Limited data are available to describe the regulation of heart rate (HR) during sleep in spaceflight. Sleep provides a stable supine baseline during preflight Earth recordings for comparison of heart rate variability (HRV) over a wide range of frequencies using both linear, complexity, and fractal indicators. The current study investigated the effect of long-duration spaceflight on HR and HRV during sleep in seven astronauts aboard the International Space Station up to 6 mo. Measurements included electrocardiographic waveforms from Holter monitors and simultaneous movement records from accelerometers before, during, and after the flights. HR was unchanged inflight and elevated postflight [59.6 ± 8.9 beats per minute (bpm) compared with preflight 53.3 ± 7.3 bpm; P < 0.01]. Compared with preflight data, HRV indicators from both time domain and power spectral analysis methods were diminished inflight from ultralow to high frequencies and partially recovered to preflight levels after landing. During inflight and at postflight, complexity and fractal properties of HR were not different from preflight properties. Slow fluctuations (<0.04 Hz) in HR presented moderate correlations with movements during sleep, partially accounting for the reduction in HRV. In summary, substantial reduction in HRV was observed with linear, but not with complexity and fractal, methods of analysis. These results suggest that periodic elements that influence regulation of HR through reflex mechanisms are altered during sleep in spaceflight but that underlying system complexity and fractal dynamics were not altered.
- heart rate
- power spectral analysis
- complexity and fractal analysis
human spaceflight dramatically alters cardiovascular demands of day-to-day work tasks (18) and eliminates the need for the normal regulatory adjustments of the cardiovascular system associated with postural change on Earth (9, 47). Without appropriate countermeasures, including regular exercise, there would be marked cardiovascular deconditioning that could be detected by changes in heart rate (HR) and heart rate variability (HRV) (55). For more than 40 years, researchers have reported that inflight HR might be reduced (19, 22, 25, 35), elevated (9, 25), or unchanged (2, 11, 21, 46, 52) compared with preflight measurements. Interpretation of these findings is complicated by uncertain conditions for data collection. However, recent investigations of astronauts on the International Space Station (ISS) reported no change from preflight resting HR under controlled conditions in space (2, 26, 52).
Measurement of HRV has likewise been complicated by the variable conditions of data collection. An early report of 24-h HRV from cosmonauts on the Russian Mir station (21) observed a reduction in low-frequency spectral power during days 3–7, but no changes over the remainder of the flight (up to 179 days). Another study from Mir reported that HRV during short sessions of paced breathing was generally reduced in long-duration spaceflights compared with preflight supine values (11). Data from three recent investigations of astronauts on the ISS have revealed some conflict. HRV was unchanged from preflight semirecumbent values except for a strong trend (P = 0.07) during 0.1 Hz controlled breathing (2), and no changes were observed from supine rest in a second study (52), but we found in astronauts investigated in the current study that high-frequency HRV was reduced (26).
To avoid some of the complications of altered patterns of daily activity and effect of body posture, sleep might yield important information on the fundamental nature of cardiovascular regulation. Several investigators have reported that compared with preflight values, there was no change in the sleep time HR during spaceflight (18, 19, 46), and one study reported no change in the long-term scaling (fractal) behavior of HR (HRV) compared with healthy controls (28). However, another study reported lengthening of the R-R interval by about 100 ms, and no change or an increase in high-frequency HRV (22). To date, no study has incorporated simultaneous measurement of movement indicators with collection of HR data during sleep to assist in interpretation of the data. The mean HR during sleep has been previously reported for the astronauts in the current study as being unchanged during flight (18), and therefore, we hypothesized that HRV will likewise be unchanged compared with preflight values, whereas postflight HRV will be reduced.
Seven astronauts (1 female, age 47 ± 4.6 years, height 176 ± 5.0 cm, weight 81 ± 9.6 kg) who participated in space missions aboard the ISS (52–199; 144 ± 49 days) were studied. Each subject was given full verbal and written details of the experiment and signed a consent form. The experiment protocol conformed to the guidelines in the Declaration of Helsinki and was approved by the Office of Research Ethics at the University of Waterloo, and the Committee for the Protection of Human Subjects at Johnson Space Center.
Protocol and data collection.
Measurements were obtained in four test sessions for each subject: 41 ± 30 days (minimum 18 days) before flight (preflight); 2–3 wk after launch (early inflight); 2–3 wk before landing (late inflight); and 1 day (n = 5), 2 days (n = 1), or 3 days (n = 1) after return to Earth (postflight). Preflight tests were performed at the Johnson Space Center, Houston, TX (n = 4) and the Gagarin Cosmonaut Training Center, Star City, Russia (n = 3). Postflight tests were completed at the Kennedy Space Center, Cape Canaveral, FL (n = 3), the Johnson Space Center (n = 1, landed at the Dryden Flight Research Center), and the Gagarin Cosmonaut Training Center (n = 3).
In each test session, subjects were asked to wear Holter monitors [for the first two subjects and the early inflight session of the third subject these were designated as HM1 (Digicorder Model 483; Del Mar Reynolds Medical, Irvine, CA); for all other sessions these were designated as HM2 (H12+ Holter; Mortara Instrument, Milwaukee, WI)] for at least 24 h to collect the electrocardiographic (ECG) waveforms. Simultaneously, movements were recorded by accelerometers (Actiwatch-L AW-7; MiniMitter, Bend, OR) placed on subjects' ankle and wrist of the dominant arm. The Actiwatch captured the highest activity or movement amplitude in each second and output the summation of counts every 15 s. The inflight raw data were downlinked and preprocessed by support personnel at Johnson Space Center.
Data collected during sleep were analyzed in this study. Data for mean HR and activity patterns at different times of the day have been described for these same subjects (18). The sleep period in each session was manually identified on the basis of markedly reduced activity level in that period. Then, a segment of steady data was taken from the sleep period for subsequent analysis (see Fig. 1). For each subject, the data length used for analysis was the same for all four test sessions (i.e., the shortest segment identified among the sessions) to diminish the influence of data length on HRV indices (48a). In this way, the data length varied between 1.5 and 4 h.
The raw ECG waveforms were first analyzed by Holter scanning software (Impresario; Del Mar Reynolds Medical, Irvine, CA and H-Scribe Rx; Mortara Instrument) to obtain beat-by-beat time series of R-R interval. The R-R interval series were inspected for inappropriate (e.g., selection of T wave as R wave) or missing (e.g., failure to detect the R wave) identification of heartbeats and filtered when necessary as previously described (18, 57). Briefly, ectopic intervals with absolute beat-to-beat HR variation >20 beats per minute (bpm) were identified and filtered by either adjusting or inserting heartbeats to the time series. The filtering process was not always necessary for the sleep data because of limited motion artifact during sleep. Indeed, only the late inflight record of one astronaut was substantially improved by the filtering. Then, indices of HR and HRV were calculated from each R-R interval series by linear time and frequency domain methods, and by complexity and fractal analysis, which were implemented in MATLAB (MathWorks, Natick, MA).
Time domain parameters included 1) mean HR; 2) standard deviation of the R-R intervals (SDRR) indicating the overall HRV; 3) standard deviation of the averaged R-R intervals over 5 min (SDARR), which estimates the slow components (i.e., low frequency) of HRV; and 4) square root of the mean squared differences of successive R-R intervals (RMSSD) representing the fast components (i.e., high frequency) of HRV (48a).
Frequency domain parameters were derived from the power spectral analysis of the R-R interval series. Four frequency bands were defined as recommended in (48a): ultralow frequency (ULF, ≤0.003 Hz); very low frequency (VLF, 0.003–0.04 Hz); low frequency (LF, 0.04–0.15 Hz); and high frequency (HF, 0.15–0.4 Hz). The R-R interval series was first resampled at 2 Hz by cubic spline interpolation. Then the power spectral density of the series was estimated via Welch's method (56). The time series was divided into four segments (Hanning windowed) with 80% overlap so that all the segments would have sufficient data length (≥3,375 s) to generate a ULF spectral component (45, 48a). Next, the respiratory frequency was estimated as the frequency corresponding to the HF peak in power spectral density. Finally, spectral powers in the four frequency bands (P_ULF, P_VLF, P_LF, and P_HF), the total power from all four frequency bands (P_TOT), and the ratio of LF to HF power (LF/HF) were calculated. Consistent with several previous studies (2, 7, 24, 45), large interindividual variance in spectral powers was observed. Therefore, spectral powers were normalized by the total power from all four test sessions (NP_ULF, NP_VLF, NP_LF, NP_HF, and NP_TOT) to minimize the intersubject variation while retaining the changes of HRV under different conditions. For example, NP_ULF was computed as follows: where session equals preflight, early inflight, late inflight, or postflight. The power spectral analysis was also performed on shorter data sets to assess the short-term modulation of HR. Specifically, each R-R interval series was divided into 5-min segments and the power spectral index was calculated for all the short segments, the median of which was taken as a representative value.
Complexity and fractal dynamics of HRV were assessed by sample entropy (SampEn) and the detrended fluctuation analysis (DFA), respectively. Sample entropy is a measure of complexity (i.e., irregularity) of a time series (30, 42). DFA is a technique to extract the pattern of signal amplitude with changing time scales (i.e., fractal dynamics) (3, 37, 42), with smaller scaling exponent α indicating higher unpredictability of the time series and vice versa. Both techniques have demonstrated abilities to reveal changes in HRV dynamics associated with various conditions such as aging, sleeping, pregnancy, and disease (3, 28, 29, 37, 53). In the current study, SampEn was computed for each R-R interval series with a pattern length of 2 data points and a tolerance threshold of 0.15, which was widely adopted in HRV studies (10, 31). Similar to power spectral analysis, short-term SampEn was also calculated from 5-min data segments using the same procedure. The DFA scaling exponent α (α-DFA) was estimated with maximum window size of one-fourth of the time series length (37).
Linear regression analysis was employed to evaluate the contribution of movements to HRV. Movement counts measured from ankle and wrist were first averaged to present the overall movement patterns during sleep. Then, the movement powers were calculated in ULF and VLF bands [LF and HF powers were not available because the movement series was sampled every 15 s (0.067 Hz)]. Briefly, the time period between each two adjacent movement events in the averaged movement series was first identified and converted to frequency. Next, the movement power at this specific frequency was calculated from the movement counts in the two adjacent events. In the end, all the obtained movement powers were integrated in ULF and VLF bands. The above method, rather than conventional power spectral analysis, was applied to cope with impulse train–shaped movement measurements during sleep. Finally, correlation coefficients (r) and P values between movement power and HRV were calculated in ULF and VLF bands pooled over all test sessions and all subjects.
Data were presented as means ± SD. One-way repeated measures ANOVA was applied to test differences across the four flight conditions (preflight, early inflight, late inflight, and postflight), followed by a Student-Newman-Keuls post hoc test. A value of P < 0.05 was considered significant. Statistical analyses were performed with SigmaPlot (Systat Software, San Jose, CA).
Inflight vs. preflight.
Inflight, mean HR was maintained at the preflight level as summarized along with all other data in Table 1. On the other hand, inflight HRV indices in time domain (SDRR, SDARR, and RMSSD) were reduced from preflight values by 33–42% on average. The reductions were significant except for SDARR during late inflight. Consistent with their time domain counterparts, inflight spectral powers were lower than the preflight values by 49–67% on average, although the ULF, VLF, and LF power reductions showed no statistical significance, likely due to large intersubject variations. Normalization of spectral powers, as expected, attenuated the standard deviations and revealed significant decrements in HRV in all frequency bands. No significant difference was found in LF/HF ratio, SampEn, or α-DFA from preflight to inflight.
Ankle movement was much lower during flight, whereas wrist movement remained unchanged compared with preflight values. The resulting inflight overall movement counts were significantly lower than those at the preflight level.
Postflight vs. preflight and inflight.
Postflight HR was greater than both preflight and inflight values by ∼5 bpm (Table 1). SDRR and SDARR recovered toward preflight levels upon return to Earth (P > 0.05 vs. preflight; P < 0.05 vs. inflight), whereas postflight RMSSD remained lower (P < 0.01 vs. preflight). The frequency domain postflight HRV indices showed similar patterns of recovery toward preflight values except for P_HF and NP_HF (P < 0.05 vs. preflight). The LF/HF ratio was slightly but not significantly elevated. SampEn and α-DFA postflight were not significantly different from either preflight or inflight values.
Upon return to Earth, the movement counts were higher than inflight values (P < 0.01). The wrist and overall movement levels after landing also exceeded preflight values (P < 0.05).
Movement and HRV.
Figure 2A illustrates time series of R-R interval and overall movements for preflight and inflight recordings from one astronaut and the R-R interval power spectra. Reductions in inflight HRV from preflight in all four frequency bands were noticeable. Figure 2A further reveals temporal agreement of events when movement counts increased and R-R interval decreased. Figure 2B shows the linear regression between movement power and HRV in ULF and VLF bands. Quantitatively, the regression analysis revealed moderate but significant correlation between movement power and HRV in both ULF (r = 0.63; P = 0.003) and VLF (r = 0.40; P = 0.03) bands, indicating a reasonable contribution of movement to slow fluctuations in HR.
Table 2 summarizes the short-term HRV indices derived from the median value of the 5-min data segments from each subject by power spectral analysis and sample entropy method. Although there were differences in the absolute values when comparing the short-term HRV values in Table 2 with the long-term HRV values in Table 1, the comparisons between the time points yielded essentially identical statistics as their counterparts in Table 1.
The novel finding of the current study was the marked reduction in time and frequency domain indicators of HRV while sleeping during spaceflight even though there was no change in mean HR. These results contrast with our hypothesis and with a previous investigation of cosmonauts sleeping on the Mir station when HR was lower but there was no change or a small increase in HF HRV (22). The results also differ from the recent report of no change in HR or HRV in volunteers confined for 105 days in a spaceflight simulation (54). The reductions in HRV power at ultralow and very low frequencies could be explained in part by parallel reductions in the patterns of movement during sleep at these same frequencies. In the low and high frequency bands for HRV the reductions in power during sleep were unlikely to be associated with movement; rather, other mechanisms associated with cardiovascular regulation need to be considered. Interestingly, the complexity and fractal dynamics of HR were unaltered by spaceflight. The spectral power indices in the low and high frequency bands and the SampEn results were further confirmed by short-term HRV analysis (see Tables 1 and 2).
As we have previously reported on different data segments from the same astronauts (18), mean HR in sleep was not different from preflight values during both early and late inflight testing. This was consistent with some previous research (19, 46) but it contrasts with the prolongation of R-R interval in Mir cosmonauts (22). Postflight (1–3 days after landing), HR during sleep was elevated by ∼5 bpm, suggesting that gravitational stimuli may affect the autonomic control system and invoke a tachycardic response even without orthostatic stress. More specifically, blood volume is reduced in space (9) and might not have recovered, whereas redistribution of blood flow occurs upon return to Earth even in a supine position (i.e., shift toward one side of the body). Either of these factors could have contributed to baroreceptor stimulation, reducing parasympathetic and/or enhancing sympathetic nerve activity to increase HR. The overall reduction in movement during sleep was not sufficient to alter the mean HR.
Inflight HRV in LF and HF.
Limited data during sleep in spaceflight have provided evidence of unchanged or a small increase in high-frequency HRV (19, 22). Also, recent data indicated no change in sleep time HRV during simulated spaceflight in the Mars 500 pilot study (54).
The HF component of HRV is considered to be related to parasympathetic nervous control on HR, whereas the LF variability is modulated by both sympathetic and parasympathetic nervous systems (38). During sleep, vagal activity is dominant in healthy humans (20, 51). In the present study, inflight mean HR was maintained at the preflight level. Thus the marked reduction of inflight HRV at LF and HF was more likely to reflect smaller fluctuations in HR regulatory systems rather than a resetting of tonic autonomic modulation level. The reduced and more evenly distributed blood volume in space is likely to cause smaller fluctuations in venous return and cardiac filling, and the resulting attenuation in blood pressure variability was translated into reduced HRV by the baroreflex regulatory system (4, 12). In these same astronauts studied under controlled conditions while they were awake, there was a significant reduction in HF HRV with no change in LF HRV (26). Coincident with the change in HF HRV were significant reductions in HF power for the estimated stroke volume and arterial pulse pressure, as well as nonsignificant reductions in LF power for these same variables. It is possible that during sleep similar changes in stroke volume and pulse pressure contributed to the reductions in both LF and HF HRV, but this cannot be confirmed because blood pressure was not measured.
HR fluctuations in LF and HF bands are also modulated by respiration (8, 15). Higher respiratory rate and smaller tidal volume are associated with decreased HRV in LF and HF with unchanged mean HR due to the kinetics of sinoatrial node responses to acetylcholine (8, 15). In the current study, inflight respiratory frequency was slightly but significantly higher than that at preflight (Table 1), which was consistent with previous observations of astronauts during sleep (44). Therefore, reduction of inflight HRV may also be attributed to higher respiratory rate. Indeed, linear regression analysis between respiratory rate and normalized HRV powers showed significant correlations in VLF (r = −0.46; P = 0.01), LF (r = −0.50; P = 0.007), and HF (r = −0.49; P = 0.008) bands. Although measurement or estimate of tidal volume was not available, recent studies have shown similar or reduced tidal volume in space compared with the preflight value measured in supine posture (40, 41, 44).
Postflight HRV in LF and HF.
In the first days after landing, LF and HF HRV powers during sleep tended to recover, but there was still a 19% reduction in LF and a 40% reduction in HF compared with preflight. The tendency of postflight recovery in LF and HF powers could be a result of combined effects from 1) restoring blood volume, 2) gravitational stimuli, 3) sustained higher respiratory rate (Table 1), and 4) possibly decreased postflight baroreflex sensitivity (26). The postflight elevation in mean HR indicated a shift in the balance between parasympathetic and sympathetic activation that affected HRV, which was not detected as a significant increase in postflight LF/HF ratio due to large intersubject variance. Indeed, the postflight LF/HF ratio was much larger than the preflight values in three subjects, marginally increased in two subjects, and lower in two subjects.
HRV in ULF and VLF.
Mechanisms causing ULF and VLF components in HRV are incompletely understood (6, 48a). VLF HRV is mainly effected by modulation of parasympathetic activity (49) and this is reflected by the dramatic reduction in HRV in these frequency bands by atropine (58). Variation in the mean level of parasympathetic activity is unlikely in the current study given the unchanged inflight HR. Additional mechanisms inducing HRV in the ULF and VLF bands are suggested to include renin-angiotensin-aldosterone system (1, 49), thermoregulation (17, 27), and physical activity (5, 36, 43, 45).
Elevated levels of the renin-angiotensin-aldosterone system during spaceflight (14) might be associated with reduced HRV in the VLF frequency band in the current study. It is unlikely that thermoregulatory effects contributed during the 1.5–4 h data segments.
Physical activity affects long-term fluctuations in HR under daily routines (5, 36, 43, 45). Although movement was limited during sleep, our results revealed moderate correlations between overall movements and HRV in both ULF and VLF.
Complexity and fractal dynamics of HRV.
The finding of no difference in α-DFA between preflight and inflight values was similar to the observation by Ivanov et al. who compared healthy subjects during sleep on Earth with astronauts during sleep in orbit (28). There was also no significant difference in the SampEn with spaceflight. The overall observations in the current study suggest that the complexity and fractal properties of HRV were not changed by spaceflight in contrast to the observed changes in the linear components. These results are consistent with the hypothesis presented by Yamamoto and colleagues that the linear and fractal mechanisms regulating HRV are distinct (58). Postflight, SampEn and α-DFA were almost identical to their preflight values, indicating stability in this aspect of cardiovascular control, even though there were small but significant increases in HR and reduced HF HRV compared with preflight.
One major limitation of the study was the lack of polysomnography recordings. Sleep structure has been documented to change during spaceflight (13, 23, 34). We noted reduced inflight movements during sleep that might be associated with reduced sleep-disordered breathing (16). Reported impairment of sleep quality during spaceflight could be associated with factors such as variable light-dark cycle, altered circadian rhythm, microgravity, confinement, and workload (13, 23, 34). The potential of better sleep on the basis of observations of reduced movement and HRV in the current study could be linked to regular exercise as a countermeasure that facilitates the adaptation of circadian rhythmicity (34). Nevertheless, without polysomnography it is likely that our data crossed sleep stages, which could alter HRV (48, 50, 53). To test for potential impact of sleep phase, the same time and frequency domain analysis methods were applied on the first 1.5 h of data during sleep (presumably within the first sleep cycle). The resulting HR and HRV indices were similar to those in Table 1 (results not shown), suggesting that the effect of microgravity on HRV was probably predominant over that of different sleep stages. However, incorporating polysomnography into studies of sleep structure effect on HRV during spaceflight is warranted in future studies.
The second limitation of the study was that the power spectral analysis assumes data stationarity, which is normally not the case for R-R interval time series (32, 39). Thus, a test for weak stationarity was implemented (39). Each R-R interval series was divided into segments with a length of 300 samples. From these, a subset was randomly selected for short-term stationarity testing. This procedure was repeated for 500 iterations and the averaged percentage of stationary segments was calculated to be 29.8 ± 9.4%, 29.4 ± 9.1%, 31.9 ± 8.7%, and 25.7 ± 7.3 for series during preflight, early inflight, late inflight, and postflight sessions, respectively, indicating short-term nonstationarities of the R-R interval series during sleep. Even smaller percentage stationarity would be expected with longer data and data collected during daytime. Although power spectral analysis is the standard method in HRV studies, methods suitable for nonstationary series (e.g., wavelet transform analysis) should be considered in HRV analysis.
The scheduling of postflight tests (1–3 days after landing) was set to investigate the immediate effect of the restoration of gravity on cardiovascular function but does include stressful readaptation. Additional tests following a recovery period after landing would be helpful to assess the cardiovascular adaption after return to Earth. The small sample size in the current study was another limitation that may diminish the statistical power of the data. Finally, the movement data were recorded by the accelerometer and sampled at a low rate (15 s per sample). Therefore, activities in LF and HF bands were not accessible. Fortunately, movements are commonly related to ULF and VLF powers of HRV, and the sampling rate of 1/15 Hz is sufficient to provide information in these two frequency bands.
Perspectives and Significance
The effects of long-duration spaceflight on the cardiovascular regulatory systems were assessed in terms of HR and HRV. The current results showed that unequivocal alterations occurred in the linear, but not the complexity and fractal, indicators of cardiac rhythm control mechanisms during sleep in astronauts living on the ISS up to 6 mo. The wide-scale impact on linear HRV from the ULF to the HF regions of the power spectrum compared with data on Earth suggest multiple rhythms are altered by spaceflight, including altered sleep pattern affecting movements, oscillations in the renin-angiotensin-aldosterone system, and gravitational effects on blood volume distribution in the body and within a respiratory cycle. These results provide new insights into the adaptations of the human cardiovascular regulatory system under microgravity and suggest that future studies need to include quantitative measurements of sleep stage and detailed investigations of mechanisms proposed for the alterations in HRV.
This work was supported by Canadian Space Agency Grant 9F007-02-0213, and by the Space Science Enhancement Program. D. Xu was supported by the Ontario Ministry of Research and Innovation.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: D.X., K.S.F., and R.L.H. analyzed data; D.X., J.K.S., A.P.B., P.A., K.S.F., and R.L.H. interpreted results of experiments; D.X. and R.L.H. prepared figures; D.X., A.P.B., and R.L.H. drafted manuscript; D.X., J.K.S., A.P.B., P.A., K.S.F., and R.L.H. edited and revised manuscript; D.X., J.K.S., A.P.B., P.A., K.S.F., and R.L.H. approved final version of manuscript; J.K.S., A.P.B., and R.L.H. conception and design of research; J.K.S., A.P.B., P.A., K.S.F., and R.L.H. performed experiments.
We thank the astronauts for their dedication to the project. Danielle Greaves provided excellent support as the project manager. Logistical support was provided by the Canadian Space Agency and National Aeronautics and Space Administration personnel in the Cardiovascular Laboratory at Johnson Space Center, Kennedy Space Center, Dryden Flight Research Center, and Gagarin Cosmonaut Training Center.
- Copyright © 2013 the American Physiological Society