Ambulatory around-the-clock 24-hour ECG records were obtained by using a two-channel Holter recorder (SM-28, Fukuda-Denshi, Tokyo) and an analyzing system (SCM-280-3, Fukuda-Denshi, Tokyo), including an interval counter of RR intervals and a built-in A/D converter. The interval resolution was 8 msec.

For the analysis of a trend with development, 19 infants (aged from 25 days to 3 months of age), 22 children (aged 3–9 years), 18 boys and girls (aged 10–14 years), 14 pubertal boys (aged 15–20 years), and 19 young men (aged from 21 to 29 years) were examined.

The mapping of circadian changes in HRV endpoints as a function of age and development was performed in Tokyo on 285 males (aged from 25 days to 92 years) and 117 women (aged 3–91 years), who were randomly selected from a larger group of clinically healthy volunteers. The clinical health status of all subjects investigated was assessed by means of a physical medical examination and laboratory data, including blood and urine examinations, chest X-ray, and ECGs at rest. Adult subjects were found to be normotensive by WHO criteria (SBP and DBP below 140 and 90 mmHg, with no history of major diseases).

Frequency-domain measures were obtained with the MemCalc/CHIRAM (Suwa Trust Co., Ltd., Tokyo) software. Time series of NN intervals covering 5 min were processed consecutively, and the spectral power in different frequency regions was computed, namely, in the “very-low-frequency” (“VLF-1,” 0.003–0.04 Hz, “VLF-2,” 0.005–0.04 Hz, SAS (90-sec)), “low-frequency” (spectral power centered around 10.5 sec: “LF-1,” 0.04–0.15 Hz, “LF-2,” 0.05–0.07 Hz, MSNA (15-sec)), and “high-frequency” (spectral power centered around 3.6 sec: “HF,” 0.15–0.40 Hz) regions of the MEM spectrum (Fig. 5.1). Next, 180-min NN interval time series were processed consecutively, progressively displaced by 5 min, to estimate the “total spectral frequency” (“TF,” 0.0001–0.50 Hz), the “ultralow-frequency” (“ULF-1,” 0.0001–0.003 Hz, “ULF-2,” 0.0001–0.0003 Hz, circacentuminutan or Kleitman (90 min), “ULF-3,” 0.0003–0.005 Hz “extremely low frequency” (ELF) (30 min)) components, and the slope of 1/f fluctuations of NN intervals (“β (beta)”), in the frequency range of 0.0001–0.01 Hz.

The dynamics of HRV measures, assessed every 5 min, were analyzed by both the Maximum Entropy Method (MEM) and by cosinor, involving the least-squares fit of cosine curves with anticipated periods. Spectral components of each MEM spectrum were expressed not only in terms of spectral power or amplitude but also as power ratio (unit, %), normalized in relation to the “total spectral power” (dividing by “TF”). Circadians, circasemidians, and circaoctohorans were defined in terms of periods ranging from 20 to 28 hours, from 10 to 14 hours, and from 6.5 to 9.5 hours, respectively. Each component was assumed to be present (detectable) when the power ratio of these three chronome components was equal to or greater than 10 %.

For assessing any trend with age, time-domain measures, SDNN, pNN50, SDANNs (1, 5, 30, and 60 min), SDNNIDXs (1, 5, 30, and 60 min), r-MSSD, NN50(+), NN50(−), total NN50, and pNN50 were computed. The mean cycle length of the normal-normal RR (NN) intervals over 24 hours and the standard deviation (SD) of NN intervals over 24 hours were calculated as 24-hour NN and SDNN, respectively. SDANNs (1, 5, 30, and 60 min) were calculated as the SD of the 1-min, 5-min, 30-min, and 60-min mean NN intervals over the entire 24-hour ECG recordings, respectively. Similarly, SDNNIDXs (1, 5, 30, and 60 min) were calculated as the mean of the SD of the 1-min, 5-min, 30-min, and 60-min NN intervals over 24 hours, respectively. The difference between adjacent NN intervals was also computed as the root mean square of successive differences (r-MSSD), calculated as the square root of the mean of the sum of the squares of differences between adjacent NN intervals over the entire 24-hour record. NN50 was assessed as follows: successive NN intervals were subtracted from each other and compared with 50 msec. If the difference was larger or smaller than 50 msec, the positive or negative count was incremented by one, with the totals over 24 hours being represented by NN50(+) and NN50(−), respectively. NN50 and pNN50 represent the total count and the percentage of differences between adjacent NN intervals that are greater than 50 msec, computed over the entire 24-hour ECG recording. Frequency-domain measures were determined as follows: The time series of NN intervals were reconstructed using a third-degree spline interpolation method to yield a new smoothed instantaneous time series sampled at 4 Hz. A 10-min span of the reconstructed time series was processed consecutively by the fast Fourier transform (FFT), with successive 10-min spans overlapping by 150 sec. Spectra obtained within a 60-min span were averaged, and the “LF,” “HF,” and the ~10.5 sec/~3.6 sec spectral power ratio (“LF/HF”) were determined.

HRV endpoints were found to be eminently circadian periodic in the 359 profiles previously analyzed (from subjects 10–92 years of age). The spectral power centered around 3.6 sec (“HF” = high frequency) peaks during the night, and the ~10.5 sec/~3.6 sec (“LF = low frequency/HF”) power ratio peaks during the day. The 24-hour component is the most prominent one in 292 profiles (81.3 %) out of the 359 around-the-clock 7-day ECG records analyzed. Circadian variation of HF and LF/HF ratio is shown in Fig. 5.2, measured in 218 healthy males aged from 3 to 92 years. The second most frequently observed component is the 8-hour variation, found to be statistically significant in 14 profiles (3.9 %), and the third one is the 12-hour component, detected in 13 profiles (3.6 %).

HRV serves not only as a noninvasive measure of autonomic nervous system activity but also as an index predicting long-term survival in patients with coronary artery disease (CAD). Here we introduce “chronodesms” (Fig. 5.3), that is, time-specified reference values that account for the circadian variation in HRV endpoints. To illustrate their qualified usefulness and to show the need for refinement by outcomes, circadian profiles of CAD patients are interpreted in the light of chronodesms derived from healthy peer groups, revealing nearly deviant features in the presence of disease.

The finding of an 8-hour component in circulating endothelin-1 [13, 14] and also in the population density of endotheliocytes [15] is one of the noteworthy observations in recent years, notably since the hypothesis was made that an 8-hour periodicity may be a key between the linear and nonlinear systems of life, as proposed by Li and Yorke in 1975 [16]. Such mechanisms may reside in endotheliocytes, since circulating cortisol did not show an about-8-hour (circaoctohoran) component [13]. As noted above, another recent finding relates to the slope (β (beta)) over the two-decade band from 0.0001 to 0.01 Hz, known as 1/f behavior. Several authors have proposed that this slope is a better predictor of all-cause mortality than the traditional power spectral bands [8–10].

The development of the chronome of different endpoints of HRV, notably of the circadian, circasemidian, and circaoctohoran components, was investigated. Results are summarized in Table 5.1 in terms of the incidence of detection of each of these three components for six different frequency-domain measures of HRV (1/f slope, “TF,” “ULF,” “VLF,” “LF,” and “HF”). The incidence of detection was compared among the following five age groups: 25 days to 3 months (n = 19), 3–9 years (n = 22), 10–14 years (n = 18), 15–20 years (n = 14), and 21–29 years (n = 19). The incidence of detection of the circadian component for 1/f changes with age (21.05, 27.27, 61.11, 57.14, and 31.58 %, respectively, p < 0.05) was mostly present around puberty. Averaged circadian profiles of the 1/f slope are shown in Fig. 5.4. Not shown in Table 5.1 is an increase with age in the power ratio of the circadian spectrum: 40.8, 39.8, 41.7, 56.1, and 51.1 %, respectively (p < 0.05). No statistically significant changes in the circadian period, assessed by MEM spectral analysis, were found as a function of age: 24.5, 21.2, 22.6, 20.9, and 26.0 hours, respectively.

Table 5.1 suggests that the about 35 % circadian rhythm detection for “TF” and “ULF” does not change appreciably with age. By contrast, the detectability of a 24-hour component for “VLF,” “LF,” and “HF” is much lower early in life than after 3 years of age. Averaged circadian profiles of “HF” are shown in Fig. 5.5. Similar changes with age are not seen for the circaoctohoran component, however. For 1/f, the incidence of detection is 52.63, 54.55, 38.89, 50.00, and 57.89 %, respectively (N.S.), and for the power ratio of the MEM spectrum, it is 27.7, 27.1, 24.9, 26.3, and 22.9 %, respectively (N.S.). Results for the circasemidian component show an increase with age only for “HF” and may reflect the development of the circadian system. Also apparent from Table 5.1 is the larger detection of a circaoctohoran component for 1/f, “TF,” and “ULF” as compared to “VLF,” “LF,” and “HF,” reflecting the autonomic nervous activity.

Changes with age of circadian components in the spectral element of HRV chronomics are highly statistically significant (Fig. 5.6) in both genders. The relation is not invariably linear, as indicated by the presence of higher-order polynomial terms selected by stepwise regression, using a sixth-order model at the outset. For instance, NN50, SDNNIDX (1-min), and “HF” reach a minimum around 66.5 years (male: 65, female: 68), 72 years (male: 70, female: 74), and 59.5 years (male: 61, female: 58), respectively, and increase thereafter, as shown in Fig. 5.7, raising the question whether the oldest age group represents a category of long-lived subjects, who by this status may have been healthier than the average subject in the younger age groups. Thus, circadian elements in HRV chronomics, obtained with the MemCalc/CHIRAM (Suwa Trust Co., Ltd., Tokyo) software, are shown to vary with age in Appendices 1, 2, 3, and 4, which are shown as an average of 24 hours, sleeping and waking time spans.

Using a linear model with age as a first approximation, gender differences are found for the slope but not for the intercept of r-MSSD, NN50, pNN50, and HF, the females showing a slower decline with age as compared to males (p < 0.05). By 1-way ANOVA, using age as covariate, females are found to have higher values than males for r-MSSD, NN50, pNN50, and HF (p < 0.05) (Fig. 5.8, shown as not NN50, but RR50).

Circadian chronodesms of “HF” and LF/HF” for different age groups of healthy males are shown in Fig. 5.9. They were computed as 90 % prediction limits on log₁₀-transformed data, with the mean values and lower and upper limits of variability back-transformed into original units for plotting. These chronodesms show that both HF and LF/HF vary greatly as a function of age.

In the chronodesms for “HF” (Fig. 5.9, top), the vertical scales are adjusted for each age group separately, so as to optimize the viewing of the reference limits in each group. The results were further summarized by population-mean cosinor in different age groups. A circadian rhythm was invariably demonstrated (p < 0.05) with a peak at night (around 03:30) for HF. The MESOR and circadian amplitude of HF were linearly regressed with age. A statistically significant decrease as a function of age was found for the MESOR (all: r = −0.590, p < 0.001; 20–60 years: r = −0.495, p < 0.001) and for the circadian amplitude (all: r = −0.530, p < 0.001; 20–60 years: r = −0.474, p < 0.001).

In the chronodesms for “LF/HF” (Fig. 5.9, bottom), the vertical scales are kept the same, so as to facilitate the comparison of the circadian profiles from one age group to another. Lack of statistical significance tended to occur in the older age groups.

The presence of a circaoctohoran component of variation, proposed to be a key between the linear and nonlinear systems of life by Li and York in 1975 [16], was found not to differ among the five age groups considered above for the case of HRV indices reflecting the autonomic nervous system activity (“LF” and “HF” components). It was found to be already present in a sizeable proportion of infants for the case of HRV indices reflecting the endocrine or immune system function (“TF,” “ULF,” “VLF,” and 1/f fluctuations). By comparison, the circadian component showed a more pronounced development as a function of age.

As noted above, fractal analysis methods have provided clinically useful information. In particular, the scaling exponent α, calculated with detrended fluctuation analysis (DFA), is available to clinical cardiology. Circadian variation of the two scaling components (α₁; <11 beats, and α₂; >11 beats) describing fractal properties of the HRV is shown as a comparison between healthy young subjects and elderly people in Appendix 5.

In clinical health, HRV is characterized by a broad time structure, including not only a prominent circadian and infradian variation, including infra-annuals such as circadecadals, but also statistically significant circaoctohoran and circasemidian components, in addition to trends as a function of age. Previously, we have reported that a circadian rhythm is demonstrated in both genders and in different age groups (<10 years, 10–20 years, 20–30 years, 30–40 years, 40–50 years, 50–60 years, 60–70 years, and >70 years) except for the oldest age group. The “HF” component may reflect the parasympathetic activity with a peak during the night. A strong circadian component is also shown for the “LF/HF” spectral power ratio, which is considered to be an indicator of sympatho-vagal interaction, and is characterized by a peak during the day.

A circaseptan component is also demonstrated for most of the HRV indices, including SDNN, observed both in middle aged and in elderly subjects. SDNN is known to be a good predictor of long-term survival, both in patients with myocardial infarction and in a cohort of elderly subjects. A meta-analysis of 47 studies shows the incidence of myocardial infarction to be increased on Mondays with a secondary peak on Thursdays and Fridays [2]. The circaseptan pattern of SDNN for the middle-aged subjects is characterized by a decrease on Mondays, with a secondary trough on Fridays, and may be at least one of the reasons underlying the circaseptan pattern of incidence of cardiovascular events. An important aspect of the circaseptan variability may be the relative prominence of the circasemiseptan (3.5-day) component, which could account for the secondary peak later in the week.

A circannual variation in “LF” and “HF” components has been observed in elderly subjects living in Shandong province, China, where the environmental temperature is hot in summer (up to 40 ° C) and cold in winter (as low as −15 ° C). A physiological interpretation for “LF” still needs to be clarified. “LF” fluctuations reflect combined sympathetic and parasympathetic activity. The “LF” peak is strikingly reduced by atropine. An increase in its power has been observed as a consequence of sympathetic activation. Thus, an increase in “LF” is widely viewed as a marker of sympathetic activation and vice versa. De Boer et al. [17] suggested that the “LF” peak was caused by the rather slow baroreflex response at 2–3 seconds in the beat-to-beat changes in arterial pressure. Thus, the decrease in “LF” in the winter may suggest an increase in arterial pressure related to an alteration in environmental temperature. It is noteworthy that β (the slope of the 1/f fluctuations of RR intervals), noted as one of the most important predictors of adverse cardiovascular outcomes, is characterized by a circannual component. “β (beta)” was most negative in the summer, when an 83-year-old man experienced a minor stroke. Nicolau et al. [18] investigated circannual variations in cardiac mortality and reported a peak in July with a broader peak during the cold season. Previously, Smolensky et al. [19] and Reinberg et al. [20] had also reported a circannual pattern in the incidence of cardiovascular events.

We also have shown previously that a relation exists between HRV endpoints and latitude, suggesting that there may be an effect of sunshine duration on the chronome mechanism that affects HRV measurements [21]. These results may also be interpreted as an association between environmental factors and human beings who have adapted to the surrounding cycles in the cosmos.

In 2000, we began a community-based study named “Longitudinal Investigation for Longevity and Aging in Hokkaido County (LILAC study)” and monitored 7-day/24-hour ambulatory BP of middle-aged subjects, aged 40–74 years, while also evaluating the cardiovascular and neurobehavioral functions of elderly people. Our goal was the prevention of stroke and myocardial infarction and of the decline in cognitive function of the elderly in a community.

CHAT (short for circadian hyper-amplitude-tension), a condition defined by an excessive circadian amplitude of blood pressure (BP), above a threshold approximated by the upper 95 % prediction limit of clinically healthy peers matched by gender, age, and ethnicity, is associated with a large increase in vascular disease risk, cerebral ischemic events, and nephropathy in particular [1, 11, 22–32]. We have already monitored 7-day/24-hour ambulatory BP of 218 middle-aged citizens in Hokkaido County from April 2001 to April 2003. We observed circaseptan components for the average systolic (S) BP value during the waking span, for the 24-hour average of HR, for the standard deviation (SD) of HR, and for the day-night (dipping) ratio of SBP. Specifically, SBP during waking was lowest on Sunday. The BP average during waking shows a circaseptan variation. Thus, we point out that ambulatory BP monitoring limited to 24 hours is not sufficient and propose a routine 7-day screen, which is continued when abnormality is found. Among the 218 records of 7-day/24-hours ambulatory BP, we also observed 32 (14.7 %) and 11 (5.0 %) cases of SBP and diastolic (D) BP CHAT, respectively. We have started several kinds of interventions, including lifestyle guidance, education from the viewpoint of clinical nutrition, and consultation of drugs for hypertension and hyperlipidemia, among other conditions.

We also determined in the longitudinal community-based study (LILAC) whether BP, HR, and HRV are predictors of cognitive function in the elderly. We initially examined 115 people older than 75 years (average, 79.6 years). BP was measured at the beginning of the study in a sitting position, and pulse wave velocity (PWV) was measured between the right arm and ankle in a supine position, using an ABI/Form instrument (Nippon Colin Co., Ltd., Komaki, Japan). The first 1-hour of ambulatory ECG recording was obtained during routine medical examination conducted each year in July. The data were processed for HRV using a Fukuda-Denshi Holter analysis system (SCM-280-3). Time-domain measures (SDNN, pNN50, SDANN, and Lorenz plot indices: length (L), width (W), and L/W ratio) and frequency-domain measures (spectral power in the “very low frequency” (VLF), 0.003–0.04 Hz; “low frequency” (LF), 0.04–0.15 Hz; and “high frequency” (HF), 0.15–0.40 Hz regions and the “LF/HF” ratio) were determined. Except for SDNN and HR, calculated over the whole 1-hour record, all indices were computed as averages over consecutive 5-min intervals. Spectral indices were obtained by the Maximum Entropy Method (MEM) with the MemCalc/CHIRAM program (Suwa Trust Co., Ltd., Tokyo, Japan). Using as reference the data obtained in July 2000, the cardiovascular coordination function of each participant was scored as 3, 2, or 1 point for each of the following three indices (SBP, HR, and “VLF” component of the HRV): SBP > 160, 140–159, or < 140 mmHg; HR > 80, 70–79, or < 70 beats/min; and “VLF” < 800, 800–1000, or > 1000 msec². Participants were classified into either the normal, mildly disordered, or disordered group when the sum of these indices was ≤ 4, 5 or 6, or ≥7, respectively.

The Japanese version of the Mini-Mental State Examination (MMSE) and the Hasegawa Dementia Scale Revised (HDSR) were used to measure the overall cognitive function, including verbal orientation, memory, and constructional ability. The Up & Go test measured (in seconds) the time it took the subject to stand up from a chair, walk a distance of 3 meters, turn, walk back to the chair, and sit down again. This test is a simple measure of physical mobility and demonstrates the subject’s balance, gait speed, and functional ability (Up & Go). A lower time score indicates better physical mobility. Functional reach, used to evaluate balance, represents the maximal distance a subject can reach forward beyond arm’s length while maintaining a fixed base of support in the standing position. A higher score indicates better balance. Manual dexterity was assessed using a panel with combinations of ten hooks, ten big buttons, and five small buttons. There were three discrete measurements of time recorded for each participant (ten “hook-ons,” ten big “button-on-and-offs,” and five small “button-on-and-offs”). Total manual dexterity time in seconds, defined as the button score (Button-S), was calculated by adding the average times for one hook-on and one big or small button-on-and-off. A lower button score indicates better manual dexterity.

In 2000 [33], the cardiovascular coordination score did not correlate with any index of neurobehavioral function, although it showed a negative correlation with SDNN, SDANN, pNN50, “LF,” or “HF” components (p < 0.0001) and a positive correlation with PWV (p < 0.01). We were able to follow up 72 of the 115 subjects. We found that between 2000 and 2002 the cognitive function, estimated by MMSE and HDSR, was maintained or improved as follows: In the cardiovascular coordination disordered group, MMSE and HDSR improved from 24.6 to 26.0 (p = 0.06) and from 23.8 to 25.9 (p = 0.04), respectively. In the mildly disordered group, these indices improved from 23.4 to 25.7 (p = 0.005) and from 23.4 to 25.1 (N.S.), respectively. In the normal cardiovascular coordination group, MMSE and HDSR were maintained from 25.6 to 26.0 (N.S.) and from 24.9 to 26.4 (N.S.), respectively. There were no statistically significant alterations in activities of daily living (ADL), assessed by Up & Go, functional reach, and button score, in any of the groups.

Although a cross-sectional study did not show any apparent correlation between cardiovascular and neurobehavioral functions in subjects 75 years of age or older, an intervention aimed at preventing stroke and a decline in cognitive function in an elderly community population induced an improvement of the cognitive function, especially in people suffering from hypertension, tachycardia, or decreased HRV. This study demonstrates a positive impact of a simple social intervention, including advising on the implementation of a medical prescription, in improving a disordered cognitive function in elderly people. People with a disordered coordination of cardiovascular function are more sensitive to such an intervention, suggesting that the cardiovascular function is a major factor affecting cognitive function.

In clinical health, both HR and BP variabilities (HRV and BPV) are characterized by a broad time structure that includes the prominent circadians and also ultradian (notably about 8 hours and about 12 hours) and infradian (notably about-weekly, about-yearly, and about-10-yearly) changes, in addition to undergoing trends with growth, maturation, and aging. Alterations in these HRV and BPV chronomes are observed in the presence of disease and at times of increasing magnetic activity. We postulate that the physiological chronomes, such as those of HRV endpoints, have counterparts in our environment and that our genetic makeup in time may have evolved to adapt to and integrate with our cosmos. Future work should focus on how phenomena in the cosmos, including helio- and geomagnetics, can affect physiological chronomes, those of the HRV in particular.

Human beings have shielded themselves from the unkind features of obvious extremes of weather and climate by building houses and have developed means, e.g., sulfonamides and antibiotics, to deal with at least some of the microbial contagions confronting us. These infections via the mutation of microorganisms causing them and by (steroidal?) changes in the host may be accelerated by solar activity. This is one more reason to explore the possibility of feedsidewards underlying phase response curves along genetically anchored built-in cycles from “then” (a billion years of evolution ago) as well as “now.”

In conclusion, there is a need for a transdisciplinary combined archival, physiological, and demographic, always inferential statistical methodology in order to resolve multifactorial interactions underlying “contagions” of the body and of the mind, which later may also be under the nonphotic influence of the sun or the galaxies, at least insofar as their cyclicity is concerned. They may lead to a heightened susceptibility of the masses to demagogues as well as to the demagogues themselves. We face the task of a scientific chronobioethic that should complement chronomics and chronobiology in everybody’s service. It may not be coincidence that circadecadal changes in religious motivation [34] and homicide [4] are nearly in antiphase (200° apart). When religiosity reigns from crime one abstains (religio reget, homicidium subsidet).