See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/6073219

Area Under the Curve and Other Summary

Indicators of Repeated Waking Cortisol
Measurements

Article in Psychosomatic Medicine September 2007

DOI: 10.1097/PSY.0b013e31814c405c Source: PubMed

CITATIONS READS

176 2,542

7 authors, including:

Michael E Andrew John M Violanti

Centers for Disease Control and Prevention University at Buffalo, The State University
235 PUBLICATIONS 3,846 CITATIONS 184 PUBLICATIONS 2,625 CITATIONS

SEE PROFILE SEE PROFILE

Luenda E Charles Diane Miller

Centers for Disease Control and Prevention Centers for Disease Control and Prevention
89 PUBLICATIONS 872 CITATIONS 159 PUBLICATIONS 6,206 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Michael E Andrew
letting you access and read them immediately. Retrieved on: 02 August 2016
Area Under the Curve and Other Summary Indicators of Repeated Waking
Cortisol Measurements
DESTA B. FEKEDULEGN, PHD, MICHAEL E. ANDREW, PHD, CECIL M. BURCHFIEL, PHD, JOHN M. VIOLANTI, PHD,
TARA A. HARTLEY, MPH, LUENDA E. CHARLES, PHD, AND DIANE B. MILLER, PHD
Objective: To derive the area under the curve and related summary measures of stress from saliva samples collected over time and
to provide insight into the interpretation of the derived parameters. In research designed to assess the health consequences of stress
these samples are often used as a physiologic indicator of the responsiveness of the hypothalamic-pituitary-adrenal (HPA) axis. To
make these repeated measurements of salivary cortisol more useful in defining the relationships between stress and health there is
a need to derive two forms of area under the curve that summarize the measurements: area under the curve with respect to ground
(AUCG) and area under the curve with respect to increase (AUCI). The latter parameters, AUCI, however, is seldom used by
research scientists. Methods: In this study, interpretation and generic definition of the area under the curve was provided through
graphical analyses and examination of its association with other summary measures using data from the Buffalo Cardio-Metabolic
Occupational Police Stress (BCOPS) Pilot Study. In generic form, AUCI is derived as the area under the curve above the baseline
value minus the area above the curve below the baseline value. Results: The sign and magnitude of AUCI are related to the profile
and the rate of change of the measurements over time. The parameter showed significant associations with other summary indicators
that measure pattern or rate of change of the measurements over time. Conclusion: Principal components analyses revealed that
summary parameters derived from repeated cortisol measurements can be grouped into two meaningful general categories:
measures of the magnitude of response and measures of the pattern of response over time. Key words: repeated cortisol
measurements, area under the curve, principal component analysis, total hormonal secretion, time course of salivary cortisol.

AUCG area under the curve with respect to ground; AUCI area
under the curve with respect to increase; AUCB area under the
curve with respect to baseline; HPA hypothalamic-pituitary-adre-
nal; PCA principal component analyses.
INTRODUCTION
M any studies utilizing the salivary cortisol measure as a
physiologic indicator of the responsiveness of the hy-
pothalamic-pituitary-adrenal (HPA) axis to determine the
health consequences of stress obtain these cortisol values in
samples collected over time from the same subject. In research
involving repeated measurements of a response variable, there
is a need to derive parameters that summarize the information
contained in the multivariate data. The area under the curve
(AUC) computed with the trapezoidal formula is one of the
widely used parameters (19) but has had little application in
work involving repeated measurements of salivary cortisol.
In repeated data, each measurement is comprised of two
types of information (Figure 1A); its distance from the ground
(i.e., the intensity or magnitude of the response) and its dis-
tance from its neighbor (i.e., changes over time). Unlike other
summary measures (e.g., baseline value, average, maximum),
From the Biostatistics and Epidemiology Branch (D.B.F., M.E.A., C.M.B.,
T.A.H., L.E.C.), Toxicology and Molecular Biology Branch (D.B.M.), Health
Effects Laboratory Division, National Institute for Occupational Safety and
Health, Centers for Disease Control and Prevention, Morgantown, West
Virginia; Department of Social and Preventive Medicine (J.M.V.), School of
Public Health and Health Professions, State University of New York at
Buffalo, Buffalo, New York.
Address correspondence and reprint requests to Desta B. Fekedulegn,
Biostatistics and Epidemiology Branch, National Institute for Occupational
Safety and Health, HELD/BEB, MS 4050, 1095 Willowdale Rd., Morgan-
town, WV 26505. E-mail: djf7@cdc.gov
Received for publication October 30, 2006; revision received May 10,
2007.
This work was supported by Contract No. 200 to 2003 to 01580 from the
National Institute for Occupational Safety and Health. The findings and
conclusions in this report are those of the authors and do not necessarily
represent the views of the National Institute for Occupational Safety and
Health.
DOI: 10.1097/PSY.0b013e31814c405c
both pieces of information can be captured through the use of
AUC. The use of AUC simplifies the statistical analyses by
transforming the multivariate data into univariate space, espe-
cially when the numbers of repeated measurements are high
and there is a need to summarize the information (7). This
approach also reduces the number of statistical comparisons
between groups, which minimizes the need for adjustment of
the significance level. With AUC, the number of statistical
comparisons only depends on the number of groups to be
compared, as opposed to the original repeated data. In addi-
tion, when the time interval between repeated measurements is
not identical, the use of AUC provides an alternative because
repeated measures analysis of variance, using the original data,
has no proven method to adjust for these differences (10).
A recent study by Pruessner et al. (10) provides a simple
formula for the computation of two types of AUC that reveal
different information embodied in the repeated multivariate
measurements (Figure 1B): area under the curve with respect
to ground (AUCG) and area under the curve with respect to
increase (AUCI). The formulas are basically simple additions
of areas consisting of triangles and rectangles. AUCG is the
total area under the curve of all measurements. It takes into
account both sensitivity (the difference between the single
measurements from each other) and intensity (the distance of
these measures from ground). AUCI is calculated with refer-
ence to the baseline measurement and it ignores the distance
from zero for all measurements and emphasizes the changes
over time. With endocrinological data, AUCG is assumed to
be a measure more related to total hormonal output, whereas
AUCI is a parameter that emphasizes the changes over time
and is more related to sensitivity of the system.
Despite the simple computational formula for AUCI, the
parameter has had limited application in research compared
with AUCG. This could be attributed to the lack of a generic
formula and clear interpretation of the parameter and the
limited literature on its association with other summary indi-
cators. In addition, graphical representation of the region

Psychosomatic Medicine 69:651 659 (2007) 651

0033-3174/07/6907-0651
Copyright 2007 by the American Psychosomatic Society
 D. B. FEKEDULEGN et al.

accounted for by AUCI is not well documented for situations

Figure 1. Plot of repeated measurements indicating magnitude of response
or intensity (I1, I2, and I3) at each time point, changes in the response over
time or sensitivity (S1 and S2) (A) and the three forms of AUC (B). AUC
area under the curve.
TABLE 1. Simulated Data on Waking Cortisol Measurements at Awakening, 15, 30, and 45 Minutes After Waking and Derived Summary
Measures of the Data
where one or more of the measurements have values lower
than the baseline measurement. The main objectives of this
study, therefore, are to provide more insight into the interpre-
tation of AUCI by defining a generic formula for AUCI
through graphical analyses, to study its association with other
well defined cortisol parameters known to measure different
aspects of the multivariate data, and to select optimal param-
eters that capture meaningful information about the original
data using dimension reduction analytical techniques.
METHODS
Artificial Data
The study used two different data sets for analyses. First, a simulated data set
(Table 1) representing a variety of different waking cortisol measurement patterns
(at awakening, 15, 30, and 45 minutes after waking) was used to derive a generic
formula for AUCI and to demonstrate how the sign and magnitude of AUCI
changes with various cortisol profiles. The data were also used to illustrate the
parameters (summary measures) that can be computed to describe the information
contained in the waking measurements. The derived parameters include the
following: average of the four waking cortisol measurements (AVE); area under
the curve with respect to the baseline (AUCB); area under the curve with respect
to the ground (AUCG); area under the curve with respect to increase (AUCI); area
under the curve above the baseline (AUCAB); peak cortisol (PK); time in minutes
from baseline to peak (TBP); slope from baseline to peak (SBP); reactivity (RT);
slope of the line through the baseline and last measurement (SP1); intercept of the
regression line fitted through the raw cortisol data (INT); slope of the regression
line fitted through the raw cortisol data (SP2); and area under the curve of the
regression line fitted through the raw cortisol data (AUR).
Average cortisol is the average of the four measurements. Peak cortisol is
the maximum value of the measurements regardless of when it occurred. Time
from baseline, where baseline is the measurement at awakening, to peak is the

Subjects (8 cases)
Characteristics
(a) (b) (c) (d) (e) (f) (g) (h)

Cortisol measurements
At awakening 9.60 9.60 9.60 9.60 14.1 14.1 14.1 14.1
15 min after waking 13.4 19.0 13.4 10.2 13.4 11.1 10.0 3.00
30 min after waking 16.5 11.1 11.1 9.10 11.1 9.10 6.10 6.10
45 min after waking 17.2 7.20 7.20 6.40 9.40 7.10 2.40 9.50
Derived cortisol parameters
AVE 14.18 11.73 10.33 8.83 12.00 10.35 8.15 8.18
AUCB 432 432 432 432 635 635 635 635
AUCG 650 578 494 410 544 462 365 314
AUCI 218 146 62 23 91 173 269 321
AUCAB 218 157 73 7 0 0 0 0
PK 17.20 19.00 13.40 10.20 14.10 14.10 14.10 14.10
TBP 45.00 15.00 15.00 15.00 0.00 0.00 0.00 0.00
SBP 0.17 0.63 0.25 0.04 0.00 0.00 0.00 0.00
RT 7.60 2.40 2.40 3.20 4.70 7.00 11.7 4.60
SP1 0.17 0.05 0.05 0.07 0.10 0.16 0.26 0.10
INT 10.29 13.99 11.75 10.43 14.46 13.80 14.00 9.78
SP2 0.17 0.10 0.06 0.07 0.11 0.15 0.26 0.07
AUR 638 528 465 397 540 466 367 368

AVE average of the four waking cortisol measurements; AUCB area under the curve with respect to the baseline; AUCG area under the curve with respect
to the ground; AUCI area under the curve with respect to increase; AUCAB area under the curve above the baseline; PK peak cortisol; TBP time in
minutes from baseline to peak; SBP slope from baseline to peak; RT reactivity (the change in salivary cortisol during the observation period); SP1 slope
of the line through the baseline and last measurement; INT intercept of the regression line fitted through the raw cortisol data; SP2 slope of the regression
line fitted through the raw cortisol data; AUR area under the curve of the regression line fitted through the raw cortisol data.

652 Psychosomatic Medicine 69:651 659 (2007)

ANALYSES OF AWAKENING CORTISOL MEASUREMENTS

number of minutes from the baseline to the peak value. Slope from baseline TABLE 2. Descriptive Summary Statistics of the Waking Cortisol
to peak is the slope of the line connecting the baseline to the peak value. The Response and Derived Cortisol Parameters, Buffalo Cardio-Metabolic
reactivity was defined as the change in salivary cortisol during the observation Occupational Police Stress (BCOPS) Pilot Study, 20012003
period and was calculated as the difference between the first and the last
sample (sample 4 sample 1). Slope of the line through the baseline and last Characteristics Minimum Maximum Mean SD
measurement is a scaled version of reactivity; it is reactivity divided by time
in minutes from baseline to last measurement. Cortisol measurements
Previous studies (3,11) used the slope from a fitted regression model and (n 68)
the time-weighted average to describe the pattern and the amount of cortisol At awakening 0.1 49.4 13.5 10.5
secreted, respectively. Hence, the last three parameters (INT, SP2, and AUR) 15 min after waking 0.1 74.5 16.6 13.7
are derived after fitting a simple linear regression model to the waking data. 30 min after waking 0.9 57.1 15.4 11.5
The area under the regression line (AUR) was computed by using the 45 min after waking 0.1 55.4 14.1 10.3
estimated equation and integrating the resulting function as follows: Derived parameters
(n 68)

AVE 0.9 47.0 14.9 9.9
AUR a bxdx a b
2
2 AUCB
AUCG
4.5
58.5
2223.0
2244.0
641.4
725.6
480.1
482.7
0
AUCI 670.5 834.8 84.2 278.1
AUCAB 0.0 834.8 167.5 195.7
where is the time interval in minutes from the baseline measurement to the PK 1.6 74.5 20.8 14.7
last measurement, x is time from baseline (predictor variable), a is the TBP 0.0 90.0 24.3 19.3
intercept, and b is the slope of the fitted regression line. SBP 0.0 1.7 0.3 0.4
RT 34.5 29.6 0.6 11.6
Data From BCOPS Study SP1 0.8 0.7 0.0 0.3
Endocrinological data from a recent study with 100 police officers were INT 1.4 63.7 14.9 12.2
chosen to study the association of AUCI with other cortisol parameters. SP2 1.0 0.7 0.0 0.3
Participants are 100 police officers from Buffalo, New York, who were AUR 57.6 2112.8 710.7 462.6
randomly selected in the Buffalo Cardio-Metabolic Occupational Police
Stress (BCOPS) Pilot Study and were examined between 2001 and 2003. The SD standard deviation; AVE average of the four waking cortisol
study design, methods, and participant characteristics are described by Vio- measurements; AUCB area under the curve with respect to the baseline;
lanti and colleagues (12). Cortisol was measured in saliva samples and was AUCG area under the curve with respect to the ground; AUCI area under
analyzed using a competitive chemiluminescence immunoassay technique. the curve with respect to increase; AUCAB area under the curve above the
The participants collected 13 saliva samples at various times during a 3-day baseline; PK peak cortisol; TBP time in minutes from baseline to peak;
period. On the second day, awakening cortisol measures were obtained from SBP slope from baseline to peak; RT reactivity (the change in salivary
four salivary samples taken at 15-minute intervals on first awakening (at cortisol during the observation period); SP1 slope of the line through the
awakening, 15, 30, and 45 minutes after waking). Morning cortisol response baseline and last measurement; INT intercept of the regression line fitted
is widely studied because it serves as a useful index of adrenocortical activity. through the raw cortisol data; SP2 slope of the regression line fitted through
Hence, the waking data were used for statistical analyses in this study. From the raw cortisol data; AUR area under the curve of the regression line fitted
the 100 participants, 68 subjects with complete (nonmissing) data on all four through the raw cortisol data.
waking cortisol samples and times of measurement were used in this study.
Summary statistics of the waking measurements and the derived parameters
for the BCOPS waking data are shown in Table 2. generic definition of AUCI suitable to all patterns. The first
We examined the association between the area measures (AUCI and AUCG) scenario (Figure 2) illustrates a pattern where all subsequent
and the other cortisol parameters by using correlation and cluster analyses. We measurements are greater than the baseline value. For such pat-
used cluster analysis (k-means clustering) to classify participants into two clus-
terns, AUCI is always the area under the curve above the baseline
ters, using AUCI as the clustering variable and Students t test to assess differ-
ences in other cortisol parameters between these two clusters. Principal compo- value and hence, AUCI AUCAB. In the second scenario
nent analysis (PCA) was employed to understand the correlation structure among (Figure 3), the last measurement is less than the baseline value.
the derived cortisol parameters and reduce the dimension of the summary mea- Generally, when one or more of the subsequent measurements
sures to a few new meaningful components. The Eigenvector loadings associated are less than the baseline value, AUCI is not the area under the
with the new components were examined graphically to assess what aspect of the
curve above the baseline value. For such patterns, AUCI is the
multivariate data the components measure. All statistical analyses were per-
formed using the SAS/STAT software, version 9.0 for Windows. difference of two areas: area under the curve above the baseline
value (AUCAB) minus area above the curve below the baseline
value. This is a generic definition for AUCI applicable to all
RESULTS patterns of measurements. The first scenario is a special case of
Generic Definition of AUCI this generic definition where the area above the curve below the
A generic formula for AUCI is illustrated using two scenarios. baseline value is zero. Note that for patterns where all subsequent
The purpose of this section is not to provide a formula for measurements are less than the baseline value, AUCAB is always
computation of AUCG and AUCI, as this is already presented by zero and consequently AUCI takes a negative value.
Pruessner and colleagues (10). Rather, we intend to show that the
existing definition of AUCI, area under the curve above the Scenario I
baseline value, is not applicable to all patterns of measurements. Four repeated measurements taken at 15-inute intervals (sam-
Here we treat the area under the curve above the baseline value ple 1 (baseline) measurement 9.6 at 0720 AM; sample 2
(AUCAB) as a separate parameter. We then provide a more 13.4 at 0735 AM; sample 3 16.5 at 0750 AM; and sample 4

Psychosomatic Medicine 69:651 659 (2007) 653

 D. B. FEKEDULEGN et al.

Figure 3. Pattern of four repeated measurements taken at 15-minute inter-

Figure 2. Pattern of four repeated measurements taken at 15-minute inter-
vals where all subsequent measurements are larger than the baseline value.
AUCI is the area under the curve above the referent baseline measurement
(shaded region).
14.2 at 0805 AM). The pattern of the measurements is shown in
Figure 2. The area under the curve with respect to the baseline
(AUCB), area under the curve with respect to the ground (AUCG),
area under the curve with respect to increase (AUCI), and area under
the curve above the baseline (AUCAB) are computed as:
AUC B 9.6 15 15 15 432
vals where the value of the last sample is less than the baseline measurement.
AUCI is the difference of two areas: area under the curve above the baseline
value (AT1 AT2 AR2 AT3) minus area above the curve below the
baseline value (AT5).
Scenario II
Four repeated measurements taken at 15-minute intervals
(sample 1 (baseline) measurement 9.6 at 0720 AM; sample
2 13.4 at 0735 AM; sample 3 11.1 at 0750 AM; and
sample 4 7.2 at 05 AM). The pattern of the measurements
is shown in Figure 3. As in scenario I, AUCB, AUCG and
AUCI are calculated using the computational formula as:
AUC B 9.6 15 15 15 432

AUC G 9.6 13.4

2 15 13.4 16.5
2 AUC G 9.6 13.4
2 15
15

16.5 14.2
2 15 627 13.4 11.1
2 15
AUC I AUC G AUC B 627 432 195
11.1 7.2
15 493.5

2
13.4 9.6
AUC AB 15 AUC I AUC G AUC B 493.5 432 61.5
2

13.4 9.6 16.5 9.6

2 15 However, to compute AUCAB and to illustrate the generic
definition of AUCI, the region between two sampling points is
partitioned into a series of triangles (T) and rectangles (R). The

16.5 9.6 14.2 9.6

2 15 area (A) of each triangle and rectangle was calculated and given as:
AT1 28.5, AT2 17.25, AT3 4.33, AT4 11.08, AT5 11.08,
AR1 144, AR2 22.5, AR3 144, AR4 13.85, and AR5 108.
195 The generic definition entails that AUCI is the difference of two

654 Psychosomatic Medicine 69:651 659 (2007)

ANALYSES OF AWAKENING CORTISOL MEASUREMENTS
areas. The first component, area under the curve above the
baseline value (AUCAB), is the sum of areas of T1, T2, T3 and
R2 (28.5 17.25 4.33 22.5 72.58). The second
component, area above the curve below the baseline, is the
area of T5 (11.08). AUCI is then 72.58 11.08 61.5; the
same result as in the computational formula.
Note that to compute the area of the triangles named T3, T4,
T5 and the rectangle named R4, the coordinates of the point
labeled a (Figure 3) need to be known. The y coordinate that
represents the cortisol level is 9.6 but the x coordinate that
represents time of measurement (in minutes) from baseline is
unknown. The x coordinate was then computed by solving the
simultaneous equation of the two intersecting lines: the hori-
zontal line through the baseline measurement (y 9.6) and
the line connecting the 3rd and 4th measurements. The equa-
tion of the line passing through the third and fourth measure-
ments is computed by solving the equation:
Y y1 y y2 y1 7.2 11.1
0.26
X x1 x x2 x1 45 30
where (x1, y1) and (x2, y2) are the coordinates of the 3rd and
4th sample points, respectively. After algebraic manipulation,
the equation of the line is
Y 18.9 0.26X.
Figure 4. (AH). Various time courses of salivary cortisol and the sign and magnitude of AUCI associated with each profile. AUCI area under the curve
with respect to increase.
Psychosomatic Medicine 69:651 659 (2007)
The x coordinate of the intersection point of the two lines
is obtained by solving the simultaneous equations: y 9.6 and
y 18.9 0.26X. This yields a value of x 35.76923
minutes from baseline.
Sign and Magnitude of AUCI
Previous studies (1315) have shown that the response
magnitude and time course of salivary cortisol levels after
awakening are significantly related to various psychological
and physical conditions (e.g., pain, stress). Therefore, the
change in the sign and magnitude of AUCI associated with
various cortisol patterns is important in understanding the
definition of the parameter. In this section, the variation of
AUCI associated with various cortisol patterns is examined
numerically and graphically. Artificial data on eight waking
cortisol patterns (Table 1) was used for illustration and the
corresponding graphical depiction of these patterns is shown
in Figure 4 (AH). Although simulated for analyses here, the
patterns in Figure 4 can also be observed in real life. In each
figure, the area represented by AUCI is shown. There are four
distinct patterns in Figure 4: a steadily increasing pattern
(Figure A); a pattern where cortisol peaks 15 minutes after
waking and decreases thereafter (Figures BD); a steadily
decreasing pattern (Figures EG); and a pattern where
655
 D. B. FEKEDULEGN et al.

cortisol drops 15 minutes after waking and increases there- TABLE 3. Correlation Coefficients of Waking AUCI and AUCG with
after (Figure H). the Raw Measurements and Other Derived Parameters, Buffalo
Cardio-Metabolic Occupational Police Stress (BCOPS) Pilot Study,
For the first example (A), AUCI is positive and clearly 20012003 (n 68)
defined; it is the area of the region under the curve above the
reference baseline measurement. The examples in (B), (C), AUCI AUCG
and (D) depict similar patterns but the AUCI values differ Cortisol
substantially. Note that, unlike the first example, AUCI is not Measurements Correlation Correlation
p p
Coefficient (r) Coefficient (r)
clearly defined in a single region but it is obtained through
subtraction of area from two regions: area above the baseline At awakening 0.296 .0143 0.810 .0001
and below the curve minus area below the baseline and above 15 min after waking 0.137 .2669 0.915 .0001
the curve. This illustrates that AUCI balances the amount of 30 min after waking 0.435 .0002 0.862 .0001
45 min after waking 0.619 .0001 0.741 .0001
increase versus decrease and can be considered as an
Derived parameters
index of change. AUCI therefore is the difference in area AVE 0.258 .0339 0.976 .0001
under and above the curve with reference to the baseline. In AUCB 0.280 .0206 0.833 .0001
(B) and (C), the amount of increase (area under the curve AUCG/AUCI 0.297 .0138 0.297 .0138
above the baseline) is larger than the amount of decrease AUCAB 0.88 .0001 0.58 .0001
PK 0.221 .0706 0.950 .0001
(area above the curve below the baseline). In (D), the amount
TBP 0.644 .0001 0.181 .1387
of increase is smaller than the amount of decrease; hence, SBP 0.493 .0001 0.655 .0001
AUCI takes a negative value. RT 0.818 .0001 0.072 .5581
The examples in (E), (F), and (G) also present similar SP1 0.809 .0001 0.091 .4590
patterns but the AUCI values show large differences in mag- INT 0.180 .1418 0.851 .0001
SP2 0.767 .0001 0.137 .2650
nitude. Note that, for these cases, the amount of increase (area
AUR 0.287 .0176 0.996 .0001
above the baseline AUCAB) is zero; hence, AUCI is the
amount of decrease and takes a negative value. The slope of AVE average of the four waking cortisol measurements; AUCB area
the regression line fitted to the patterns in (E), (F), and (G) are under the curve with respect to the baseline; AUCG area under the curve
0.11, 0.15, and 0.26, respectively (Table 1). The rate of with respect to the ground; AUCI area under the curve with respect to
increase; AUCAB area under the curve above the baseline; PK peak
decrease is highest for (G) because the cortisol value de- cortisol; TBP time in minutes from baseline to peak; SBP slope from
creases by 0.26 nmol/l for 1- minute increase in time. AUCI baseline to peak; RT reactivity (the change in salivary cortisol during the
has the largest magnitude for the curve that has the steepest observation period); SP1 slope of the line through the baseline and last
slope. This illustrates that AUCI also measures or indicates the measurement; INT intercept of the regression line fitted through the raw
rate of decrease relative to the baseline. The large differences cortisol data; SP2 slope of the regression line fitted through the raw cortisol
data; AUR area under the curve of the regression line fitted through the raw
in magnitude of AUCI for these three examples are related to cortisol data.
the differences in the rate of decrease. The pattern depicted in
(H) is a good example illustrating that large and negative
AUCI does not necessarily imply that the pattern of the mea- ground; rather, it only considers distance of the measurements
surements is strictly decreasing. In this example, cortisol de- from the reference baseline. A similar result was reported by
creased 15 minutes after waking and then picked up. There is Pruessner et al. (10). Note that area under the curve above the
a stronger decrease than increase and, hence, a negative AUCI. baseline (AUCAB) shows stronger correlation to AUCI than
In summary, AUCI is an index of change that takes negative AUCG because it is the first component in the generic formula
or positive values and translating the magnitude of AUCI into for AUCI.
patterns over time in reference to the baseline requires caution. Five parameters largely measure the pattern of the mea-
surements over time: area under the curve above the baseline
Association of AUCI with Other Cortisol Parameters (AUCAB); time from baseline to peak; reactivity; slope from
A correlation analysis of AUCI and AUCG with each of the baseline to last measurement; and slope of the regression line
original measurements as well as with other derived parame- fitted to the data. The slope of the regression line fitted to the
ters was performed to gain further insight in interpreting the data can be thought of as a more informative measure of
parameter (Table 3). The correlations with the raw measure- pattern than the other four because it utilizes all the informa-
ments indicate that AUCI is negatively and significantly cor- tion in the multivariate data. Large positive values for all four
related to the baseline measurement. This can be seen by parameters generally indicate an increasing trend in the mea-
examining the computational formula for AUCI (AUCI surements over time and, hence, a positive or large AUCI.
AUCG AUCB). AUCB is an area solely determined by the Two observations can be ascertained based on the correla-
baseline value and larger baseline measurements yield larger tion of the two area measures with the other derived param-
AUCB and, hence, smaller AUCI. It also shows that the eters (1). AUCI showed strong positive correlations with the
association between AUCI and the raw measurements is four parameters (reactivity, slope from baseline to last mea-
weaker compared with AUCG. This is because AUCI does not surement, slope of the regression line fitted to the data, and
take into account the distance of the measurements from the time from baseline to peak) although the corresponding cor-

656 Psychosomatic Medicine 69:651 659 (2007)

ANALYSES OF AWAKENING CORTISOL MEASUREMENTS

relations with AUCG were weak or negligible (2). AUCG
showed significant positive correlations with the intercept of
the regression line fitted to the data (r .85) and the corre-
sponding area under the regression line (AUR) (r .99)
although the associations of AUCI with these two parameters
were weaker. The intercept from the regression line is a model
estimate of the baseline value; hence, the two parameters (INT
and baseline cortisol) are strongly correlated (r .97, p
.001).
The two clusters of participants created using AUCI as a
clustering variable were compared with respect to the mea-
surements and the derived parameters (Table 4). The first
cluster contained 38 participants and the average AUCI for the
subjects in the first cluster was 109.5 167. The second
cluster had 30 subjects and an average AUCI of 329.6 178.
The largest differences between the two clusters of partici-
pants, as shown by the test statistic, were observed for the
following parameters: area under the curve above the baseline;
time from baseline to peak; reactivity; slope between the 1st
and 4th sample; and slope of the regression line. The result is
in agreement with the data in Table 3 that these parameters are
strongly correlated to AUCI and they all measure a common
aspect of the repeated measurements: time course of salivary
cortisol (profile). The baseline measurement, the area under
the curve with respect to the baseline (AUCB), and the inter-
cept from the fitted regression model (INT) showed the least
differences between the two clusters (Table 4).
total hormonal secretion and parameters indicating the time
course of the measurements. A formal analytical tool, princi-
pal component analysis (PCA), was used to uncover if there
are two underlying sets of parameters. The Eigenvalues and
the Eigenvectors of the correlation matrix of the 13 derived
cortisol parameters were computed. Only the first two princi-
pal components had Eigenvalues 1.0 and the majority of the
variability (nearly 90%) among the 13 parameters was ac-
counted by these two principal components. The magnitude
and sign of the Eigenvector loadings were examined to assign
meaningful interpretation to the two components. Several obser-
vations can be made based on the results of the PCAs. The first
observation is that the first principal component has large load-
ings for the average of the waking measurements, AUCB, AUCG,
peak cortisol, intercept of the regression line, and AUR. Thus,
the first principal component, which accounts for 49.3% of the
variability, seems to dominantly measure total hormonal se-
cretion. Second, the second principal component has large
loadings for AUCI, reactivity, SP1, SP2, and AUCAB. The
second component that accounts for 39.5% of the variabil-
ity in the original data seems to dominantly measure time
course of salivary cortisol. Finally, the results from PCAs
demonstrate that the large number of parameters can be
reduced to two components: the first principal component
measures intensity or total hormonal secretion and a second
component measures sensitivity or profile of cortisol over
time (Table 5).

Principal Component Analyses DISCUSSION

Based on the data in Tables 3 and 4, there seem to be two This study focused on interpretation of AUCI, the associ-
broad categories of derived parameters: parameters related to ation of area measures with other cortisol parameters, and

TABLE 4. Characteristics (Mean Standard Deviation, Test Statistic, and p Value) of Cortisol Measurements and Derived Parameters by the Two
Clusters of Participants Created Using AUCI as the Clustering Variable

Cortisol measurements Cluster I (n 38) Cluster II (n 30) t Value p (t test)

At awakening 15.3 12.5 11.3 6.8 1.65 .1036

15 min after waking 15.1 15.8 18.4 10.3 1.02 .3139
30 min after waking 11.6 10.2 20.3 11.4 3.31 .0015
45 min after waking 8.6 6.3 21.1 10.3 5.81 .0001
Derived parameters
AVE 12.6 8.5 17.7 6.7 2.18 .0332
AUCB 626.9 852.9 477.5 368.6 0.95 .3469
AUCG 605.5 508.6 877.7 406.7 2.39 .0198
AUCAB 35.1 63.3 335.3 176.7 8.87 .0001
PK 17.7 15.9 24.9 12.2 2.05 .0446
TBP 14.5 15.4 36.7 16.5 5.73 .0001
SBP 0.14 0.3 0.45 0.4 3.73 .0004
RT 6.6 8.8 9.7 7.6 8.09 .0001
SP1 0.14 0.19 0.20 0.16 7.87 .0001
INT 16.2 14.7 13.1 8.1 1.09 .2803
SP2 0.15 0.22 0.20 0.17 7.35 .0001
AUR 594.3 480.6 858.2 399.4 2.42 .0183

AVE average of the four waking cortisol measurements; AUCB area under the curve with respect to the baseline; AUCG area under the curve with respect
to the ground; AUCI area under the curve with respect to increase; AUCAB area under the curve above the baseline; PK peak cortisol; TBP time in
minutes from baseline to peak; SBP slope from baseline to peak; RT reactivity (the change in salivary cortisol during the observation period); SP1 slope
of the line through the baseline and last measurement; INT intercept of the regression line fitted through the raw cortisol data; SP2 slope of the regression
line fitted through the raw cortisol data; AUR area under the curve of the regression line fitted through the raw cortisol data.

Psychosomatic Medicine 69:651 659 (2007) 657

 D. B. FEKEDULEGN et al.

TABLE 5. Wakening Cortisol Summary Parameter Ranks Based on substantial differences in the magnitude of their AUCI values.
Their Relative Importance as Measures of Total Hormonal Secretion This is because AUCI takes into account the vertical distances
and Pattern of Response
of each measurement from the reference baseline.
Summary Measures Magnitude Rank Pattern Rank The effect of both measurement error and biological vari-
ability are important to consider. As in any other scientific
Average of the four waking 3 12 measurements, each awakening cortisol observation is com-
cortisol measurements (AVE)
Area under the curve with respect 6 8
posed of the true value plus some random error value (X
to the baseline (AUCB) T e). Random error is caused by any factor that affects the
Area under the curve with respect 2 11 measurement of the variable across the sample (subjects) and
to the ground (AUCG) is always part of any measurement. The accuracy of a cortisol
Area under the curve with respect 12 1
measurement may be as important as the measurement itself.
to increase (AUCI)
Area under the curve above the 8 5 In addition, the inherent (biological) variability associated
baseline (AUCAB) with each measurement is another important source of varia-
Peak cortisol (PK) 1 13 tion. For example, the baseline waking cortisol measurement
Time in minutes from baseline to 13 6 may have the largest variability of all cortisol measurements
peak (TBP)
Slope from baseline to peak (SBP) 7 7
within a day as it is obtained during the time of rapid rise from
Reactivity (RT) 11 2 the late night nadir to the morning peak. Multiple-day awak-
Slope of the line through the 10 3 ening cortisol sampling would be ideal to estimate and under-
baseline and last measurement stand the biological variation associated with each measurement.
(SP1)
Such extensive observations involving multiple day sampling
Intercept of the regression line 4 9
fitted through the data (INT) of cortisol may not be economically feasible and contribute to
Slope of the regression line fitted 9 4 participant burden. The analyses in this study were based on 1
through the data (SP2) day of awakening cortisol sampling (no retest reliability);
Area under the fitted regression 5 10 hence, we cannot provide an estimate of the variability asso-
line (AUR)
ciated with each of the waking measurements. However, we
The summary measures were ranked using the first and second principal believe that the variability in the baseline waking measure-
components. Rank 1 is the most important and rank 13 is the least important. ment has the largest influence in the magnitude of AUCI. This
is because AUCI is the difference between AUCG and AUCB
and the baseline measurement solely determines AUCB. For
application of PCAs to reduce the large number of summary
example, if we assume a random error of 1.5 nmol/l for the
indicators of repeated cortisol measurements to a smaller
baseline value of the data shown in Figure 3, the value of
meaningful set. Generically, AUCI was defined as the area
under the curve above the baseline minus the area above the AUCI could change by 56.3 area units.
curve below the baseline. The formula implies that AUCI In analyses of multivariate data, PCA can serve a dual role:
takes into account both sensitivity and intensity (the distance 1) it is often used to reduce the dimension of a correlated set
of the measurements from baseline value). It is a measure of of variables; and 2) PCA is also used to partition experimental
the pattern or rate of change of the repeated measurements units into subgroups so that units that are similar are in the
over time. The statistical analyses using the BCOPS data same group. In the latter case, the principal components can be
demonstrated that AUCI is highly and significantly correlated used as input into clustering programs (16). In this study, the
to those summary indicators that are known to measure the technique significantly reduced the number of summary mea-
pattern or rate of change of the measurements over time (e.g., sures and the new components explained the majority of the
reactivity and slope of the fitted regression line). Based on the variability in the original measures and had meaningful bio-
results of this study, the following conclusions can be made logical interpretation. According to PCA, the most important
regarding AUCI as a cortisol parameter. First, AUCI is always measures of total hormonal secretion are area under the curve
negative for patterns that are steadily decreasing over time. with respect to the ground (AUCG), peak cortisol, average of
The variability in magnitude of AUCI among these cases is the measurements, and intercept of the fitted regression line
directly related to variation in slope; the highest magnitude (INT), whereas the corresponding measures of time course of
being associated with a pattern having the steepest slope. salivary cortisol are area under the curve with respect to
Second, AUCI is always positive for patterns that are steadily increase (AUCI), reactivity (RT), SP1 and SP2. This result is
increasing over time. The magnitude of AUCI increases with consistent with previous studies (3,11). Note that the impor-
the rate of increase (slope) of the pattern. Third, for patterns tant measures of total hormonal secretion (AUCG, PK, AVE,
that exhibit a series of increases and decreases over time, and INT) are strongly correlated with the baseline measure-
AUCI can be positive or negative depending on the amount of ment (r .81 for AUCG; r .83 for PK; r .84 for AVE;
increase relative to decrease. Fourth, subjects with similar and r .97 for INT; p .0001 for all cases). This suggests
patterns do not necessarily have the same AUCI values. Two that the baseline measurement is a natural measure of total
patterns can have a strong positive correlation and still show waking hormonal concentration. PCA was rerun including the

658 Psychosomatic Medicine 69:651 659 (2007)

ANALYSES OF AWAKENING CORTISOL MEASUREMENTS
baseline measurement in the list of cortisol parameters and the
result was consistent with the previously mentioned conclusion.
The use of AUC is advantageous from both statistical and
biological stand points. Statistically, it simplifies the analyses
by creating a single summary response from multiple mea-
surements and also increases the power of testing without
sacrificing the information contained in the multiple measure-
ments. Biologically, it provides a means to incorporate both
pieces of information (intensity and sensitivity) contained in
repeated measurements into the statistical analyses. The re-
sults of this study (generic definition and interpretation) re-
garding AUCI are applicable in other designs; for example,
when studying cortisol secretion over the course of a day
(diurnal rhythm) or when studying cortisol response after a
specific challenge. One limitation of the study is that it is
based on a small sample size of 68 participants.
Given the high level of intercorrelation among derived
summary measures of cortisol data, it seems prudent to reduce
to a manageable number the set of parameters to be considered
in analyses. This has implications for ease in presentation and
interpretation of analysis results as well as for issues of
multiple testing. We have provided an approach that allows
the number of parameters to be reduced by choosing a smaller
set from those most highly loaded on the first two principal
components.
REFERENCES
1. Pruessner JC, Wolf OT, Hellhammer DH, Buske-Kirschbaum A, Von
Auer K, Jobst S, Kaspers F, Kirschbaum C. Free cortisol levels after
awakening: a reliable biological marker for the assessment of adrenocor-
tical activity. Life Sci 1997;61:2539 49.
2. Schmidt-Reinwald A, Pruessner JC, Hellhammer DH, Federenko I,
Rohleder N, Schurmeyer TH, Kirschbaum C. The cortisol response to
Psychosomatic Medicine 69:651 659 (2007)
awakening in relation to different challenge tests and a 12-hour cortisol
rhythm. Life Sci 1997;64:1653 60.
3. Sephton SE, Sapolsky RM, Kraemer HC, Spiegel D. Diurnal cortisol
rhythm as a predictor of breast cancer survival. JNCI 2000;92:994 1000.
4. Wust S, Wolf J, Hellhammer DH, Federenko I, Schommer N, Kirschbaum C.
The cortisol awakening response-normal values and confounds. Noise Health
2000;7:79 88.
5. Kudielka BM, Kirschbaum C. Awakening cortisol response are influ-
enced by health status and awakening time but not by menstrual cycle
phase. Psychoneuroendocrinology 2003;28:35 47.
6. Buchanan TW, Kern S, Allen JS, Tranel D, Kirschbaum C. Circadian
regulation of cortisol after hippocampal damage in humans. Biol Psychi-
atry 2004;56:651 6.
7. Watamura SE, Donzella B, Kertes DA, Gunnar MR. Developmental
changes in baseline cortisol activity in early childhood: relations with
napping and effortful control. Dev Psychobiol 2004;45:12533.
8. Bhagwagar Z, Hafizi S, Cowen PJ. Increased salivary cortisol after
waking in depression. Psychopharmacology 2005;182:54 7.
9. Neylan TC, Brunet A, Pole N, Best SR, Metzler TJ, Yehuda R, Marmar
CR. PTSD symptoms predict waking salivary cortisol levels in police
officers. Psychoneuroendocrinology 2005;30:373 81.
10. Pruessner JC, Kirschbaum C, Meinlschmid G, Hellhammer DH. Two
formulas for computation of the area under the curve represent measures
of total hormone concentration versus time-dependent change. Psycho-
neuroendocrinology 2003;28:916 31.
11. Kurina LM, Schneider B, Waite LJ. Stress, symptoms of depression and
anxiety, and cortisol patterns in working parents. Stress and Health
2004;20:53 63.
12. Violanti JM, Burchfiel CM, Miller DB, Andrew ME, Dorn J, Wactawski-
Wende J, Beighley CM, Pierino K, Joseph PN, Vena JE, Sharp DS,
Trevisan M. The Buffalo cardio-metabolic occupational police stress
(BCOPS) pilot study: methods and participants characteristics. Ann Epi-
demiol 2006;16:148 56.
13. Geiss A, Varadi E, Steinbach K, Bauer HW, Anton F. Psychoneuroim-
munological correlates of persisting sciatic pain in patients who under-
went discectomy. Neurosci Lett 1997;237:65 8.
14. Schulz P, Kirschbaum C, Pruessner JC, Hellhammer DH. Increased free
cortisol secretion after awakening in chronically stressed individuals due
to work overload. Stress Med 1998;14:917.
15. Pruessner JC, Hellhammer DH, Kirschbaum C. Burnout, perceived stress,
and cortisol response to awakening. Psychosom Med 1999;61:197204.
16. Johnson DE. Applied multivariate methods for data analysts. New York:
Duxbury Press; 1998.
659