Pediat. Res. 4: 63-70 (1970)
Age
circadian rhythms
blood pressure
respiration
body temperature
A Study of the Development of Human Circadian
Periodicity
G.T.BRYAN [ 1 6 ] and J.E. OVERALL
Division of Endocrinology, Department of Pediatrics, The Clinical Study Center,
and the Department of Neurology and Psychiatry, University of Texas Medical Branch, Galveston, Texas, USA
Extract
Circadian periodicity of bcdy temperature, pulse rate, respiratory rate, systolic, and diastolic blood
pressure has been studied in 143 patients. The ages of the patients varied from 3 months to 21 years.
A type of power spectrum analysis was used to reduce the data to a single quantity for each of the
five measurements in each patient. This quantity was examined by various methods of statistical
analysis such as correlation, regression, and analysis of variance. Evidence from these analyses suggested that a single circadian periodicity indicator could be calculated for each patient by combining
the variance in circadian periodicity for temperature, pulse, respiration, and systolic blood pressure.
Analysis of this indicator, 'composite circadian periodicity', revealed highly significant correlations
with age, mean levels of the physiological measurements, and duration of the study. The relation
with age was best defined by a third-degree polynomial equation. The smoothed wave that was
derived from the data demonstrated a peak in composite circadian pericdicity at 6 years of age, with
a tendency to form a plateau at about 16 years of age.
Speculation
A rational extension of this study would be to apply the data to the investigation of human disease.
Although 'normal ranges' for age are published for temperature, pulse rate, respiratory rate, and
blood pressure [1, 8], many physicians modify these ranges intuitively (lowered body temperature
during the early morning hours, higher temperatures during the late afternoon and early evening).
It would seem appropriate to develop normograms for these functions that incorporate the changes
in mean level that occur with age, sex, and time of the day. In this way, the increased periodicity
demonstrated in the child of 5 or 6 years of age, as well as the decreased periodicity in the younger
and older subjects, would be accounted for in the normograms. The practicality of these techniques
could be tested in computer-oriented diagnostic evaluations.
Introduction
The presence of circadian periodicity [12] in the
mature human subject has been demonstrated repeatedly. Studies of these human rhythms were
summarized recently by MILLS [7]; however, the span
of life from conception in utero to adulthood has been
a difficult time to encompass and, therefore, the number of studies concerned with the developmental aspects
of human circadian periodicity are few. Significant
64
BRYAN and OVERALL
contributions to this problem have been presented by
HELLBRUGGE and his collaborators [2, 3], JUNDELL
50
[5], IRWIN [4], and PARMEIXE et al. [10]. HELLBRUGGE
et al. [3] have suggested that the amplitude of oscillation of various functions tends to increase during infancy and early childhood, and that the onset of each
circadian periodic function was independent of other
functions and dependent upon the maturity of the
subject. We attempted to define the problem by proposing that the amount of periodicity observed in the
adult human subject was the result of a gradual, systematic, and measurable increase in certain periodic
phenomena with a definable order of progression. The
studies presented in this paper were designed to test
whether or not there were gradual and measurable
increases in circadian periodicity of temperature, pulse
rate, respiratory rate, and blood pressure. A series of
analyses are presented which permit quantitation of
circadian periodicity and an evaluation of variables
which contribute to this quantity. Furthermore, a
statistical model which demonstrates circadian changes
during maturation is presented.
1 30
120
0-36
37-72
73-108 109-144 145-180 181-216 217-252 253-268
Age (months)
Fig. 1. Histogram showing distribution of ages of
patients.
Duration of hospitalization
60
50
J30H
Methods and Materials
Patients
This report concerns 143 subjects whose ages varied
from 3 months to 21 years with a peak in distribution
at 9-12 years of age (fig. 1). They represented many
different diagnostic categories (vide infra). Hospitalization for each patient varied from 7 to 117 days. A histogram showing the distribution of the duration of hospitalization is shown infigure2.
Body temperature, pulse rate, respiratory rate,
systolic, and diastolic blood pressure were obtained
by trained nursing personnel at intervals of precisely
6 h throughout the period of hospitalization. The temperature was measured to the nearest 0.01° with a
mercury thermometer which remained in the mouth
or anal canal for 3 min. The same thermometer was
used for the same subject throughout his period of
hospitalization. Pulse and respiratory rates were obtained by counting each for 1 min. Blood pressure was
obtained by the indirect auscultatory method using a
mercury manometer with the cuff size adjusted to a
width of one-half the circumference of the patient's
arm. All measurements were obtained after the patient
had been resting for 15 min.
Data Analysis
The primary purpose of the present investigation
was to examine the relation of circadian periodicity to
age. To accomplish this, a quantitative index of circadian periodicity was desired. We chose to employ
a form of power spectrum analysis from which esti-
I2°|io-
7-22
23-37 38-52 53-67
68-82 83-97 98-112 113-127
Total number of days
Fig. 2. Histogram showing distribution of duration of hospitalization for each patient.
mates of the proportion of total variation at each harmonic frequency could be obtained. It should be
stressed that the power spectrum analysis was employed solely as a data reduction technique. From numerous measurements taken atfixedintervals across time,
the power spectrum analysis yielded an index of the
relative prominence of circadian periodicity. The computer program used for this data reduction regressed
each complex waveform (e.g., pulse rate over time)
on orthogonal sine-cosine waveforms of various harmonic frequencies. The index of circadian periodicity
derived from this analytic technique represented the
proportion of total variability in each clinical measure
that could be accounted for by regression on orthogonal
sine-cosine waveforms having a frequency of 24 h.
After quantitative indices of the degree of circadian
periodicity were computed for each patient, their relation to age and other clinical variables were studied
by correlation, regression, and analysis of variance
methods.
Development of human circadian periodicity
Table I. Fractional variance in periodicity of temperature measurements of one patient
Results
An analysis was obtained on each set of measurements;
temperature, pulse, respiration, systolic, and diastolic
blood pressure. In utilizing the method of harmonic
analysis, the total variance of all observations for a
given function (e.g., temperature) was examined;
then, the proportion of the variance that could be
accounted for by a sine-cosine waveform of a given
frequency was determined. The duration of this waveform (its frequency) was examined in successive decrements, with the longest period coinciding with the
total duration of observation, and the shortest period
being less than 24 h. In this way, each harmonic was
independent of the other harmonics in the series, and
variance for each harmonic could be determined. An
example of the resulting computer output is seen in
table I and graphically displayed in figure 3. In this
example, the circadian variation accounted for 0.4723
of the total variance. Thus, the fraction of total variance accounted for by a recurring cycle of 24 h becomes a quantitative measurement of the circadian
activity for a given variable, and can be studied in
relation to other variables.
Correlations between the circadian periodicity of
1) temperature, 2) pulse, 3) respiration, 4) systolic
blood pressure, and 5) diastolic blood pressure were
then examined (i.e., 1 vs. 2, 1 vs. 3, 1 vs. 4,1 vs. 5,2 vs. 3,
2 vs. 4,2 vs. 5, 3 vs. 4, 3 vs. 5,4 vs. 5). It was found that
significant positive correlations (P<0.05) existed between the circadian periodicity of temperature, pulse,
50-
65
Temperature
40*
iati
g30-
Cycles/week
Fraction of variance
0.18
0.37
0.56
0.75
0.94
0.0096
0.0114
0.0003
0.0197
0.0057
1.13
1.32
1.51
1.70
1.89
0.0007
0.0005
0.0054
0.0125
0.0050
2.08
2.27
2.45
2.64
2.83
0.0038
0.0076
0.0023
0.0039
0.0000
3.02
3.21
3.40
3.59
3.78
3.97
0.0054
0.0005
0.0028
0.0209
0.0108
0.0030
4.16
4.35
4.54
4.72
4.91
0.0013
0.0001
0.0011
0.0079
0.0003
5.10
5.29
5.48
5.67
5.86
0.0102
0.0027
0.0048
0.0004
0.0044
6.05
6.24
6.43
6.62
6.81
6.99
0.0045
0.0206
0.0042
0.0032
0.0065
0.47231
7.18
0.0088
Total
0.6851
Residual
0.3149
>
tot
20fd
s= 10-
Cycles/week.251.0 2.0
Duratbn(h) 67216884
4.0
42
6.0 7.0 8.0
28 24 21
* Percentage of total variation accounted for
by a sinusoidal function of stated frequency
Fig. 3. Graph of the percentage of total variation in
sine-cosine waveforms with a cycle of stated frequency
in one patient.
1
Orcadian variance.
66
BRYAN and OVERALL
Table II. Summary of correlation analysis among circadian periodicity of temperature, pulse, respiration,
and blood pressure1
Circadian
Circadian
Direction
Temperature
Temperature
Temperature
Pulse
Pulse
Respiration
Systolic pressure
Pulse
Respiration
Systolic pressure
Respiration
Systolic pressure
Systolic pressure
Diastolic pressure
Positive
Positive
Positive
Positive
Positive
Positive
Positive
1
P< 0.05 for each pair.
Table III. Summary of correlation analysis among
mean levels and circadian periodicity of temperature,
pulse, respiration, blood pressure, and composite circadian periodicity1
Mean level
Gircadian variation Direction
Respiration
Systolic blood
pressure
Temperature
Negative
Pulse
Negative
Diastolic blood
pressure
Diastolic blood
pressure
Diastolic blood
pressure
1
Negative
Composite circadian Negative
periodicity
P< 0.05 for each pair.
Table IV. Summary of correlation analysis among
mean levels of temperature, pulse, respiration, and
blood pressure1
Mean level
Mean level
Direction
Temperature
Temperature
Temperature
Pulse
Respiration
Systolic blood
pressure
Respiration
Systolic blood
pressure
Diastolic blood
pressure
Diastolic blood
pressure
Positive
Positive
Positive
Pulse
Respiration
Respiration
Systolic blood
pressure
1
P< 0.05 for each pair.
Positive
Positive
Positive
Positive
respiration, and systolic blood pressure in various combinations (table II). The circadian periodicity of the
diastolic blood pressure correlated to a high degree only
with the circadian periodicity of systolic pressure
(r = 0.53). Since it is well known that the mean levels
of physiological functions of an individual change
significantly with age (pulse rate tends to decline, and
systolic and diastolic pressures tend to increase), it was
necessary to test the degree to which these mean levels
contributed to the circadian periodicity of each function. Table III lists the correlations in which the likelihood of occurrence by chance was less than 5%. The
intercorrelations between the mean levels of temperature, pulse, respiration, systolic, and diastolic pressures
are shown in table IV.
Since the circadian periodicity of four of the five
physiologic functions being studied correlated positively with each other, a combination of these functions
should yield a more powerful basis for the analysis of
circadian phenomena. The advantage sought in defining a composite score was increased reliability of
measurement of the underlying process. Each peripheral measurement is subject to a variety of unspecified influences, as well as simple error of measurement. It can be shown that the additive combination
of several positively correlated variables enhances reliability. In fact, if several variables have essentially
equal reliability, it can be shown that simple weighting
by reciprocals of the standard deviations is optimal to
maximize reliability of the composite [9]. Thus, a
composite function of circadian periodicity was calculated that represented a more comprehensive estimate
of circadian activity in a given individual than any
single measurement. This function was called 'composite circadian periodicity' (CCP) and is derived as
follows:
P
+
SDt
+
SD p
+
SDr
SD sys
where T = circadian variance of temperature; P =
circadian variance of pulse rate; R = circadian variance of respiratory rate; SYS = circadian variance of
systolic blood pressure; and SD = standard deviation
of indicated circadian variance. Thus, CCP is a weighted sum of the circadian variances of temperature,
pulse, respiration, and systolic blood pressure for the
given individual. The weights (reciprocals of standard
deviations) were chosen to equate the contributions of
the four components to the CCP [13].
A multiple classification analysis of variance-covariance was computed for the circadian periodicity in
temperature, pulse rate, respiratory rate, systolic and
diastolic blood pressure, and CCP. The dependence
upon 1) the mean levels for each of the single functions,
2) the duration of hospitalization, 3) the age of the
patient at the time of the studies, 4) the sex, and 5) the
67
Development of human circadian periodicity
diagnosis, was determined. The results of the multiple
classification analysis of variance-covariance for circadian temperature are shown in table V. A summary
of the same analysis for each of the physiologic functions and CCP is shown in table VI. The diagnostic
groups were acute nephritis, nephrotic syndrome,
hypopituitary disorders, essential hypertension, metabolic bone disease, adrenal disorders, diabetes mellitus,
control subjects, and 'other'. An additional multiple
classification analysis was done which included the
symptom 'hypertension' instead of the 'other' category
as one of the diagnostic variables. The summary of
that analysis is shown in table VI. In these analyses of
variance-covariance, the effect of age was most significant (J*< 0.01) in its contribution to GGP and was clearly
distinguishable from the effect of mean levels. This relation of CCP to age was now studied in greater detail.
A polynomial regression analysis testing the contribution of age, age2, and age3 to the CCP revealed
that each of these terms contributed significantly
(P<0.05) to the CCP. It was found that the age trend
in circadian periodicity was best expressed by a thirddegree polynomial equation. The equation derived
by least squares fit was:
CCP = 96.8x—0.85x2+0.0017x3+4810.14
where x is the age of the patient in months. The smoothed curve representing this equation is shown infigure4.
It should be noted that there was considerable variation in the CCP as a function of age, but the smoothed
curve was the best expression of the data. The F test
revealed that the probability is less than 0.001 that
CCP is not related to age. The figure suggests that
change in circadian periodicity during maturation is
curvilinear, and that there is an increase in periodicity
up to 5-7 years of age, followed by a decrease to a
plateau at 16-18 years of age. Since these data were
obtained in a 'cross-sectional' sample, from different
patients at different ages, we can only infer that longitudinal data from the same patient would show similar
changes. The number of subjects in the youngest and
oldest groups made those portions of the curve less
Table V. Summary of multiple classification analysis of variance-covariance for circadian temperature1
Effects of
Mean levels
Duration of hospitalization
Age
Diagnosis
Sex
Residual error
1
SS
df
MS
F
P
0.96 xlO 6
0.13X106
0.52 XlO6
0.32 xlO 6
0.16xl0 5
0.62 xlO 7
5
1
3
6
1
126
0.19xl0 6
0.13xl0 6
0.17xl0 e
0.53 xlO 5
0.16xl0 5
0.49 xlO 5
3.95
2.69
3.51
1.08
0.32
—
<0.01
>0.05
<0.05
>0.05
>0.05
—
SS = sum of squares; df = degrees of freedom; MS = mean squares; F = F ratio; and P = probability.
Table VI. Summary of multiple classification analyses of variance-covariance for independent effects of mean
levels, duration, age, diagnosis, and sex on circadian variance of temperature, pulse rate, respiratory rate,
systolic and diastolic blood pressure, and composite circadian periodicity (CCP)
Circadian-dependent
variable
Temperature
Pulse rate
Respiratory rate
Systolic blood pressure
Diastolic blood pressure
CCP
CCP with hypertension
1
ns = not significant at 5 % level.
Independent effects of
Mean levels
Duration of
hospitalization
Age
Diagnosis
Sex
<0.01
<0.01
<0.05
<0.01
ns
<0.01
<0.01
ns 1
<0.05
ns
ns
ns
<0.05
<0.05
<0.05
<0.05
<0.05
ns
ns
<0.01
<0.01
ns
ns
ns
ns
<0.05
ns
ns
ns
ns
ns
ns
<0.05
ns
ns
68
BRYAN and
stable. It was recognized that the mean levels of the
functions studied, the diagnostic categories of the patients, the duration of hospitalization, the sex, as well
as unknown factors, may influence this curve.
Thus, several means of evaluation have demonstrated the interrelation among circadian periodicity of the
functions examined in this study, the age of the subject,
and the average levels of the various functions. The
studies also demonstrated that the circadian periodicity
of certain functions was related to the length of hospitalization and to sex of the patient. Of additional interest was the consistently negative correlation of circadian periodicity with mean levels of temperature,
respiratory rate, systolic, and diastolic blood pressure
(table III).
OVERALL
iodicr
o
cussed in previous studies [2, 6]. The prevailing concepts of the development of circadian periodicity were
presented at the 39th Ross Conference for Pediatric
Research [11]. The present studies add an additional
dimension to the work, and establish certain interrelations that must be included in subsequent evaluations
of the development of circadian periodicity. The use
of power spectrum analysis for quantitation of circadian periodicity makes these analyses possible.
The rationale for combining the variance of the
circadian periodicity of temperature, pulse rate, respiratory rate, and systolic blood pressure into a single,
more powerful function, CCP, developed from an
intuitive evaluation of the correlation analyses. If one
may assume that the circadian variation in each of the
physiologic functions is an indirect reflection of an underDiscussion
lying or exogenously impressed 'circadian' function,
then the best estimate of the primary changes would
The developmental aspects of human circadian period- be obtained by a combination of the physiologic indiicity have been summarized recently [7] and dis- cators. This concept was partially confirmed by the
increased statistical probability of the contribution of
'age' seen with CCP in table VI, compared with the
individual function. The probability that circadian
14 periodicity was not related to age was less than 1 in
*/
.
100 with the CCP analyses, whereas the probability
£•12was
less than 5 in 100 for the individual functions of
• • • ' . - .
•*
circadian variation of temperature, pulse rate, respira*
•• •
tory rate, systolic, and diastolic blood pressure.
• -•
The relation between circadian periodicity and the
£ 6age
of the subject was particularly important. Previous
a
studies in mature subjects have not evaluated the ino
"
fluence of aging on circadian periodicity, and thereI 2"
fore,
the contribution of aging to the variation between
(So- 1
individuals
was unknown. Although our initial hypo30 60 90 120 50 180 210 240 270
thesis concerning the development of circadian periodAge (months)
icity proposed a gradual increase during the matura4 0 - Distribution of devistions about regression line in fig.4
tion process, it was not surprising that the changes related to age were much more complex. The extension
of the present studies to include individuals in the 3rd
30and 4th decades of life may complicate the relation
even further. The present studies included several
individuals who were less than 3 years of age, but it
—
20will be necessary to examine this portion of the life
span in much greater detail. Although previous studies have not demonstrated circadian periodicity in the
first few days of life, this may result from difficulties in
measurement rather than a complete lack of periodicity
E
1
z 0
[2]. Thus, it appears illogical to predict that a newborn
15 1.5-1.01.0-0.5 0.5-M-0.5
infant,
coming from an environment that has a well1nches from smoothed curve in fig.4
established temperature periodicity, would lose that
Fig. 4. Plot of values of CCP related to age of the pa- periodic function immediately, only to assume it again
tient. The curve represents the predicted change in within about 3 weeks. If, however, the maintenance
CCP with increasing age. This shows that the peak of of periodicity is a function of neurological maturation,
circadian periodicity occurs at approximately 6 years of then it is reasonable to predict that the nervous system
age (72 months). Graph illustrates distribution of data. of the infant would not maintain the maternal period•
•
(b
•
• • • • •
*
• • * •••
-
lian
c
ft
•
4
*
*
ber of point
o
*
5.5-1.0,1.0-1.5 > 1 5
Development of human circadian periodicity
icity until a critical amount of neurological maturation had occurred. This suggestion is supported by
studies indicating that circadian periodicity develops
later in premature infants than in children born at
term [2].
It is not yet possible to formulate an hypothesis
which includes the observation that the mean level
of the various functions contributes or subtracts from
the circadian periodicity distinct from the relation to
age. The changes of pulse rate, respiratory rate, and
blood pressure with age are well known [8]; body temperature may be related to age [1]. It can be seen that
increments of respiratory rate and diastolic blood
pressure are associated with decrements in circadian
periodicity (table III). It has been reported that increments of temperature are associated with increments
of circadian activity [2], but our studies did not confirm this observation as a phenomenon distinct from
the increment associated with age. Since the mean
levels of these functions are to some degree age-dependent, additional studies with the specific alteration of
single functions in subjects of a given age will be required to clarify this relation.
The conclusions that may be drawn from studies of
hospitalized patients will remain tenuous until they
are applied to normal individuals in normal surroundings. The validity of the present study is enhanced by
the demonstration that the diagnostic category for the
individual patient neither contributed nor detracted
a significant amount from the CCP. This was shown
conclusively in the multiple classification analysis of
variance-covariance (table VI). Although individual
patients in this study occasionally exhibited bouts of
hyperthermia or hypothermia, these episodes reduced
the total quantity of variance that could be contributed
to the circadian periodicity, but did not ablate it. Obviously, in studies of extremely short duration (less
than 7 days), the influence of such alterations would be
much greater.
Summary
In 143 patients, hospitalized for various disorders,
temperature, pulse rate, respiratory rate, systolic, and
diastolic blood pressures were obtained at 6-h intervals. The data were analyzed by a method of harmonic
analysis, and a measure of circadian periodicity for
each of these functions was calculated. Gircadian
periodicity was shown to be highly correlated with age;
the correlation assumed a relation which was best
described by a third-degree polynomial equation.
Circadian periodicity also was related to the mean
level of the function measured, and this relation was
69
at least partially independent of age. The application
of these findings to the interpretation of measurements
of temperature, pulse rate, respiratory rate, systolic,
and diastolic blood pressure was proposed.
References and Motes
1. DuBois, E. F.: Fever and the regulation of body
temperature, p. 7 (Thomas, Springfield, 111. 1948).
2. HELLBRUGGE, T.: The development of circadian
rhythms in infants. Cold Spr. Harb. Symp. quant.
Biol. 25.- 311 (1960).
3. HELLBRUGGE, T.; LANGE, J.E.; RUTENFRANZ, J.
and STEHR, K..: Circadian periodicity of physiological functions in different stages of infancy and
childhood. Ann.N.Y.Acad.Sci. 117: 361 (1964).
4. IRWIN, O. C.: The amount and nature of activities
of newborn infants under constant external stimulating conditions during the first ten days of life.
Genet. Psychol. Monogr. 8: 1 (1930).
5. JUNDELL, I.: tiber die nykthemeralen Temperaturschwankungen im ersten Lebensjahre des Menschen. Jb. Kinderheilk. 59: 521 (1904).
6. KLEITMAN, N. and RAMSAROOP, A.: Periodicity in
body temperature and heart rate. Endocrinology
43: 1 (1948).
7. MILLS, J. N.: Human circadian rhythms. Physiol.
Rev. 46: 128(1966).
8. NELSON, W. E. (ed.): Textbook of pediatrics, 8th
ed., pp.40, 881 (Saunders, Philadelphia 1964).
9. OVERALL, J. E.: Reliability of composite ratings.
Educat. Psychol. Measurement 25: 1011 (1965).
10. PARMELEE, A.H.; SCHULZ, H.R.; DISBROW, M.A.
and LITT, M.: Sleep patterns of the newborn. J.
Pediat. 58: 241 (1961).
11. RUTENFRANZ, J.: The development of circadian
system functions during infancy and childhood;
in: Circadian Systems, 39th Ross Conference
Pediatric Research, pp. 38-41 (Ross Lab., Columbus, Ohio 1961).
12. Circadian refers to a period of time of approximately 24 h. A cycle refers to a complete set of
events or phenomena recurring in the same sequence. A period is the time occupied by a single
cycle. Periodicity and rhythm are used synonymously, and refer to the phenomenon of recurring
cycles.
13. As an example of calculation of the composite
CCP, consider an individual whose circadian
variances of the four components of temperature,
pulse, respiration, and systolic blood pressure were
average values of 0.46, 0.23, 0.30, and 0.14, respectively. Multiplying each component by the
70
BRYAN and OVERALL
reciprocal of its standard deviation across the total
sample, the CCP for this individual would be:
The mean of CGP for patients in our sample was
approximately X = 6.4 with effective range from
0.5 to 12.0. Assuming a normal distribution, 95%
of the values should be expected to fall within ± 2
SD units of the mean, or 6.4±6.1.
14. The authors gratefully recognize the cooperation
of other investigators with children in the Clinical
Study Center and tolerance of the patients themselves; Mr. RON KLEIBRINK and Mr. HARVEY
BUNCE, Jr., assisted with the computor analyses;
Mrs. RUTH MCBEE provided invaluable clerical
assistance. The nursing personnel of the Clinical
Study Center deserve special thanks for their
tireless effort in making the primary observations.
15. Supported by Public Health Service Research
Grants nos. DHEW 5M01 FR-00073, 5 P07 FR00024, and 5 R01 HE-09366 from the National
Institutes of Health, Division of Research Facilities
and Resources. Some of these studies were presented to the Southern Society of Pediatric Research,
November, 1968. (Sth med. J. 61: 1339-40, 1968.)
16. Requests for reprints should be addressed to:
GEORGE T.BRYAN, M.D., Clinical Study Center,
University of Texas Medical Branch, Galveston,
TE 77550 (USA).
17. Accepted for publication August 20, 1969.