Eur J Appl Physiol (1993) 66:207-213
,u... A p p l i e d Physiology
Joumal of
and OccupationalPhysiology @ Springer-Verlag 1993
Body position affects the power spectrum of heart rate variability during dynamic exercise Renza Perinil, Claudio Orizio 1, Stefania Milesil, Luca Biancardi 1, Giuseppe Baselli a, and Arsenio Veicsteinas 1 1 Cattedra di Fisiologia Umana, Dipartimento di Scienze Biomediche e Biotecnologie, Universit~ di Brescia, via Valsabbina 19, 1-25123 Brescia, Italy a Dipartimento di Elettronica per l'Automazione, UniversitA di Brescia, Brescia, Italy Accepted October 29, 1992
Summary. The power spectrum analysis of R-R interval variability (RRV) has been estimated by means of an autoregressive method in six men in supine (S) and sitting (C) postures at rest and during steady-state cycle exercise at about 1407o, 28%, 45%, 6707o of the maximal oxygen consumption (07o 1202max). The total power of RRV decreased exponentially as a function of exercise intensity in a similar way in both postures. Three components were recognized in the power spectra: firstly, a high frequency peak (HF), an expression of respiratory arrhythmia, the central frequency (fccntral) of which increased in both S and C from a resting value of about 0.26 Hz to 0.42 Hz at 6707o I202max; secondly, a low frequency peak (LF) related to arterial pressure control, the fcentral of which remained constant at 0.1 Hz in C, whereas in S above 28070 I?O2m~x decreased to 0.07 Hz; and thirdly, a very low frequency component (VLF; less than 0.05 Hz, nOfcentral). The power of the three components (as a percentage of the total power) depended on the body posture and the metabolic demand. H F % at rest was 30.3 (SEM 6.6) 07o in S and 5.0 (SEM 0.8) 07o in C. During exercise H F % decreased by about 3007o in S and increased to 19.7 (SEM 5.5) % at 2807o P'O2m~x in C. LF% was lower in S than in C at rest [31.6 (SEM 5.7) 070 vs 44.9 (SEM 6.4) 07o; P < 0 . 0 5 ] , remaining constant up to 2807o P-O2ma~. At the highest intenstities it increased to 54.0 (SEM 15.6) 07o in S whereas in C it decreased to 8.5 (SEM 1.6) 070. VLF represented the remaining power and the change was in the opposite direction to LF. The changes in power spectrum distribution of RRV during exercise depended on the intensity and the body posture. In particular, the LF peak showed opposite trends in S and C tasks, thus suggesting a different readjustment of arterial pressure control mechanisms in relation to the blood distribution and peripheral resistances. Key words: Supine exercise - Power spectrum analysis Heart rate control system - Posture and circulatory system Correspondence to: A. Veicsteinas
Introduction The spontaneous fluctuations of heart rate (fc) around the mean value, i.e. fc variability (fo .... ), have been found to decrease as a function of exercise intensity (Arai et al. 1989; Bernardi et al. 1990; Perini et al. 1990). Moreover, changes in the power spectrum distribution occur with respect to rest. In the resting upright position, a dominant component at about 0.1 Hz (low frequency, LF) and a second component at respiratory frequency (about 0.24 Hz, high frequency, HF) have been observed (Pomeranz et al. 1985; Pagani et al. 1986). With increasing exercise intensity, a tendency in LF peak to disappear has been shown both during incremental tasks (Bernardi et al. 1990) and at constant loads above about 30°70 of the maximal oxygen consumption (12Ogmax) (Perini et al. 1990). HF peak, in contrast, has been shown to shift towards higher frequencies and account for the major part of the total power at almost maximal intensity (Bernardi et al. 1990). The LF component has been shown to relate to the arterial pressure control system (Sayers 1973; Akselrod et al. 1981). The change in control mechanisms due to exercise - baroreceptor reset and decrease in peripheral resistance - have been suggested to be responsible for the disappearance of LF fluctuation in fo . . . . during exercise (Perini et al. 1990). The HF component has been shown to be an expression of respiratory arrhythmia (Pomeranz et al. 1985) and during exercise, the H F peak to persist due to the increased ventilation (Arai et al. 1989; Bernardi et al. 1990). At rest the body posture has been found to affect the fc .... power spectrum (Pomeranz et al. 1985; Pagani et al. 1986), which reflects different sympatho-vagal interactions on the heart, which in turn depends on the different cardiovascular adjustments to the influence of gravity on blood distribution (Astrand and Rodahl 1986). With the aim of clarifying some aspects of the interaction among factors which modulate f~, we have studied the effect of the supine posture during dynamic exer-
208
cise on the changes in the power spectrum of fc, var, and in particular on the LF rhythm. It is noteworthy that at the intensities of sitting exercise at which the LF peak disappeared, in supine tasks a marked increase in LF percentage power was found.
Methods Subjects. Six sedentary men [mean age 25.3 (SEM 1.2) years; body mass 69.7 (SEM 3.5) kg; height 178.5 (SEM 3.8) cm] volunteered for the study. The subjects were untrained medical students, familiar with laboratory tests.
Exercise tests. Exercise consisted of pedalling for 6 min on a cycle ergometer (Ergomedic 818, Monark) at a pedal frequency of 50 rpm at 0, 50, 100, 150 W (unloaded, low, medium and high intensity, respectively). Each tasks was performed in the supine (S) posture. Control values (C) were obtained in the sitting posture. During S exercise the subject lay on a bed with his shoulders against a fixed support to avoid isometric contractions of the arms. The crank axis of the pedals was at the level of the bed. A series of trial runs was performed to measure the individual's oxygen consumption (1202; open circuit method) during the last minute of exercise at each intensity in both postures. In separate sessions the 1202~x of each subject was determined during cycle ergometer exercise in both S and C. After a 5-rain period at 100 W, the exercise intensity was increased by 25 W every 4 min, measuring 1202 during the last minute at each intensity. The 1202m~x was considered to have been reached by establishing a plateau of 1202 with increasing intensity (Astrand and Rodahl 1986).
Experimental procedure. The actual experiments were carried out in the morning at the same time of day, in a room at constant temperature (19-21 ° C). The different intensities were performed in a randomized sequence on different days. The electrocardiographic (ECG) signal was continuously recorded: 1. At rest for 10 min while the subject was sitting on the cycle in C and supine on the bed in S. No successive task was required. 2. Again at rest for a few minutes and during exercise. The ECG was fed on-line on a computer (Olivetti 290 S) via a 12 bit A - D converter at a sampling rate of 256 Hz.
Signal analysis. The further analysis of the signals was carried out on a personal computer. The occurrence of the R-wave in the ECG was detected by an automatic procedure using a derivativethreshold algorithm (Bartoli et al. 1985). The sequences of R-R intervals at rest and during exercise were then obtained and stored. Each period was analysed separately. A trend was subtracted from the signal, either linear at rest or exponential during exercise, after having discarded the initial transient, i.e. about 2 min. The series {ti} was then obtained. Its variance represented the total power offc.v~r and the root mean square quantified the amplitude off~ . . . . The RR-interval variability (RRV) percentage (RRV%) was calculated as:
Individual spectrum components were recognized via a decomposition procedure, which relates each spectrum component to a pair of conjugate poles in the AR model (Zetterberg 1978; Baselli et al. 1986). The components were classified as very low frequency (VLF), below 0.05 Hz; LF between 0.05 and 0.15 Hz; HF between 0.15 and 1.0 Hz. If two or more components were present in each band, the sum of their power and the mean frequency were considered the power and the central frequency ffce,tr~l), respectively, of that band. Both absolute and percentage power, i.e. the percentage of the total power as the sum of the three components, and the fcentral of each component were calculated.
Statistical analysis. One-way variance analysis and paired student's t-test were used to verify the significance of the difference at P
Results
The ~rO2max w a s on average 43.5 (SEM 2.3) m l . k g - l . m i n -1, and 46.5 (SEM 2.4) m l . k g - l . m i n -1 in S and C, respectively. The 1702 at rest was similar in the two positions, being 9.3°70 (SEM 0.9) 1702. . . . and 10.3070 (SEM 1.1) I?O2m~x, in S and C, respectively. The oxygen demand of the four tasks corresponded to 15.0°70 (SEM 1.3) VO2m~; 30.7°70 (SEM 2.6) I?O2max; 47.0°70 (SEM 4.1) l?O2m~, and 68.5°70 (SEM 5.8) l?O2m,x in S and 12.7070 (SEM 0.7) l?O2.~,x; 23.6°70 (SEM 2.7) IkO2max; 44.2070 (SEM 3.9) VO2max and 64.0°70 (SEM 5.1) l?O2m~x in C. The changes in mean R-R interval are reported as a function of °7olkO2max in Fig. 1 for both positions. From a resting value of 919 (SEM 56) ms, and 808 (SEM 39) ms, in S and C respectively, at 0 W the mean R-R interval value did not change in C, but decreased to 742 (SEM 39) ms in S. Thereafter similar decreases occurred in each of the two positions. The total power o f f ° . . . . (Fig. 2, top) at rest was not significantly different in S and C [3210 (SEM 410) ms 2,
1000
800 R-R
<~
(ms) 6O0
RRV% = rms (tO/T" 100 where T is the mean R-R value and rms is the root mean square (Perini et al. 1990). The power spectrum density of discrete series {ti} in each period was estimated by an autoregressive method (AR), in which coeficients were calculated via the Levinson Durbin algorithm (Kay and Marple 1981). The validity of the model was checked by testing the whiteness of the prediction error, Anderson's test (Kay and Marple 1981), and the choice of the AR model order, was performed using the Akaike Information Criterion (Akaike 1970).
400 0
. 20
.
.
. 60
40
% V02
80
max
Fig. 1. Mean and SEM values of average R-R interval as a function of relative exercise intensity (percentage maximal oxygen consumption, % 1202max) in supine (0) and sitting (O) postures
209
I suP,NE ]
4000
[S,TT,NG ]
°°k2 L2 o.oLZ ,
Total power
s 2. Hz -1
(ms')
0
2000
s 2. Hz -1 o
0 10
°°1 L2 L2 o.oo L o.ooo1
S2" Hz-1
0
RRV % "-.
SITTING
$2. HE-1
SUPINE
0
0
O__t~
i
i
i
i
20
40
60
80
S2. Hz -1
% V02" mQx
Fig. 2. Mean and SEM values of the total power of R-R variability (RRV) in absolute (ms z, top) and normalized units (RRV%, bottom) as a function of relative exercise intensity (percentage maximal oxygen consumption, % l~Ozm,~)in supine (e) and sitting (©) postures and 3640 (SEM 660) ms2], respectively. With increasing exercise intensity, the total power decreased exponentially at the same rate in both postures. In Fig. 2 (bottom), the RRV% is given as a function of °7012Oz. . . . At rest RRV% was not significantly different in S and C [6.2 (SEM 0.33) %, and 7.6 (SEM 0.90) %, respectively]. A marked decrease occurred during the zero intensity tasks in both positions, particularly in S where a halved value was found. Then, a further linear decline occurred to 1.6 (SEM 0.1) % and 1.4 (SEM 0.2) % in C and S, respectively, at the highest intensities. The slope of the line RRV% versus % [ ~ r O z m a x was steeper in C than in S, - 0 . 7 RRV% and - 0 . 4 RRV% per 10% lkO2max increment, respectively. In Fig. 3 the power spectra of RRV at rest and at steady-state of the four intensities are shown for a representative subject in both S and C. Note the decreasing scale on the y axis. In comparison to rest, a fivefold decrease in total power occurred at 0 and 50 W, a 50-fold at 100 W and a 500-fold at 150 W in both positions. At rest three components can be detected in the spectra. A large part of the power is below 0.05 Hz (VLF), where a peak is not recognizable. The second component is a well-defined peak at 0.1 Hz (LF), which is the dominant one in C. Finally in S, a peak particularly high and sharp is present at about 0.24 Hz (HF), while in C it is less evident at 0.19 Hz.
0
• 0.0
0.4 Hz
0.8
0.0
0.4
0.8
Hz
Fig. 3. Power spectrum distribution of R-R variability at rest and during steady-states of the four exercise intensities in supine (left panel) and in sitting (rightpanel) postures in a representative subject. The dominant autoregressive spectrum components are shown as very low frequency (D), low frequency (N), high frequency (11). Note the different scale of y axes, decreasing with exercise intensity
With exercise, the three components are recognizable at each of the intensities in S. The LF peak tends to shift towards the left and the HF one towards the right of the spectrum. In addition, the LF peak is sharper with increasing exercise intensity. In C, in contrast, LF enlarges and tends to disappear with increasing intensity, whereas the HF peak is easily recognizable at increasing frequencies, as in S. Power spectra distributions comparable to those given for the representative subject were found in all others. In particular, the tendency for LF peak to disappear in C tasks and to persist in S at increasing intensities was constantly observed. Detailed analyses of each component in the following figures show the results of the group. Figure 4 shows the fcentral of the LF (top) and HF (bottom) peaks as a function of exercise intensity. At rest the fcentral were 0.103 (SEM 0.008) Hz for LF and 0.24 (SEM 0.03) Hz for HF in S, and 0.099 (SEM 0.01) Hz, and 0.28 (SEM 0.03) Hz in C. With exercise, the fcentral of the LF peak did not change in C, whereas in S
210 0.15
LF
80
VLF
SITTING
power
Hz
%
J
SITTING
40
s"
......
0.10 SUPINE
I 0 SUPINE 80
0.05 0.5
SUPINE/
Hz
LF SUPINE [
power %
L
4O
0,4
SITTINa" ' 6 -
0 4O
0.3
power
% 0.2
i
0
20
i
40 % 902
i
i
60
80
20
! I
L ./
-
max
(~
Fig. 4. Mean and SEM values of the central frequencies of low frequency (top) and high frequency (bottom) peaks as a function of relative exercise intensity (percentage maximal oxygen consumption, O7oI2Ozm~) in supine (0) and sitting (O) postures
above 35o7o ~rO2max it decreased sharply, being at 70% 1202m~x 0.074 (SEM 0.005) Hz. HF fcentral increased linearly as a function of exercise intensity in S and C, respectively, i.e.: fcentral (Hz) : 0.239 + 0.0031 (% ~rO2max) /~< 0.05 r=0.93, n=5; fcentral (Hz) ~--0.262 + 0.0019 (% lkO2max) P < 0.05 r = 0.98, n = 5 No statistical difference was found when student's paired t-tests on individual values in the two positions were performed. It is noteworthy that in the S zero intensity the fcentral was found to be significantly higher than at rest and, surprisingly, at 50 W [0.334 (SEM 0.05) Hz, vs 0.317 (SEM 0.048) Hz]. No f~entra] in the VLF component can be detected. In Fig. 5, from the top downwards, the average values of percentage powers of the VLF, LF and FIF components in S and C are shown as a function of % f'O2m,x. At rest VLF represents the major part of the power, being in S 38.0 (SEM 7.8) % and in C 50.1 (SEM 6.4) %; LF is significantly lower in S [31.6 (SEM 5.7) %] than in C [44.9 (SEM 6.4) %]. For HF the opposite is true [in S 30.3 (SEM 6.6) % and in C 5.0 (SEM 0.8) %]. Practically no changes were observed during exercise in comparison to rest for VLF% and LFOT0UP to about 35% 12Ozmaxin the two postures. Then, opposite trends
HF
SITTING I
I
I
l
20
40
60
80
% (/02 mox
Fig. 5. From top to bottom, mean and SEM values of the power of very low frequency, low frequency and high frequency components expressed as percentages of the total (VLF%, LFOT0,HFO7o) as a function of relative exercise intensity (% 1202m,x) in supine (0) and sitting ((3) postures
were found in S and C. In S a significant decrease in VLFO70 occurred at 23.3 (SEM 13.3) o70 at 70O/o 1202ma~, whereas in C an increase to 73.6 (SEM 5.8) °7o was found at high intensity. Over the same range of intensities in S, an increase in LF °7o up to 54.0 (SEM 15.6) o70 and in C a decrease to 8.5 (SEM 1.6) % occurred. With exercise in S, H F % showed a tendency to be lower than at rest, in particular at medium intensity, being on average 20°7o. In C a clear increase in HFO70 from rest to low intensities [19.7 (SEM 5.5) °7o] occurred and then a tendency to decrease slightly was observed, with values no different to those in S.
Discussion
The present study aimed to analyse the effect of posture during steady-state cycle exercise up to 70% 1202max on the power spectrum of fc ..... The same procedure and algorithm used by our group (Perini et al. 1990) during C exercise over the same range of intensity were adopted. The results obtained in this new set of experi-
211
ments on the one hand confirmed our previous findings on changes in fc .... power spectrum during C exercise and on the other provided new data on a topic not thoroughly examined in the current literature, i.e. the combined effect of S posture and exercise intensity on the power spectrum distribution of RRV. In addition, the new set of C tasks allowed us to verify the reliability of the quantitative estimation of the spectrum parameters. The mean values and the interindividual variability of both fcentral and percentage powers of the spectrum components found during C exercises are quite similar to those observed in different subjects in our previous paper (Perini et al. 1990). These results add further evidence to previous observations on the reproducibility of the results of power spectrum analysis by the AR method (Pagani et al. 1986). At rest the mean R-R interval in S was significantly higher than in C. General agreement has been found to exist that the lower fo in S [66 (SEM 3.7) beats.min -1] than in C [75 (SEM 3.7) beats.min -1] can be ascribed to a higher vagal tone, which should lead to a higher total power infc .... (Pomeranz et al. 1985; Pagani et al. 1986). In contrast, absolute and percentage power values were similar in the two positions. This difference may have been due to the fact that C and not standing was compared with S. However, it might be hypothesized that mechanical inputs arising from muscles which have been shown to be involved info control during voluntary contractions (Mitchell et al. 1983; Rowell and O'Leary 1990) are active also at rest in relation to the body posture and thus can affect f~,var. The role of muscle afferents is reinforced when zero intensity exercise is considered. In fact, in both positions a large similar decrease in fc .... was found. This occurred even when different responses inf~ were observed in S versus C, i.e. a significant increase in S versus no change in C. Therefore, following the suggestion of Bernardi et al. (1990), we believe that the equivalence between the vagal tone and the total power of f~ .... - commonly reported in the literature - has to be considered with caution. Nevertheless, the main role of vagal activity in determining the total power o f f ° .... was confirmed when our results from exercise were taken into account. In fact, about three-quarters of the total power decreased in both body postures within 35% lkO2 . . . . i.e. the range of intensity in which the increase info has been shown to occur due to the withdrawal of vagal tone (Robinson et al. 1966; Rowell and O'Leary 1990). Even if posture did not influence at rest the magnitude o f f c ..... it influenced the power spectrum distribution. In agreement with previous data (Pomeranz et al. 1985; Pagani et al. 1986), while resting supine we found a H F peak six times higher than when upright, which in contrast was characterized by a dominant LF peak. With C exercise at steady-state (Perini et al. 1990) or at continuously increasing intensities (Bernardi et al. 1990) changes in the power spectrum distribution have been shown. The present study has demonstrated that the power spectrum distribution during exercise was largely influenced also by the body position (see Fig. 3).
The HF peak, which is a quantitative expression of respiratory arrhythmia (Pomeranz et al. 1985), has been found to be constantly present during exercise, due to the increased ventilation. In particular, the fcen*ra~of the HF peak was found to be linearly related to % 1202max with a similar slope in the two postures. This should reflect the increase in respiratory rate, as has been suggested by observations both at rest with controlled breathing (Pomeranz et al. 1985; Pagani et al. 1986) and during exercise (Arai et al. 1989; Bernardi et al. 1990). Also the "excessive" increase in the fcentral of the HF peak during unloaded S pedalling in respect to the corresponding resting value could have been an expression of the "extra" increase in respiratory rate, which often occurs during low intensity exercise. In fact, when the metabolic demand is low, the respiratory response can be influenced by other factors, i n particular by the rate of limb movements (,&strand and Rodhal 1986). With exercise, the H F % power showed no significant changes during S tasks. On the contrary, in C exercise H F % increased up to 30% lkO2 . . . . reaching values as high as those found in S. This increase of H F % from rest to exercise had not previously been found. This discrepancy might depend on the fact that in the present study the resting values were obtained during separate sessions and not immediately before exercise (Perini et al. 1990; Bernardi et al. 1990), when psychological factors a n d / o r breathing into a mouthpiece could affect the respiratory pattern, cause hyperventilation and thus high HF values. The sharp increase in H F % observed in sitting - and its absence in S exercise - could be related to the much higher increase in venous return occurring with exercise in C than in S (Poliner et al. 1980). In fact fluctuations in venous return related to respiration have been suggested as a mechanism generating respiratory arrhythmia (Melcher 1976). Moreover, recently it has been postulated that during exercise the changes in venous return might be a basis for a non-autonomic mechanism modulating fo at respiratory frequency (Bernardi et al. 1990). The LF peak has been shown to be due to the baroreceptor reflex activated by oscillations in arterial pressure (Sayers 1973; Akselrod et al. 1981), in turn related to a spontaneous activity of the vessels (Akselrod et al. 1985) a n d / o r to a central sympathetic discharge (Gebber 1980). The LF peak remained unchanged compared to rest up to about 30% ~rO2maxin C exercises, confirming our previous observations (Perini et al. 1990). The same was true in S exercises. At higher intensities, a different trend in LF% occurred while pedalling supine and sitting. In fact, in C a tendency for LF peak to disappear was found, in agreement with previous data (Perini et al. 1990; Bernardi et al. 1990). In S, in contrast, we observed an increase in LF% coupled with a shift of the foentra~ to lower values. It is noteworthy that the total power in the two exercise modes was absolutely comparable (see Fig. 2). Therefore, the detection (S) or not (C) of an LF peak was not affected by the extremely low variability of the signal, i.e. the accuracy of the method was maintained.
212 The different pattern of LF peak observed in the different postures therefore requires further comment. The lack of changes in LF peak up to about 30°7o I202max in both body postures would seem to suggest that the modulation o f f c at 0.1 Hz was unaffected until the cardiovascular response to exercise was determined mainly by the increase in cardiac output via the vagal withdrawal (Rowell and O ' L e a r y 1990). In contrast, significant changes in LF component were observed at intensities at which, in addition, increases in sympathetic activity have been shown to occur (Galbo 1983; Rowell and O ' L e a r y 1990). The consequent readjustments in the cardiovascular system would seem to have influenced the LF rhythm. The effect, however, seems to depend dramatically on the body posture. We have hypothesized in our previous paper (Perini et al. 1990) that the decrease in LF component observed during C exercise at medium and high intensities could be related to the changes in the baroreceptor mechanisms a n d / o r to the decrease in peripheral resistances. In S exercise, because of gravitational effects, a lower blood flow in the leg muscles (Folkow et al. 1971) and reduced muscle vasodilatation (Rost 1987) have been found to occur. Therefore, both the reset of baroreceptors, which has been shown to be affected by muscle afferents (Rowell and O ' L e a r y 1990), and the decrease in resistances, could be different to that in the sitting pedalling position. This might justify the opposite trend in LF rhythm observed during exercise in the two body postures. In addition, this different trend might be the expression of the combined effect of exercise and body posture on the resonance properties of the loops involved in arterial pressure control which have been suggested as a generation mechanism of the rhythm (Baselli et al. 1988). At variance with others workers (Pagani et al. 1986; Bernardi et al. 1990), we did not use normalized units (the percentage of the sum of LF and H F components), i.e. we did not discard the VLF component. This allowed to us to recognize both the effective and the relative changes in LF and H F power in respect to rest. The VLF power represented a not negligible part of the total power both at rest and during exercise, confirming previous observations (Saul 1990; Perini et al. 1990; Y a m a m o t o et al. 1991). The interpretation of the physiological meaning of VLF is beyond the scope of this study. Nevertheless, it has been suggested that fluctuations in f~ at frequencies below 0.05 Hz are the expression of the influence of humoral and central factors of cardiovascular control a n d / or of slow respiratory oscillations (Saul 1990), which are presumably changed by muscle exercise depending on the body posture. In conclusion, during dynamic submaximal exercise the power spectrum distribution of f¢ .... was found to be affected not only by exercise intensity but also and substantially by body posture. In particular, at intensities above about 30% ~rO2maxthe LF peak tended to disappear in C whereas it increased in S. Therefore, the changes in fo at 0.1 Hz during exercise would seem to be
the result of m a n y factors, the intensity of exercise being neither only one nor the main one. Th e analysis of the variability of arterial pressure and its correlation with fc .... during exercise could contribute to the clarification of the mechanisms controlling the LF rhythm.
Acknowledgements. This work was supported in part by the Consiglio Nazionale deUe Ricerche, CNR, Roma, Italy and by Centro di Studio e Ricerca di Fisiologia Muscolare e dello Sport, Brescia, Italy. References Akaike H (1970) Statistical predictor identification. Am Int Stat Math 22: 203-217 Akselrod S, Gordon D, Ubel FA, Shannon DC, Barger AC, Cohen RJ (1981) Power spectrum analysis of heart rate fluctuation: a quantitative probe of beat-to-beat cardiovascular control. Science 213:220-222 Akselrod S, Gordon D, Madwed JB, Snidman NC, Shannon DC, Cohen RJ (1985) Hemodynamic regulation: investigation by spectral analysis. Am J Physiol 249:H867-H875 Arai Y, Saul JP, Albrecht P, Hartley LH, Lilly LS, Cohen RJ, Colucci WS (1989) Modulation of cardiac autonomic activity during and immediately after exercise. Am J Physiol 256: H132-H141 Astrand PO, Rodahl K (1986) Textbook of work physiology. McGraw Hill, New York, pp 168-173,358-360 Bartoli F, Baselli G, Cerutti S (1985) AR identification and spectral estimate applied to the R-R interval measurements. Int J Biomed Corflput 16: 201-215 Baselli G, Cerutti S, Civardi S, Liberati D, Lombardi F, Malliani A, Pagani M (1986) Spectral and cross-spectral analysis of heart rate and arterial blood pressure variability signals. Comput Biomed Res 19: 520-534 Baselli G, Cerutti S, Civardi S, Malliani A, Pagani M (1988) Cardiovascular variability signals: towards the identification of a closed-loop model of the neural control mechanisms. IEEE Trans Biomed Eng 35 : 1033-1046 Bernardi L, Salvucci F, Suardi R, Sold/t PL, Calciati A, Perlini S, Falcone C, Ricciardi L (1990) Evidence for an intrinsic mechanism regulating heart rate variability in the transplanted and the intact heart during submaximal dynamic exercise? Cardiovasc Res 24: 969-981 Folkow B, Haglund U, Jodal M, Lundgren O (1971) Blood flow in the calf muscle of man during heavy rhythmic exercise. Acta Physiol Scand 81 : 157-163 Galbo H (1983) Hormonal and metabolic adaptation to exercise. Thieme, Stuttgart, pp 5-28 Gebber GL (1980) Central oscillators responsible for sympathetic nerve discharge. Am J Physiol 239: H143-H155 Kay SM, Marple SL (1981) Spectrum analysis: a modern perspective. Proc IEEE 69:1380-1419 Melcher A (1976) Respiratory sinus arrhythmia in man. A study in heart rate regulating mechanisms. Acta Physiol Scand [Suppl] 435 : 1-31 Mitchell JH, Kaufman MP, Iwamoto GA (1983) The exercise pressor reflex: its cardiovascular effects, afferent mechanisms, and centr/tl pathways. Annu Rev Physiol 45 : 229-242 Pagani M, Lombardi F, Guzzetti S, Rimoldi O, Furlan R, Pizzinelli P, Sandrone G, Malfatto G, Dell'Orto S, Piccaluga E, Turiel M, Baselli G, Cerutti S, Malliani A (1986) Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circ Res 59:178-193 Perini R, Orizio C, Baselli G, Cerutti S, Veicsteinas A (1990) The influence of exercise intensity on the power spectrum of heart rate variability. Eur J Appl Physiol 61 : 143-148
213 Poliner LR, Dehmer G3, Lewis SE, Parkey RW, Blomqvist CG, Willerson JT (1980) Left ventricular performance in normal subjects: a comparison of the responses to exercise in the upright and supine positions. Circulation 62:528-534 Pomeranz B, Macaulay RJB, Caudill MA, Kutz I, Adam D, Gordon D, Kilborn KM, Barger AC, Shannon DC, Cohen RJ, Benson H (1985) Assessment of autonomic function in humans by heart rate spectral analysis. Am J Physiol 248:H151H153 Robinson BF, Epstein SE, Beiser GD, Braunwald E (1966) Control of heart rate by the autonomic nervous system. Circ Res 19:400-411 Rost R (1987) Athletics and the heart. Year Book Medical Publisher, Chicago, pp 10-16
Rowell LB, O'Leary DS (1990) Reflex control of the circulation during exercise: chemoreflexes and mechanoreflexes. J Appl Physiol 69: 407-418 Sayers BMcA (1973) Analysis of heart rate variability. Ergonomics 16:17-32 Saul JP (1990) Beat-to-beat variations of heart rate reflect modulation of cardiac autonomic outflow. News Physiol Sci 5:3237 Yamamoto Y, Hughson RL, Peterson JC (1991) Autonomic control of heart rate during exercise studied by heart rate variability spectral analysis. J Appl Physiol 71 : 1136-1142 Zetterberg LH (1978) Estimation of parameters for a linear difference equation with application to EEG analysis. Math Biosci 5 : 227-275