonomic neuropathy determined from conventional transient changes of HRV are sparse and still under
measures. mathematical development. However, it may be
Besides their increased sensitivity, these 24-h assumed that proper assessment of HRV dynamics will
time domain methods are strongly correlated with other lead to substantial improvements in our understanding
established HRV measurements and have been found of both the modulations of heart period and their
to be reproducible and stable over time. Similar to physiological and pathophysiological correlates.


372 Task Force

Table 4 Summary of selected studies investigating the clinical value of HRV in cardiological diseases other than
myocardial infarction

Author Population Investigation
Disease state Clinical finding Potential value
of study (no. of patients) parameter

Hypertension Guzzetti 49 hypertensive Spectral AR LF found in Hypertension is
1991[149] 30 normals hypertensives as compared characterized by a
to normals with blunting depressed circadian
of circadian patters rhythmicity of LF
Langewitz 41 borderline Spectral FFT Reduced parasympathetic Support the use of
1994[150] hypertensive in hypertensive patients non-pathological therapy of
34 hypertensive hypertension that vagal
54 normals tone (e.g. exercise)

Congestive heart Saul 25 chronic CHF Spectral ? spectral power all In CHF, there is ? vagal, but
failure 1988[155] NYHA III, IV Blackman–Turkey frequencies, especially relatively preserved
21 normals 15 min 0·04 Hz in CHF patients sympathetic modulation of
acquisition HR
Casolo 20 CHF Time–domain RR Low HRV Reduced vagal activity in
1989[102] NYHA II, III, IV interval histogram CHF patients
20 normals with 24 h-Holter
Binkley 10 dilated Spectral FFT ? m high frequency power Withdrawal of
1991[152] cardiomyopathy 4 min supine (0·1 Hz) in CHF parasympathetic tone
(EF 14 to 40%) acquisition LF/HF observed in CHF
10 normals CHF has imbalance of
autonomic tone with ?
parasympathetic and a
predominance of
sympathetic tone
Kienzle 23 CHF Spectral FFT Alterations of HRV not
1992[104] NYHA II, III, IV Time-domain tightly linked to severity of
24–48 h-Holter CHF
? HRV was related to
sympathetic excitation
Townend 12 CHF Time domain HRV during ACE
1992[153] NYHA III, IV 24 h-Holter inhibitor treatment
Binkley 13 CHF Spectral FFT 12 weeks of ACE inhibitor A significant augmentation
1993[154] NYHA II, III 4 min supine treatment high of parasympathetic tone
acquisition frequency HRV was associated with ACE
inhibitor therapy
Woo 21 CHF Poincarè plots Complex plots are Poincarè plots may assist
1994[155] NYHA III Time-domain associated with analysis of sympathetic
24 h-Holter norepinephrine levels and influences
greater sympathetic
activation

Heart Alexopoulos 19 transplant Time–domain Reduced HRV in
transplantation 1988[156] 10 normals 24 h-Holter denervated donor hearts;
recipient innervated
hearts had more HRV
Sands 17 transplant Spectral FFT HRV from 0·02 to 1·0 Hz — Patients with rejection
1989[100] 6 normals 15 min supine 90% reduced documented biopsy show
acquisition significantly more variability

Chronic mitral Stein 38 chronic mitral Spectral FFT HR and measures of May be prognostic indicator
regurgitation 1993[157] regurgitation Time-domain ultralow frequency by of atrial fibrillation, mortality,
24 h-Holter SDANN correlated with and progression to valve
ventricular performance surgery
and predicted clinical events

Mitral valve Marangoni 39 female mitral Spectral AR MVP patients had ? high MVP patients had low vagal
prolapse 1993[158] valve prolapse 10 min supine frequency tone
24 female acquisition
controls
Continued



Table 4 Continued.

Author Population Investigation
Disease state Clinical finding Potential value
of study (no. of patients) parameter

Cardiomyopathies Counihan 104 HCM Spectral FFT Global and specific vagal HRV does not add to the
1993[159] Time-domain tone measurements of predictive accuracy of
24 h-Holter HRV were ? in known risk factors in HCM
symptomatic patients

Sudden death or Dougherty 16 CA survivors Spectral AR HRV as measured by low HRV is clinically useful to
cardiac arrest 1992[160] 5 CA Time-domain frequency power and risk stratify CA survivors for
nonsurvivors 24 h-Holter SDNN were significantly 1 year mortality
5 normals related to 1 year mortality

Huikuri 22 CA survivors Spectral AR ? High frequency power in
1992[161] 22 control Time-domain CA survivors — low
24 h-Holter frequency power did not
discriminate CA survivors
Circadian pattern of HRV
found in all patients
Algra 193 SD cases Time–domain ? short-term variation HRV may be used to
1993[110] 230 24 h-Holter (0·05–0·50 Hz) estimate the risk of SD
symptomatic independently increased
patients the risk of SD by a factor
2·6
? long-term variation
(0·02–0·05 Hz) increased
the risk of SD by a factor
of 2
Myers 6 normals Time and Both time and frequency HF power may be a useful
1986[162] 12 patients with frequency domain domain indices separated predictor of SD
structural heart 24 h-Holter normals from SD patients
disease (6 with ? HF power (0·35–0·5 Hz)
and 6 without was the best separator
SD) between heart disease
patients with and without SD

Martin 20 normals Time-domain SDNN index significantly Time domain indices may
1988[163] 5 patients 24 h-Holter lower in SD patients Identify increased risk of SD
experiencing SD
during Holter
monitoring

Ventricular Vybiral 24 VF Time-domain HRV indices do not
arrhythmias 1993[164] 19 CAD 24 h-Holter change consistently
before VF
Huikuri 18 VT or CA Spectral AR all power spectra of HRV A temporal relation exists
1993[165] 24 h-Holter were significantly ? before between the decrease of
the onset of sustained VT HRV and the onset of
than before nonsustained VT sustained VT
Hohnloser 14 post MI with Spectral FFT HRV of post MI-CA Baroreflex sensitivity, not
1994[166] VF or sustained Time-domain survivors do not differ HRV, distinguished post MI
14 post MI 24 h-Holter from other post MI patients with and without VF
(matched) patients they differ strikingly and VT
in term of baroreflex
sensitivity

Supraventricular Kocovic 64 SVT Spectral FFT HR, ? HRV and ? Parasympathetic ganglia
arrhythmias 1993[167] Time-domain parasympathetic and fibres may be more
acquisition components after dense in the mid and
24 h-Holter radiofrequency ablation anterior low septum

AR=autoregressive; CA=cardiac arrest; CAD=coronary artery disease; CHF=congestive heart failure; EF=ejection fraction; FFT=fast
Fourier transform; HCM=hypertrophic cardiomyopathy; HF=high frequency; HRV=heart rate variability; LF=low frequency;
NYHA=New York Heart Association classification; SD=sudden death; SVT=supraventricular tachycardia; VF=ventricular fibrillation;
VT=ventricular tachycardia.


374 Task Force

It remains to be seen whether the methods of dards for various age and gender subsets[110]. Recently,
non-linear dynamics will be appropriate for the investigators from the Framingham Heart Study
measurement of transient changes in RR intervals or reported on the time– and frequency–domain measures
whether new mathematical models and algorithmic of HRV in 736 elderly subjects, and the relationship of
concepts will be needed to tailor the principles of these HRV measures to all-cause mortality during 4
measurement more closely to the physiological nature of years of follow-up[175]. These investigators concluded
cardiac periodograms. In any case, the task of assessing that HRV offers prognostic information independent of
transient changes in HRV seems to be more important and beyond that provided by traditional risk factors.
than further refinements of the current technology used Additional population-based HRV studies involving the
to analyse stable stages of heart period modulations. full age spectrum in males and females need to be
performed.
PP and RR intervals
Little is known about the interplay between the PP Physiologic phenomena
and PR autonomic modulations. For these reasons, the It would be of interest to evaluate HRV in various
sequence of PP intervals should also be studied[168]. circadian patterns such as normal day–night cycles,
Unfortunately, precise location of a P wave fiducial sustained reversed day–night cycles (evening-night shift
point is almost impossible to achieve in surface ECGs work), and transiently altered day–night cycles, such as
recorded with the current technology. However, might occur with international travel. The autonomic
developments in the technology should allow PP interval fluctuations occurring during various stages of sleep
and PR interval variability to be investigate in future including rapid eye movement (REM) sleep have been
studies. studied in only a few subjects. In normal subjects, the
Multisignal analysis HF vagal component of the power spectrum is aug-
The modulations of heart periods are naturally not mented only during non-REM sleep, whereas in post-MI
the only manifestation of the autonomic regulatory patients with this increase in HF is absent[176].
mechanisms. Currently, commercial or semi-commercial The autonomic nervous system response to ath-
equipment exists which enables simultaneous recording letic training and rehabilitative exercise programmes
of ECG, respiration, blood pressure, etc. However, in after various disease states is thought to be a condition-
spite of the ease with which the signals can be recorded, ing phenomenon. HRV data should be useful in under-
no widely accepted method exists for comprehensive standing the chronological aspects of training and the
multisignal analysis. Each signal can be analysed separ- time to optimal conditioning as it relates to the auto-
ately, e.g. with parametric spectral methods, and the nomic influences on the heart. Also, HRV may provide
results of the analysis compared. Analysis of coupling important information about deconditioning with pro-
between physiological signals allow the properties of the longed bed rest, with weightlessness and with the zero g
coupling to be measured[169–174]. that accompanies space flight.

Pharmacological responses
Studies needed to improve physiological Many medications act directly or indirectly on the
understanding autonomic nervous system, and HRV can be used to
explore the influence of various agents on sympathetic
Efforts should be made to find the physiological corre- and parasympathetic activity. It is known that para-
lates and the biological relevance of various HRV sympathetic blockade with full dose atropine produces
measures currently employed. In some cases, e.g. the HF marked diminution of HRV. Low dose scopolamine has
component, this has been achieved. In other cases, vagotonic influences and is associated with increased
e.g. the VLF and ULF components, the physiological HRV, especially in the HR range. -adrenergic blockade
correlates are still largely unknown. was observed into increase HRV and to reduce the
This uncertain knowledge limits the interpreta- normalized units of the LF component[15]. Considerably
tion of associations between these variables and the risk more research is needed to understand the effects and
of cardiac events. The use of markers of autonomic clinical relevance of altered vagotonic and adrenergic
activity is very attractive. However, unless a tenable tone on total HRV power and its various components in
mechanistic link between these variables and cardiac health and disease.
events is found, there is an inherent danger of concen- At present, few data exist on the effects of
trating therapeutic efforts on the modification of these calcium channel blockers, sedatives, anxiolytics, analge-
markers[111,112]. This may lead to incorrect assumptions sics, anaesthetics, antiarrhythmic agents, narcotics,
and serious misinterpretations. and chemotherapeutic agents such as vincristine and
doxorubicin on HRV.
Possibilities of future clinical utility
Risk stratification
Normal standards Both time and frequency measures of HRV calculated
Large prospective population studies with longitudinal from long 24-h and short 2 to 15-min ECG recordings
follow-up are needed to establish normal HRV stan- have been used to predict time to death after MI, as well



as the risk of all-cause mortality and sudden cardiac cardiovascular and non-cardiovascular disorders. HRV
death in patients with structural heart disease[162,163,177] studies should enhance our understanding of physiologi-
and a number of other pathophysiological condi- cal phenomena, the actions of medications, and disease
tions[177]. Using diagnostic instruments that can measure mechanisms. Large prospective longitudinal studies are
HRV, together with the frequency and complexity needed to determine the sensitivity, specificity, and pre-
of ventricular arrhythmias, signal-averaged ECG, ST dictive value of HRV in the identification of individuals
segment variability, and repolarization heterogeneity, it at risk for subsequent morbid and mortal events.
should be possible to markedly improve the identifica-
tion of patients at risk for sudden cardiac death and
arrhythmic events. Prospective studies are needed to References
evaluate the sensitivity, specificity, and predictive
accuracy of combined testing. [1] Lown B, Verrier RL. Neural activity and ventricular fibrilla-
tion. N Engl J Med 1976; 294: 1165–70.
Fetal and neonatal heart rate variability is an
[2] Corr PB, Yamada KA, Witkowski FX. Mechanisms control-
important area of investigation, and it might provide ling cardiac autonomic function and their relation to arrhyth-
early information about fetal and neonatal distress mogenesis. In: Fozzard HA, Haber E, Jennings RB, Katz AN,
and identify those at risk for the sudden infant death Morgan HE, eds. The Heart and Cardiovascular System. New
syndrome. Most of the preliminary work in this field was York: Raven Press, 1986: 1343–1403.
[3] Schwartz PJ, Priori SG. Sympathetic nervous system and
carried out in the early 1980s before the more sophisti- cardiac arrhythmias. In: Zipes DP, Jalife J, eds. Cardiac
cated power spectral techniques became available. In- Electrophysiology. From Cell to Bedside. Philadelphia: W.B.
sight into autonomic maturation in the developing fetus Saunders, 1990: 330–43.
might also be possible through the proper application of [4] Levy MN, Schwartz PJ eds. Vagal control of the heart:
Experimental basis and clinical implications. Armonk: Future,
these techniques.
1994.
[5] Dreifus LS, Agarwal JB, Botvinick EH et al. (American
Disease mechanisms College of Cardiology Cardiovascular Technology Assess-
A fertile area of research is to use HRV techniques to ment Committee). Heart rate variability for risk stratification
explore the role of autonomic nervous system alterations of life-threatening arrhythmias. J Am Coll Cardiol 1993; 22:
948–50.
in disease mechanisms, especially those conditions in [6] Hon EH, Lee ST. Electronic evaluations of the fetal heart rate
which sympathovagal factors are thought to play an patterns preceding fetal death, further observations. Am J
important role. Recent work suggests that alterations in Obstet Gynec 1965; 87: 814–26.
autonomic innervation to the developing heart might [7] Sayers BM. Analysis of heart rate variability. Ergonomics
1973; 16: 17–32.
be responsible for some forms of the long QT syn-
[8] Penaz J, Roukenz J, Van der Waal HJ. Spectral analysis of
drome[178]. Fetal HRV studies in pregnant mothers with some spontaneous rhythms in the circulation. In: Drischel H,
this disorder is certainly feasible and might be very Tiedt N, eds. Leipzig: Biokybernetik, Karl Marx Univ, 1968:
informative[179]. 233–41.
The role of the autonomic nervous system in [9] Luczak H, Lauring WJ. An analysis of heart rate variability.
Ergonomics 1973; 16: 85–97.
essential hypertension is an important area of investiga- [10] Hirsh JA, Bishop B. Respiratory sinus arrhythmia in humans;
tion[180]. The question regarding the primary or second- how breathing pattern modulates heart rate. Am J Physiol
ary role of enhanced sympathetic activity in essential 1981; 241: H620–9.
hypertension might be answered by longitudinal studies [11] Ewing DJ, Martin CN, Young RJ, Clarke BF. The value of
cardiovascular autonomic function tests: 10 years experience
of subjects who are initially normotensive. Does essen-
in diabetes. Diabetic Care 1985; 8: 491–8.
tial hypertension result from augmented sympathetic [12] Wolf MM, Varigos GA, Hunt D, Sloman JG. Sinus arrhyth-
activity with altered responsiveness of neural regulatory mia in acute myocardial infarction. Med J Australia 1978; 2:
mechanisms? 52–3.
Several primary neurological disorders including [13] Akselrod S, Gordon D, Ubel FA, Shannon DC, Barger AC,
Cohen RJ. Power spectrum analysis of heart rate fluctuation:
Parkinson’s disease, multiple sclerosis, Guillain-Barre a quantitative probe of beat to beat cardiovascular control.
syndrome, and orthostatic hypotension of the Shy- Science 1981; 213: 220–2.
Drager type are associated with altered autonomic func- [14] Pomeranz M, Macaulay RJB, Caudill MA. Assessment of
tion. In some of these disorders, changes in HRV may be autonomic function in humans by heart rate spectral analysis.
Am J Physiol 1985; 248: H151–3.
an early manifestation of the condition and may be
[15] Pagani M, Lombardi F, Guzzetti S et al. Power spectral
useful in quantitating the rate of disease progression analysis of heart rate and arterial pressure variabilities as a
and/or the efficacy of therapeutic interventions. This marker of sympatho-vagal interaction in man and conscious
same approach may also be useful in the evaluation of dog. Circ Res 1986; 59: 178–93.
secondary autonomic neurologic disorders that accom- [16] Kleiger RE, Miller JP, Bigger JT, Moss AJ, and the Multi-
center Post-Infarction Research Group. Decreased heart rate
pany diabetes mellitus, alcoholism, and spinal cord variability and its association with increased mortality after
injuries. acute myocardial infarction. Am J Cardiol 1987; 59: 256–62.
[17] Malik M, Farrell T, Cripps T, Camm AJ. Heart rate variabil-
Conclusion ity in relation to prognosis after myocardial infarction: selec-
tion of optimal processing techniques. Eur Heart J 1989; 10:
Heart rate variability has considerable potential to as- 1060–74.
sess the role of autonomic nervous system fluctuations in [18] Bigger JT, Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger
normal healthy individuals and in patients with various RE, Rottman JN. Frequency domain measures of heart


376 Task Force

period variability and mortality after myocardial infarction. [41] Eckberg DL. Human sinus arrhythmia as an index of vagal
Circulation 1992; 85: 164–71. cardiac outflow. J Appl Physiol 1983; 54: 961–6.
[19] Saul JP, Albrecht P, Berger RD, Cohen RJ. Analysis of long [42] Fouad FM, Tarazi RC, Ferrario CM, Fighaly S, Alicandri C.
term heart rate variability: methods, 1/f scaling and implica- Assessment of parasympathetic control of heart rate by
tions. Computers in Cardiology 1987. IEEE Computer Society a noninvasive method. Heart Circ Physiol 1984; 15:
press, Washington 1988: 419–22. H838–42.
[20] Malik M, Xia R, Odemuyiwa O, Staunton A, Poloniecki J, [43] Schechtman VL, Kluge KA, Harper RM. Time-domain sys-
Camm AJ. Influence of the recognition artefact in the auto- tem for assessing variation in heart rate. Med Biol Eng
matic analysis of long-term electrocardiograms on time- Comput 1988; 26: 367–73.
domain measurement of heart rate variability. Med Biol Eng [44] Courmel Ph, Hermida JS, Wennerblöm B, Leenhardt A,
Comput 1993; 31: 539–44. Maison-Blanche P, Cauchemez B. Heart rate variability in
[21] Bjökander I, Held C, Forslund L et al. Heart rate variability myocardial hypertrophy and heart failure, and the effects of
in patients with stable angina pectoris. Eur Heart J 1992; 13 beta-blocking therapy. A non-spectral analysis of heart rate
(Abstr Suppl): 379. oscillations. Eur Heart J 1991; 12: 412–22.
[22] Scherer P, Ohler JP, Hirche H, Höpp H-W. Definition of a [45] Grossman P, Van Beek J, Wientjes C. A comparison of three
new beat-to-beat-parameter of heart rate variability (Abstr). quantification methods for estimation of respiratory sinus
Pacing Clin Electrophys 1993; 16: 939. arrhythmia. Psychophysiology 1990; 27: 702–14.
[23] Kay SM, Marple, SL. Spectrum analysis: A modern perspec- [46] Shin SJ, Tapp WN, Reisman SS, Natelson BH. Assessment of
tive Proc IEEE 1981; 69: 1380–1419. autonomic regulation of heart rate variability by the method
[24] Malliani A, Pagani M, Lombardi F, Cerutti S. Cardiovascular of complex demodulation. IEEE Trans Biomed Eng 1989; 36:
neural regulation explored in the frequency domain. Circula- 274–83.
tion 1991; 84: 1482–92. [47] Kobayashi M, Musha T. 1/f fluctuation of heart beat period.
[25] Furlan R, Guzetti S, Crivellaro W et al. Continuous 24-hour IEEE Trans Biomed Eng 1982; 29: 456–7.
assessment of the neural regulation of systemic arterial pres- [48] Yamamoto Y, Hughson RL. Coarse-graining spectral analy-
sure and RR variabilities in ambulant subjects. Circulation sis: new method for studying heart rate variability. J Appl
1990; 81: 537–47. Physiol 1991; 71: 1143–50.
[26] Berger RD, Akselrod S, Gordon D, Cohen RJ. An efficient [49] Babloyantz A, Destexhe A. Is the normal heart a periodic
algorithm for spectral analysis of heart rate variability. IEEE oscillator? Biol Cybern 1988; 58: 203–11.
Trans Biomed Eng 1986; 33: 900–4.
[50] Morfill GE, Demmel V, Schmidt G. Der plötzliche Herztod:
[27] Rottman JN, Steinman RC, Albrecht P, Bigger JT, Rolnitzky Neue Erkenntnisse durch die Anwendung komplexer Diagno-
LM, Fleiss JL. Efficient estimation of the heart period power severfahren. Bioscope 1994; 2: 11–19.
spectrum suitable for physiologic or pharmacologic studies.
[51] Schmidt G, Monfill GE. Nonlinear methods for heart rate
Am J Cardiol 1990; 66: 1522–4.
variability assessment. In: Malik M, Camm AJ, eds. Heart
[28] Malik M, Camm AJ. Components of heart rate
rate variability. Armonk: Futura, 1995: 87–98.
variability — What they really mean and what we really
measure. Am J Cardiol 1993; 72: 821–2. [52] Kleiger RE, Bigger JT, Bosner MS et al. Stability over time of
variables measuring heart rate variability in normal subjects.
[29] Bendat JS, Piersol AG. Measurement and analysis of random
Am J Cardiol 1991; 68: 626–30.
data. New York: Wiley, 1966.
[30] Pinna GD, Maestri R, Di Cesare A, Colombo R, Minuco G. [53] Van Hoogenhuyze DK, Weinstein N, Martin GJ et al. Repro-
The accuracy of power-spectrum analysis of heart-rate vari- ducibility and relation to mean heart rate of heart rate
ability from annotated RR list generated by Holter systems. variability in normal subjects and in patients with congestive
Physiol Meas 1994; 15: 163–79. heart failure secondary to coronary artery disease. Am J
Cardiol 1991; 68: 1668–76.
[31] Merri M, Farden DC, Mottley JG, Titlebaum EL. Sampling
frequency of the electrocardiogram for the spectral analysis of [54] Kautzner J. Reproducibility of heart rate variability measure-
heart rate variability, IEEE Trans Biomed Eng 1990; 37: ment. In: Malik M, Camm AJ, eds. Heart rate variability.
99–106. Armonk: Futura, 1995: 165–71.
[32] Bianchi AM, Mainardi LT, Petrucci E, Signorini MG, [55] Bigger JT, Fleiss JL, Rolnitzsky LM, Steinman RC. Stability
Mainardi M, Cerutii S. Time-variant power spectrum analysis over time of heart period variability in patients with previous
for the detection of transient episodes in HRV signal. IEEE myocardial infarction and ventricular arrhythmias. Am J
Trans Biomed Eng 1993; 40: 136–44. Cardiol 1992; 69: 718–23.
[33] Friesen GM, Jannett TC, Jadalloh MA, Yates SL, Quint SR, [56] Bailey JJ, Berson AS, Garson A Jr et al. Recommendations
Nogle HT. A comparison of the noise sensitivity of nine QRS for standardization and specifications in automated electro-
detection algorithms. IEEE Trans Biomed Eng 1990; 37: cardiography. Circulation 1990; 81: 730–9.
85–98. [57] Kennedy HN. Ambulatory (Holter) electrocardiography tech-
[34] Kamath MV, Fallen EL. Correction of the heart rate vari- nology. Clin Cardiol 1992; 10: 341–56.
ability signal for ectopics and missing beats. In: Malik M, [58] Malik M, Cripps T, Farrell T, Camm AJ. Prognostic value of
Camm AJ, eds. Heart rate variability. Armonk: Futura, 1995: heart rate variability after myocardial infarction—a compari-
75–85. son of different data processing methods. Med Biol Eng
[35] De Boer RW, Karemaker JM, Strackee J. Comparing spectra Comput 1989; 27: 603–11.
of a series of point events, particularly for heart-rate vari- [59] Jalife J, Michaels DC. Neural control of sinoatrial pacemaker
ability spectra. IEEE Trans Biomed Eng 1984; 31: 384–7. activity. In: Levy MN, Schwartz PJ, eds. Vagal Control of
[36] Harris FJ. On the use of windows for harmonic analysis with The Heart: Experimental Basis And Clinical Implications.
the Discrete Fourier Transform. IEEE Proc 1978; 66: 51–83. Armonk: Futura, 1994: 173–205.
[37] Box GEP, Jenkins GM. Time series analysis: Forecasting and [60] Noma A, Trautwein W. Relaxation of the ACh-induced
control. San Francisco: Holden Day, 1976. potassium current in the rabbit sinoatrial node cell Pflügers
[38] Akaike H. A new look at the statistical model identification, Arch 1978; 377: 193–200.
IEEE Trans Autom Cont 1974; 19: 716–23. [61] Osterrieder W, Noma A, Trautwein W. On the kinetics of the
[39] Kaplan DT. The analysis of variability. J Cardiovasc Electro- potassium channel activated by acetylcholine in the S-A node
physiol 1994; 5: 16–19. of the rabbit heart. Pflügers Arch 1980; 386: 101–9.
[40] Katona PG, Jih F. Respiratory sinus arrhythmia: a non [62] Sakmann B, Noma A, Trautwein W. Acetylcholine activation
invasive measure of parasympathetic cardiac control. J Appl of single muscarinic K+ channels in isolated pacemaker cells
Physiol 1975; 39: 801–5. of the mammalian heart. Nature 1983; 303: 250–3.



[63] DiFrancesco D, Tromba C. Inhibition of the hyperpolarizing- [86] Malliani A, Schwartz PJ, Zanchetti A. A sympathetic reflex
activated current lf, induced by acetycholine in rabbit elicited by experimental coronary occlusion. Am J Physiol
sino-atrial node myocytes. J Physiol (Lond) 1988; 405: 477– 1969; 217: 703–9.
91. [87] Brown AM, Malliani A. Spinal sympathetic reflexes initiated
[64] DiFrancesco D, Tromba C. Muscarinic control of the hyper- by coronary receptors. J Physiol 1971; 212: 685–705.
polarizing activated current lf in rabbit sino-atrial node [88] Malliani A, Recordati G, Schwartz PJ. Nervous activity of
myocytes. J Physiol (Lond) 1988; 405: 493–510. afferent cardiac sympathetic fibres with atrial and ventricular
[65] Irisawa H, Brown HF, Giles WR. Cardiac pacemaking in the endings. J Physiol 1973; 229: 457–69.
sinoatrial node. Physiol Rev 1993; 73: 197–227. [89] Bigger JT Jr, Fleiss JL, Rolnitzky LM, Steinman RC,
[66] Irisawa H, Giles WR. Sinus and atrioventricular node cells: Schneider WJ. Time course of recovery of heart period
Cellular electrophysiology. In: Zipes DP, Jalife J, eds. Cardiac variability after myocardial infarction. J Am Coll Cardiol
Electrophysiology: From Cell to Bedside. Philadelphia: W. B. 1991; 18: 1643–9.
Saunders, 1990: 95–102. [90] Lombardi F, Sandrone G, Pempruner S et al. Heart rate
[67] DiFrancesco D. The contribution of the pacemaker current variability as an index of sympathovagal interaction after
(lf) to generation of spontaneous activity in rabbit sino-atrial myocardial infarction. Am J Cardiol 1987; 60: 1239–45.
node myocytes. J Physiol (Lond) 1991; 434: 23–40. [91] Lombardi F, Sandrone G, Mortara A et al. Circadian
[68] Trautwein W, Kameyama M. Intracellular control of calcium variation of spectral indices of heart rate variability after
and potassium currents in cadiac cells. Jpn Heart J 1986; 27: myocardial infarction. Am Heart J 1992; 123: 1521–9.
31–50. [92] Kamath MV, Fallen EL. Diurnal variations of neurocardiac
[69] Brown HF, DiFrancesco D, Noble SJ. How does adrenaline rhythms in acute myocardial infarction. Am J Cardiol 1991;
accelerate the heart? Nature 1979; 280: 235–6. 68: 155–60.
[70] DiFrancesco D, Ferroni A, Mazzanti M, Tromba C. Proper- [93] Bigger JT Jr, Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger
ties of the hyperpolarizing-activated current (lf) in cells iso- RE, Rottman JN. Frequency domain measures of heart
lated from the rabbit sino-atrial node. J Physiol (Lond) 1986; period variability and mortality after myocardial infarction.
377: 61–88. Circulation 1992; 85: 164–71.
[71] Levy MN. Sympathetic-parasympathetic interactions in the [94] Ewing DJ, Neilson JMM, Traus P. New method for assess-
heart. Circ Res 1971; 29: 437–45. ing cardiac parasympathetic activity using 24-hour electro-
[72] Chess GF, Tam RMK, Calaresu FR. Influence of cardiac cardiograms. Br Heart J 1984; 52: 396–402.
neural inputs on rhythmic variations of heart period in the cat.
[95] Kitney RI, Byrne S, Edmonds ME, Watkins PJ, Roberts VC.
Am J Physiol 1975; 228: 775–80.
Heart rate variability in the assessment of autonomic diabetic
[73] Akselrod S, Gordon D, Madwed JB, Snidman NC, Shannon neuropathy. Automedica 1982; 4: 155–67.
DC, Cohen RJ. Hemodynamic regulation: investigation by
[96] Pagani M, Malfatto G, Pierini S et al. Spectral analysis of
spectral analysis. Am J Physiol 1985; 249: H867–75.
heart rate variability in the assessment of autonomic diabetic
[74] Saul JP, Rea RF, Eckberg DL, Berger RD, Cohen RJ. Heart
neuropathy. J Auton Nerv System 1988; 23: 143–53.
rate and muscle sympathetic nerve variability during reflex
changes of autonomic activity. Am J Physiol 1990; 258: [97] Freeman R, Saul JP, Roberts MS, Berger RD, Broadbridge
H713–21. C, Cohen RJ. Spectral analysis of heart rate in diabetic
neuropathy. Arch Neurol 1991; 48: 185–90.
[75] Schwartz PJ, Pagani M, Lombardi F, Malliani A, Brown AM.
A cardio-cardiac sympatho-vagal reflex in the cat. Circ Res [98] Bernardi L, Ricordi L, Lazzari P, et al. Impaired circulation
1973; 32: 215–20. modulation of sympathovagal modulation of sympathovagal
[76] Malliani A. Cardiovascular sympathetic afferent fibers. Rev activity in diabetes. Circulation 1992; 86: 1443–52.
Physiol Biochem Pharmacol 1982; 94: 11–74. [99] Bernardi L, Salvucci F, Suardi R et al. Evidence for an
[77] Cerati D, Schwartz PJ. Single cardiac vagal fiber activity, intrinsic mechanism regulating heart rate variability in the
acute myocardial ischemia, and risk for sudden death. Circ transplanted and the intact heart during submaximal dy-
Res 1991; 69: 1389–1401. namic exercise? Cardiovasc Res 1990; 24: 969–81.
[78] Kamath MV, Fallen EL. Power spectral analysis of heart rate [100] Sands KE, Appel ML, Lilly LS, Schoen FJ, Mudge GH Jr,
variability: a noninvasive signature of cardiac autonomic Cohen RJ. Power spectrum analysis of heart rate variability
function. Crit Revs Biomed Eng 1993; 21: 245–311. in human cardiac transplant recipients. Circulation 1989; 79:
[79] Rimoldi O, Pierini S, Ferrari A, Cerutti S, Pagani M, Malliani 76–82.
A. Analysis of short-term oscillations of R-R and arterial [101] Fallen EL, Kamath MV, Ghista DN, Fitchett D. Spectral
pressure in conscious dogs. Am J Physiol 1990; 258: H967– analysis of heart rate variability following human heart
H976. transplantation: evidence for functional reinnervation. J
[80] Montano N, Gnecchi Ruscone T, Porta A, Lombardi F, Auton Nerv Syst 1988; 23: 199–206.
Pagani M, Malliani A. Power spectrum analysis of heart rate [102] Casolo G, Balli E, Taddei T, Amuhasi J, Gori C. Decreased
variability to assess the changes in sympathovagal balance spontaneous heart rate variability on congestive heart failure.
during graded orthostatic tilt. Circulation 1994; 90: 1826–31. Am J Cardiol 1989; 64: 1162–7.
[81] Appel ML, Berger RD, Saul JP, Smith JM, Cohen RJ. Beat to [103] Nolan J, Flapan AD, Capewell S et al. Decreased cardiac
beat variability in cardiovascular variables: Noise or music? J parasympathetic activity in chronic heart failure and its rela-
Am Coll Cardiol 1989; 14: 1139–1148. tion to left ventricular function. Br Heart J 1992; 69: 761–7.
[82] Malliani A, Lombardi F, Pagani M. Power spectral analysis of [104] Kienzle MG, Ferguson DW, Birkett CL, Myers GA, Berg
heart rate variability: a tool to explore neural regulatory WJ, Mariano DJ. Clinical hemodynamic and sympathetic
mechanisms. Br Heart J 1994; 71: 1–2. neural correlates of heart rate variability in congestive heart
[83] Malik M, Camm AJ. Heart rate variability and clinical failure. Am J Cardiol 1992; 69: 482–5.
cardiology. Br Heart J 1994; 71: 3–6. [105] Mortara A, La Rovere MT, Signorini MG et al. Can power
[84] Casolo GC, Stroder P, Signorini C et al. Heart rate variability spectral analysis of heart rate variability identify a high risk
during the acute phase of myocardial infarction. Circulation subgroup of congestive heart failure patients with excessive
1992; 85: 2073–9. sympathetic activation? A pilot study before and after heart
[85] Schwartz PJ, Vanoli E, Stramba-Badiale M, De Ferrari GM, transplantation. Br Heart J 1994; 71: 422–30.
Billman GE, Foreman RD. Autonomic mechanisms and [106] Gordon D, Herrera VL, McAlpine L et al. Heart rate
sudden death. New insights from the analysis of baroreceptor spectral analysis: a noninvasive probe of cardiovascular
reflexes in conscious dogs with and without a myocardial regulation in critically ill children with heart disease. Ped
infarction. Circulation 1988; 78: 969–79. Cardiol 1988; 9: 69–77.


378 Task Force

[107] Inoue K, Miyake S, Kumashiro M, Ogata H, Yoshimura O. [126] O’Connor GT, Buring JE, Yusuf S et al. An overview of
Power spectral analysis of heart rate variability in traumatic randomized trials of rehabilitation with exercise after myo-
quadriplegic humans. Am J Physiol 1990; 258: H1722–6. cardial infarction. Circulation 1989; 80: 234–44.
[108] Koh J, Brown TE, Beightol LA, Ha CY, Eckberg DL. [127] Furlan R, Piazza D, Dell’Orto S et al. Early and late effects
Human autonomic rhythms: vagal cardiac mechanisms in of exercise and athletic training on neural mechanisms con-
tetraplegic patients. J Physiol 1994; 474: 483–95. trolling heart rate. Cardiovasc Res 1993; 27: 482–8.
[109] Guzzetti S, Cogliati C, Broggi C et al. Heart period and [128] Arai Y, Saul JP, Albrecht P, et al. Modulation of cardiac
arterial pressure variabilities in quadriplegic patients. Am J autonomic activity during and immediately after exercise.
Physiol 1994; 266: H1112–20. Am J Physiol 1989; 256: H132–41.
[110] Algra A, Tijssen JGP, Roelandt JRTC, Pool J, Lubsen J. [129] Hull SS JR, Vanoli E, Adamson PB, Verrier RL, Foreman
Heart rate variability from 24-hour electrocardiography and RD, Schwartz PJ. Exercise training confers anticipatory
the 2-year risk for sudden death. Circulation 1993; 88: 180–5. protection from sudden death during acute myocardial
[111] Schwartz PJ, De Ferrari GM. Interventions changing heart ischemia. Circulation 1994; 89: 548–52.
rate variability after acute myocardial infarction. In: Malik [130] La Rovere MT, Mortara A, Sandrone G, Lombardi F.
M, Camm, AJ, eds. Heart rate variability. Armonk: Futura, Autonomic nervous system adaptation to short-term exercise
1995: 407–20. training. Chest 1992; 101: 299–303.
[112] De Ferrari GM, Vanoli E, Schwartz PJ. Cardiac vagal [131] Odemuyiwa O, Malik M, Farrell T, Bashir Y, Poloniecki J,
activity, myocardial ischemia and sudden death. In: Zipes Camm J. Comparison of the predictive characteristics of
DP, Jalife J, eds. Cardiac electrophysiology. From cell to heart rate variability index and left ventricular ejection
bedside. Philadelphia: W. B. Saunders, 1995: 422–34. fraction for all-cause mortality, arrhythmic events and sud-
[113] Molgaard H, Mickley H, Pless P, Bjerregaard P, Moller M. den death after acute myocardial infarction. Am J Cardiol
Effects of metoprolol on heart rate variability in survivors 1991; 68: 434–9.
of acute myocardial infarction. Am J Cardiol 1993; 71: [132] Greene HL, Richardson DW, Barker AH et al. and the
1357–9. CAPS Investigators. Classification of deaths after myocar-
[114] Sandrone G, Mortara A, Torzillo D, La Rovere MT, dial infarction as arrhythmic or nonarrhythmic (the Cardiac
Malliani A, Lombardi F. Effects of beta blockers (atenolol or arrhythmia Pilot Study). Am J Cardiol 1989; 63: 1–6.
metoprolol) on heart rate variability after acute myocardial [133] Malik M, Camm AJ. Significant of long-term components of
infarction. Am J Cardiol 1994; 74: 340–5. heart rate variability for the further prognosis after acute
myocardial infarction. Cardiovasc Res 1990; 24: 793–803.
[115] Adamson PB, Huang MH, Vanoli E, Foreman RD,
[134] Malik M, Farrell T, Camm AJ. Circadian rhythm of heart
Schwartz PJ, Hull SS Jr. Unexpected interaction between
rate variability after acute myocardial infarction and its
-adrenergic blockade and heart rate variability before and
influence on the prognostic value of heart rate variability.
after myocardial infarction: a longitudinal study in dogs at
Am J Cardiol 1990; 66: 1049–54.
high and low risk for sudden death. Circulation 1994; 90:
[135] Bigger JT, Fleiss JL, Rolnitzky LM, Steinman RC. The
976–82.
ability of several short-term measures of RR Variability to
[116] Zuanetti G, Latini R, Neilson JMM, Schwartz PJ, Ewing DJ,
predict mortality after myocardial infarction. Circulation
and the Antiarrhythmic Drug Evaluation Group (ADEG).
1993; 88: 927–34.
Heart rate variability in patients with ventricular arrhyth-
[136] Fei L, Malik M. Short- and long-term assessment of heart
mias: effect of antiarrhythmic drugs. J Am Coll Cardiol 1991;
rate variability for postinfarction risk stratification. In:
17: 604–12.
Malik M, Camm AJ, eds. Heart rate variability. Armonk:
[117] Lombardi F, Torzillo D, Sandrone G et al. Beta-blocking Futura, 1995: 341–6.
effect of propafenone based on spectral analysis of heart rate [137] Bigger JT, Kleiger RE, Fleiss JL, Rolnitzky LM, Steinman
variability. Am J Cardiol 1992; 70: 1028–34. RC, Miller JP, and the Multicenter Post-Infarction Research
[118] Bigger JT Jr, Rolnitzky LM, Steinman RC, Fleiss JL. Group. Components of heart rate variability measured dur-
Predicting mortality after myocardial infarction from the ing healing of acute myocardial infarction. Am J Cardiol
response of RR variability to antiarrhythmic drug therapy. J 1988; 61: 208–15.
Am Coll Cardiol 1994; 23: 733–40. [138] Bigger JT, Fleiss JL, Rolnitzky LM, Steinman RC. Fre-
[119] Casadei B, Pipilis A, Sessa F, Conway J, Sleight P. Low quency domain measures of heart period variability to assess
doses of scopolamine increase cardiac vagal tone in the acute risk late after myocardial infarction. J Am Coll Cardiol 1993;
phase of myocardial infarction. Circulation 1993; 88: 353–7. 21: 729–36.
[120] De Ferrari GM, Mantica M, Vanoli E, Hull SS Jr, Schwartz [139] Camm AJ, Fei L. Risk stratification following myocardial
PJ. Scopolamine increases vagal tone and vagal reflexes in infarction: Heart rate variability and other risk factors. In:
patients after myocardial infarction. J Am Coll Cardiol 1993; Malik M, Camm AJ, eds. Heart rate variability. Armonk:
22: 1327–34. Futura, 1995; 369–92.
[121] Pedretti R, Colombo E, Sarzi Braga S, Car B. Influence of [140] Bannister R. Autonomic Failure. A textbook of clinical
transdermal scopolamine on cardiac sympathovagal interac- disorders of the autonomic nervous system. Oxford, New
tion after acute myocardial infarction. Am J Cardiol 1993; York: Oxford University Press, 1988.
72: 384–92. [141] Ewing DJ, Campbell IW, Clarke BF. The natural history of
[122] Vybiral T, Glaser DH, Morris G et al. Effects of low dose diabetic autonomic neuropathy. Q J Med 1980; 193: 95–108.
scopolamine on heart rate variability in acute myocardial [142] Smith S. Reduced sinus arrhythmia in diabetic autonomic
infarction. J Am Coll Cardiol 1993; 22: 1320–6. neuropathy: diagnostic value of an age related normal range.
[123] LaRovere MT, Mortara A, Pantaneleo P, Maestri R, Cobelli Br Med J 1982; 285: 1599–1601.
F, Tavazzi L. Scopolamine improves autonomic balance in [143] O’Brien IA, O’Hare P, Corrall RJM. Heart rate variability in
advanced congestive heart failure. Circulation 1994; 90: healthy subjects: effect of age and the derivation of normal
838–43. ranges for tests of autonomic function. Br Heart J 1986; 55:
[124] Hull SS Jr, Vanoli E, Adamson PB, De Ferrari GM, 348–54.
Foreman RD, Schwartz PJ. Do increase in markers of vagal [144] Ewing DJ, Neilson JMM, Shapiro JA, Reid W. Twenty four
activity imply protection from sudden death? The case of hour heart rate variability: effects of posture, sleep and time
scopolamine. Circulation 1995; 91: 2516–9. of day in healthy controls and comparison with bedside tests
[125] Zabel M, Klingenheben T, Hohnloser SH. Changes in auto- of autonomic function in diabetic patients. Br Heart J 1991;
nomic tone following thrombolytic therapy for acute myo- 65: 239–44.
cardial infarction: assessment by analysis of heart rate [145] Malpas SC, Maling TJB. Heart rate variability and cardiac
variability. J Cardiovasc Electrophysiol 1994; 4: 211–18. autonomic function in diabetes. Diabetes 1990; 39: 1177–81.



[146] Bianchi A, Bontempi B, Cerutti S, Gianogli P, Comi G, nonsustained and sustained ventricular tachycardia in pa-
Natali Sora MG. Spectral analysis of heart rate variability tients with coronary artery disease. Circulation 1993; 87:
signal and respiration in diabetic subjects. Med Biol Eng 1220–8.
Comput 1990; 28: 205–11. [166] Hohnloser SH, Klingenheben T, van de Loo A, Hablawetz
[147] Bellavere F, Balzani I, De Masi G et al. Power spectral E, Just H, Schwartz PJ. Reflex versus tonic vagal activity as
analysis of heart rate variation improves assessment of a prognostic parameter in patients with sustained ventricular
diabetic cardiac autonomic neuropathy. Diabetes 1992; 41: tachycardia or ventricular fibrillation. Circulation 1994; 89:
633–40. 1068–1073.
[148] Van den Akker TJ, Koelman ASM, Hogenhuis LAH, [167] Kocovic DZ, Harada T, Shea JB, Soroff D, Friedman PL.
Rompelman G. Heart rate variability and blood pressure Alterations of heart rate and of heart rate variability after
oscillations in diabetics with autonomic neuropathy. Auto- radiofrequency catheter ablation of supraventricular tachy-
medica 1983; 4: 201–8. cardia. Circulation 1993; 88: 1671–81.
[149] Guzzetti S, Dassi S, Pecis M et al. Altered pattern of [168] Lefler CT, Saul JP, Cohen RJ. Rate-related and autonomic
circardian neural control of heart period in mild hyperten- effects on atrioventricular conduction assessed through beat-
sion. J Hypertens 1991; 9: 831–838. to-beat PR interval and cycle length variability. J Cardiovasc
[150] Langewitz W, Ruddel H, Schachinger H. Reduced parasym- Electrophys 1994; 5: 2–15.
pathetic cardiac control in patients with hypertension at rest [169] Berger RD, Saul JP, Cohen RJ. Assessment of autonomic
and under mental stress. Am Heart J 1994; 127: 122–8. response by broad-band respiration. IEEE Trans Biomed
[151] Saul JP, Arai Y, Berger RD et al. Assessment of autonomic Eng 1989; 36: 1061–5.
regulation in chronic congestive heart failure by the heart [170] Berger RD, Saul JPP, Cohen RJ. Transfer function analysis
rate spectral analysis. Am J Cardiol 1988; 61: 1292–9. of autonomic regulation: I — The canine atrial rate response.
[152] Binkley PF, Nunziata E, Haas GJ, Nelson SD, Cody RJ. Am J Physiol 1989; 256: H142–52.
Parasympathetic withdrawal is an integral component of [171] Saul JP, Berger RD, Chen MH, Cohen RJ. Transfer function
autonomic imbalance in congestive heart failure: Demonstra- analysis of autonomic regulation: II — Respiratory sinus
tion in human subjects and verification in a paced canine arrhythmia. Am J Physiol 1989; 256: H153–61.
model of ventricular failure. J Am Coll Cardiol, 1991; 18: [172] Saul JP, Berger RD, Albrecht P, Stein SP, Chen MH, Cohen
464–72. RJ. Transfer function analysis of the circulation: Unique
[153] Townend JN, West JN, Davies MK, Littles WA. Effect of insights into cardiovascular regulation. Am J Physiol 1991;
quinapril on blood pressure and heart rate in congestive 261: H1231–45.
heart failure. Am J Cardiol 1992; 69: 1587–90. [173] Baselli G, Cerutti S, Civardi S, Malliani A, Pagani M.
Cardiovascular variability signals: Towards the identification
[154] Binkley PF, Haas GJ, Starling RC et al. Sustained augmen-
of a closed-loop model of the neural control mechanisms.
tation of parasympathetic tone with angiotensin converting
IEEE Trans Biomed Eng 1988; 35: 1033–46.
enzyme inhibitor in patients with congestive heart failure. J
[174] Appel ML, Saul JP, Berger RD, Cohen RJ. Closed loop
Am Coll Cardiol 1993; 21: 655–61.
identification of cardiovascular circulatory mechanisms.
[155] Woo MA, Stevenson WG, Moser DK, Middlekauff HR.
Computers in Cardiology 1989. Los Alamitos: IEEE Press,
Complex heart rate variability and serum norepinephrine
1990: 3–7.
levels in patients with advanced heart failure. J Am Coll
[175] Tsuji H, Venditti FJ, Manders ES et al. Reduced heart rate
Cardiol 1994; 23: 565–9.
variability and mortality risk in an elderly cohort: The
[156] Alexopoulos D, Yusuf S, Johnston JA, Bostock J, Sleight P, Framingham Study. Circulation 1994; 90: 878–83.
Yacoub MH. The 24 hour heart rate behavior in long-term [176] Vanoli E, Adamson PB, Lin B, Pinna GD, Lazzara R, Or
survivors of cardiac transplantation. Am J Cardiol 1988; 61: WC. Heart rate variability during specific sleep stages: a
880–4. comparison of healthy subjects with patients after myo-
[157] Stein KM, Bores JS, Hochreites C et al. Prognostic value and cardial infarction. Circulation 1995, 91: 1918–22.
physiological correlates of heart rate variability in chronic [177] Singer DH, Ori Z. Changes in heart rate variability associ-
severe mitral regurgitation. Circulation 1993; 88: 127–35. ated with sudden cardiac death. In: Malik M, Camm AJ, eds.
[158] Marangoni S, Scalvini S, Mai R, Quadri A, Levi GF. Heart Heart rate variability. Armonk: Futura, 1995: 429–48.
rate variability assessment in patients with mitral valve [178] Malfatto G, Rosen TS, Steinberg SF et al. Sympathetic
prolapse syndrome. Am J Noninvas Cardiol 1993; 7: 210–14. neural modulation of cardiac impulse initiation and repolari-
[159] Counihan PJ, Fei L, Bashir Y, Farrel TG, Haywood GA, zation in the newborn rat. Circ Res 1990; 66: 427–37.
McKenna WJ. Assessment of heart rate variability in hyper- [179] Hirsch M, Karin J, Akselrod S. Heart rate variability in the
trophic cardiomyopathy. Association with clinical and prog- fetus. In: Malik M, Camm AJ, eds. Heart rate variability.
nostic features. Circulation 1993; 88: 1682–90. Armonk: Futura, 1995: 517–31.
[160] Dougherty CM, Burr RL. Comparison of heart rate variabil- [180] Parati G, Di Rienzo M, Groppelli A, Pedotti A, Mancia G.
ity in survivors and nonsurvivors of sudden cardiac arrest. Heart rate and blood pressure variability and their inter-
Am J Cardiol 1992; 70: 441–8. action in hypertension. In: Malik M, Camm AJ, eds. Heart
[161] Huikuri HV, Linnaluoto MK, Seppanen T et al. Circadian rate variability. Armonk: Futura, 1995; 465–78.
rhythm of heart rate variability in survivors of cardiac arrest. [181] Bigger JT Jr, Fleiss JL, Steinman RC, Rolnitzky LM,
Am J Cardiol 1992; 70: 610–15. Schneider WJ, Stein PK. RR variability in healthy, middle-
[162] Myers GA, Martin GJ, Magid NM et al. Power spectral age persons compared with patients with chronic coronary
analysis of heart rate variability in sudden cardiac death: heart disease or recent acute myocardial infarction. Circula-
comparison to other methods. IEEE Trans Biomed Eng tion 1995; 91: 1936–43.
1986; 33: 1149–56.
[163] Martin GJ, Magid NM, Myers G et al. Heart rate variability
and sudden death secondary to coronary artery disease
during ambulatory ECG monitoring. Am J Cardiol 1986; 60: Appendix A
86–9.
[164] Vybiral T, Glaeser DH, Goldberger AL et al. Conventional Normal values of standard measures of heart
heart rate variability analysis of ambulatory electrocardio-
graphic recordings fails to predict imminent ventricular rate variability
fibrillation. J Am Coll Cardiol 1993; 22: 557–65.
[165] Huikuri HV, Valkama JO, Airaksinen KEJ et al. Frequency As no comprehensive investigations of all HRV indices
domain measures of heart rate variability before the onset of in large normal populations have yet been performed,


380 Task Force

the recording and the analytical part of the device. An
Normal values
Variable Units appropriate technology should be used to record a fully
(mean SD)
Time domain analysis of nominal 24 h[181] reproducible signal with precisely known HRV param-
eters, e.g. the test signal should be computer and/or
SDNN ms 141 39 hardware generated. Both brand new recorders as well
SDANN ms 127 35 as recorders which have been routinely used and rou-
RMSSD ms 27 12 tinely serviced for approximately a half of their lifetime
HRV triangular index 37 15 should be used in the tests if this is feasible (the testing
should not be delayed for newly introduced systems).
Spectral analysis of stationary supine 5-min recording
If a manufacturer claims that the device is capable
of analysing ECG records (such as Holter tapes) ob-
total power ms2 3466 1018 tained with recorders of other manufacturers, each
LF ms2 1170 416
HF ms2 975 203
combination should be tested independently.
LF n.u. 54 4 As the analysis of HRV by implantable devices
HF n.u. 29 3 may be foreseen, similar procedures as described further
LF/HF ratio 1·5–2·0 should be used to generate simulated intracardiac sig-
nals. If feasible, implantable devices with fully charged
batteries as well as devices with partly discharged
batteries should be tested.
some of the normal values listed in the following table
Test recordings
were obtained from studies involving small number of
It is intrinsically difficult to know precisely the HRV
subjects. The values should therefore have been consid-
parameters of any real ECG recordings independently of
ered as approximate and no definite clinical conclusions
equipment used to analyse the recording. Therefore,
should be based on them. The adjustment of normal
simulated ECG signals are preferable. However, the
limits for age, sex, and environment which is also needed
morphology of such simulated ECG signals as well as
has been omitted here because of the limited sources of
the HRV characteristics must closely reflect the mor-
data.
phology of real recordings. The discrete frequency used
The table lists only values of those measures of
to generate such signals must be substantially higher
HRV which might be suggested for standardisation of
than the sampling frequency of the tested device. Fea-
further physiological and clinical studies.
tures which should be introduced into such recordings
should include different factors known to influence or
potentially influence the precision of HRV assessment,
Appendix B e.g. variable noise levels, variable morphology of QRS
signals which may cause jitter of the fiducial point,
Suggestion of procedures for testing of randomly alternating noise in different channels of the
commercial equipment designed to measure signal, gradual and abrupt changes of HRV character-
heart rate variability istics, and different frequencies of atrial and ventricular
ectopic beats with realistic morphologies of the signal.
Concept The quality of records on magnetic tape based
In order to achieve comparable accuracy of measure- systems may not be constant during long-term recording
ments reported by different commercial equipment, each due to spool torque control, back tension, and other
device should be tested independently of the manufac- factors. Performance of all recorders can be influenced
turer (e.g. by an academic institution). Each test should by changes of the outside environment. Long-term
involve several short-term and, if applicable, long-term (e.g. full 24-h test) rather than short-term tests should
test recordings with precisely known HRV parameters therefore be used.
and with different morphological characteristics of the
ECG signal. If the involvement of the manufacturer is Testing procedures
required during the testing procedure (e.g. for manual Each device or each configuration of the device should
editing of the labels of QRS complexes), the manufac- be tested with several different recordings having differ-
turer must be blinded in respect of both the true HRV ent mixtures of features and different HRV characteris-
parameters of the testing recordings and the features tics. For each test record and for each selected portion of
used to obtain the signal. In particular, when the results the test record, the HRV parameters obtained from the
of the test are disclosed to the manufacturer for further commercial device should be compared with the known
improvement of the device or other wise, new tests characteristics of the initial signal. Any discrepancies
should involve a completely new set of test recordings. found should be itemised in respect of features intro-
duced into the recording, e.g. errors caused by increased
Technical requirements noise, errors due to fiducial point wander, etc. System-
Each device should be tested comprehensively including atic bias introduced by the equipment as well as its
all its parts. In particular, the test should involve both relative errors should be established.



Reporting the results Milano, Italy, Richard J. Cohen, Cambridge, U.S.A.,
A technical report of the testing should be prepared Philippe Coumel, Paris, France, Ernest L. Fallen,
solely at the testing site independent of the manufac- Hamilton, Canada, Harold L. Kennedy, St. Louis,
turer(s) of the tested device. U.S.A., Robert E. Kleiger, St. Louis, U.S.A., Federico
Lombardi, Milano, Italy, Alberto Malliani, Milano,
Italy, Arthur J. Moss, Rochester (NY), U.S.A., Jeﬀrey
Appendix C N. Rottman, St. Louis, U.S.A., Georg Schmidt,
München, Germany, Peter J. Schwartz, Pavia, Italy,
Members of the Task Force Donald H. Singer, Chicago, U.S.A.
While the text of this report was contributed to
The Task Force was composed of 17 members: and approved by all members of the Task Force, the
Co-chairmen: structure of the text was designed by the Writing Com-
A. John Camm, London, U.K, Marek Malik, London, mittee of the Task Force composed of the following
U.K. members:
Members: Marek Malik (Chairman), J. Thomas Bigger, A. John
J. Thomas Bigger, Jr., New York, U.S.A., Günter Camm, Robert E. Kleiger, Alberto Malliani, Arthur
Breithardt, Münster, Germany, Sergio Cerutti, J. Moss, Peter J. Schwartz.


Guidelines heart rate_variability_ft_1996[1]

More Related Content

What's hot (20)

Similar to Guidelines heart rate_variability_ft_1996[1] (20)

More from ES-Teck India (20)

Recently uploaded (20)

Guidelines heart rate_variability_ft_1996[1]