Effect of 3 caffeine doses on plasma catecholamines and alertness during prolonged wakefulness

Abstract

Determine the relationship between caffeine, catecholamines, and alertness during prolonged wakefulness.
Following 49 h of prolonged wakefulness, each of 50 healthy males (18-32 years) orally ingested either a placebo or one of three doses of caffeine, 2.1 (low), 4.3 (medium), or 8.6 mg kg-1 body weight (high), in a randomized double-blind design. Wakefulness continued for an additional 12 h during which venous blood samples were collected for catecholamine and caffeine analysis [determined using high-performance liquid chromatography (HPLC)]. A sleep latency test, the Stanford sleepiness scale, and a choice reaction time test were administered periodically during the post-dosing period and served as measures of alertness (physiological, subjective, and behavioral, respectively).
Caffeine had no significant effect on noradrenaline, but adrenaline was significantly increased between 1 h and 4 h post-dosing in the high dose group compared with a placebo group. Following caffeine administration, responses to sleep latency, sleepiness scores, and reaction time scores showed dose-related changes that were exhibited by significant correlation coefficients.
The results indicate that high doses of caffeine have a significant and beneficial effect on alertness during prolonged wakefulness.

Full text

CLINICAL TRIALS
G. H. Kamimori á D. M. Penetar á D. B. Headley
D. R. Thorne á R. Otterstetter á G. Belenky
Effect of three caffeine doses on plasma catecholamines
and alertness during prolonged wakefulness
Received: 9 November 1999 / Accepted: 6 June 2000 / Published online: 3 August 2000
Ó Springer-Verlag 2000
Abstract Objective: Determine the relationship between
caeine, catecholamines, and alertness during prolonged
wakefulness.
Methods: Following 49 h of prolonged wakefulness,
each of 50 healthy males (18±32 years) orally ingested
either a placebo or one of three doses of caeine, 2.1
(low), 4.3 (medium), or 8.6 mg kg
)1
body weight (high),
in a randomized double-blind design. Wakefulness
continued for an additional 12 h during which venous
blood samples were collected for catecholamine and
caeine analysis [determined using high-performance
liquid chromatography (HPLC)]. A sleep latency test,
the Stanford sleepiness scale, and a choice reaction time
test were administered periodically during the post-
dosing period and served as measures of alertness
(physiological, subjective, and behavioral, respectively).
Results: Caeine had no signi®cant eect on noradren-
aline, but adrenaline was signi®cantly increased between
1 h and 4 h post-dosing in the high dose group com-
pared with a placebo group. Following caeine admin-
istration, responses to sleep latency, sleepiness scores,
and reaction time scores showed dose-related changes
that were exhibited by signi®cant correlation coecients.
Conclusion: The results indicate that high doses of caf-
feine have a signi®cant and bene®cial eect on alertness
during prolonged wakefulness.
Key words Caeine á Catecholamines á Sleep latency
Introduction
The demands of modern work and life style fre-
quently require individuals to sacri®ce their regular
sleep schedules resulting in the fragmentation or loss
of signi®cant amounts of sleep. Decrements in cog-
nitive performance, mood, and alertness associated
with prolonged wakefulness have been previously
documented [1±3]. Pharmacologic agents, including
D
-amphetamine and caeine, are often used to alle-
viate these performance decrements. Although
D
-am-
phetamine can eectively counteract the performance
de®cits associated with prolonged wakefulness, its
highly addictive nature and negative physiological side
eects render it unacceptable for general use [4]. In
contrast, caeine is the most widely used over the
counter stimulant in the world and is commonly
found in soft drinks, chocolate, and in combination
with a number of other pharmacological agents [3, 5,
6]. Caeine has been shown to stimulate the release of
the catecholamines, adrenaline, and noradrenaline,
which are associated with physiological arousal and
stress [7].
Although a number of studies have examined the
relationship between caeine, catecholamines, and
alertness, this is the ®rst study to focus on these rela-
tionships during an extended period of prolonged
wakefulness. At the present time, we have been unable to
identify any published studies which have focused on the
use of high doses of caeine (greater than 300 mg) in
conjunction with prolonged wakefulness. We hypothe-
sized that caeine would stimulate speci®c components
of the sympathetic nervous system in a dose-dependent
fashion, thereby altering the noradrenaline or adrenaline
response, and concomitantly increase alertness as mea-
sured using sleep latency (SL), a self-rating sleepiness
scale, and a reaction time test.
Eur J Clin Pharmacol (2000) 56: 537±544
DOI 10.1007/s002280000186
G. H. Kamimori (&) á D. B. Headley á D. R. Thorne
R. Otterstetter á G. Belenky
Department of Neurobiology and Behavior,
Division of Neuropsychiatry,
Walter Reed Army Institute of Research,
Washington, DC 20307-5100, USA
e-mail: gary.kamimori@na.amedd.army.mil
Tel.: +1-301-3199714; Fax: +1-301-3199979
D. M. Penetar
U.S. Army Research Institute of Environmental Medicine,
Natick, MA, USA

Materials and methods
Subjects
Fifty healthy, nonsmoking, males, 18±32 years (mean 23.6 years),
volunteered to participate in this study and signed an informed
consent. All subjects were within acceptable weight limits for their
height, not currently taking any medication, did not normally
consume in excess of 300 mg of caeine per day, and had regular
sleep patterns (6±8 h per night without diculty falling asleep).
Subjects were randomly assigned in a double blind design to
one of four drug groups: placebo, low (2.1 mg kg
)1
), medium
(4.3 mg kg
)1
), or high (8.6 mg kg
)1
). These doses correspond to
150, 300, and 600 mg per 70 kg body weight. Characteristics of the
four groups are presented in Table 1.
This research was conducted in conformity with AR 70±25,
United States Army Medical Research and Development Com-
mand Reg. 70±25 on the use of human volunteers in research, and
the ``Guiding Principles for Research Involving Animals and Hu-
man Beings''. Human volunteers participated after giving free and
informed consent.
Procedure
Subjects arrived in the laboratory in groups of three or four the
evening prior to the initiation of the sleep deprivation period
having refrained from the use of caeine, alcohol, or any phar-
macologic agents during the previous 24 h. Blood and urine sam-
ples were collected and assayed to ensure compliance with these
restrictions.
Electrodes were attached to the scalp and face, using the in-
ternational 10±20 system of electrode placement, for collection of
the multiple sleep latency test (MSLT) data [8]. Electroenceph-
alograms (EEG), electrooculogram (EOG), and sub-mental
electromyograms (EMG) were recorded continuously using an
eight-channel Oxford Medilog ambulatory cassette recorder. Sub-
jects received a standardized diet throughout the study and water
was available ad libitum.
Subjects retired at 2300 hours on day 0 and were allowed to
sleep until 0700 hours the next morning (day 1). They were kept
awake for the following 62 h except for brief periods during ad-
ministration of the MSLT (described below). Measures of mood
and cognitive function were also administered at various times
throughout the study and the results have been reported elsewhere
[3, 9].
Following 48 h of prolonged wakefulness (0700 hours, day 3) a
Te¯on catheter was inserted into a forearm vein and maintained
with a heparin lock (heparin sodium, 20 U cc
)1
). At 0800 hours
(time 0) control samples were collected immediately prior to the
administration of placebo or one of the three doses of caeine
(caeine anhydrous, USP, City Chemical Corporation, NY).
Caeine or placebo was administered orally in 250 ml of an arti-
®cially sweetened lemon juice drink in a double-blind design. Blood
samples were then collected 12 times post-administration, namely
at 0.25, 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 10, and 12 h.
Catecholamines and caeine concentrations
Samples were collected into chilled, heparinized Vacutainers con-
taining glutathione and ethylene glycol-bis(b-aminoethyl ether)-
N,N,N¢,N¢-tetraacetic acid (EGTA), centrifuged at 2000 g for
15 min, the plasma separated into polyethylene tubes, and stored at
)70 °C until analysis. Plasma noradrenaline and adrenaline were
isolated using alumina extraction as previously described [10] and
high-performance liquid chromatography (HPLC) with electro-
chemical detection to determine the plasma catecholamine con-
centrations. The sensitivity of these assays was 10 pg ml
)1
with a
signal-to-noise ratio of 4:1. The within-day (interday) variation was
less than 1% and the between-day (intraday) variation was less
than 3% for standards in the range 5.0±5000 pg ml
)1
. Plasma
caeine concentrations were also determined using HPLC on
samples collected during the same period; the methodology and
pharmacokinetics have been previously reported [6].
Alertness measures
On eight occasions post-dosing (beginning at 1, 2, 3, 4, 6, 8, 10, and
12 h), a series of tests were given. Three alertness measures from
this test battery are reported here, namely the Stanford sleepiness
scale (SSS), a choice reaction time test (CRT), and the MSLT. The
SSS represents a subjective self-rating measure, the CRT a per-
formance-based measure of alertness, and the MSLT a physiologi-
cal measure.
Stanford sleepiness scale
The SSS [11] is a quick-to-administer scale in which subjects self-
rate their current state of alertness by choosing one of seven de-
scriptive statements ranging from 1 (``feeling active and vital; alert;
wide awake'') to 7 (``almost in reverie; sleep onset soon; lost
struggle to remain awake''). Over days 1 and 2, this scale was ad-
ministered periodically on 22 occasions, then given as the ®rst part
of the test battery at the eight times indicated above on day 3.
Choice reaction time test
This computer-based, single-choice reaction time test (a component
of the Walter Reed Performance Assessment Battery [12]) consisted
of the subject matching a digit (from 0 to 9) that appeared on the
monitor by pressing as soon as possible the appropriate numbered
keyboard key. In each administration of this test of the battery, 50
numbers were presented. Dependent variables available from this
test were speed (the reciprocal of mean response time), accuracy
(percentage of correct matches), and throughput (the speed±accu-
racy product, a measure of useful work output [13]). This latter
measure was used for analysis in the present study. Periodically
during day 1 and day 2, this test was administered 22 times, then
given on the post-dosing schedule as indicated above.
Multiple sleep latency test
For each MSLT, subjects were asked to lie on a bed in a darkened,
sound-attenuated room and were instructed to close their eyes,
relax, and allow themselves to fall asleep. EEG, EOG, and EMG
were displayed on a Grass Electroencephalograph (Model 8-10D)
for on-line scoring. Subjects were awakened by an investigator
following the identi®cation of 30 s of stage-2 sleep or the onset of
rapid eye movement (REM) sleep (per standards in Rechtschaen
and Kales [8]). If sleep did not occur within 20 min according to
Table 1 Subject descriptors by
dose group (ranges)
No. in group Age (years) Height (cm) Weight (kg)
Placebo 12 22.1 (19±26) 177.3 (170.2±186.7) 73.8 (64.1±79.5)
Low 13 23.5 (21±31) 180.3 (165.1±203.2) 77.8 (54.1±95.5)
Medium 12 25.2 (20±29) 173.9 (162.6±182.9) 71.7 (61.4±78.2)
High 13 23.3 (18±32) 175.5 (165.1±186.7) 72.8 (64.5±89.1)
538

procedures developed by Carskadon and Dement [14, 15], the test
was terminated. The MSLT measure is de®ned as the elapsed time
from initiation of the test to awakening or termination by an in-
vestigator at 20 min. Twelve tests were conducted intermittently
during the initial 48 h of prolonged wakefulness. The last baseline
MSLT was administered at the 37-h mark (day 2, 2230 hours).
Drug (or placebo) was administered at time 0 (0800 hours, day 3)
and additional MSLTs were administered as the last part of the test
battery during the eight oerings which began at the times indi-
cated above.
Statistical analysis
To scrutinize the time course of the catecholamines, two sets of
analyses of variance were performed on each of the adrenaline- and
noradrenaline-dependent variables. The ®rst analysis was run on
the blood samples taken at 0800 hours (day 3) to see whether there
were any dierences among the four dose groups just before drug
administration. The second analysis looked at group dierences
and the eects of the sleep-deprivation period after dosing. The key
analysis of variance term of interest was the interaction eect. To
guard against a potential positive bias of the F ratio, which can
occur in terms involving a repeated measure, adjustment of degrees
of freedom procedures were adopted [16]. A simple main-eects
analysis was run on each of the 12 time points following a signi®-
cant interaction, and, if this analysis was signi®cant, a comparison
of the four group means against each other was made using the
Newman±Keuls multiple comparison procedure [16].
As a follow-up to analyses of variance on the catecholamine
data, Pearson correlation coecients were run on the catechol-
amine and the caeine serum concentration values (thus the coef-
®cients were run on only the three dose groups). Data pairs were
the matched individual values at each of the observation points.
One correlation involved 12 data pairs per subject (corresponding
to all the 12 post-drug occasions of measurement). A second cor-
relation focused on the time points 15 min through 90 min post-
dosing (early phase), a third on time points 120 min through
240 min (middle phase), and the ®nal set on 360 min through
720 min (late phase). These time blocks were chosen because they
most closely corresponded to the pharmacokinetic pro®le of ab-
sorption, distribution, and elimination for caeine. Each of the
three focused correlations involved four data pairs per subject.
To assess changes in the three alertness measures (MSLT, SSS,
and CRT test), each was correlated with the caeine, adrenaline,
and noradrenaline concentration levels, and also correlated with
each other. Pearson correlations were used for all comparisons.
Correlations were only performed on data from the three groups
that received caeine. Data pairs were the matched individual
values at each of the observation points. An overall correlation
involved 12 data pairs per subject (that is, all the 12 post-drug
occasions of measurement). Three additional correlations were run
that closely corresponded respectively to the pharmacokinetic
pro®le of absorption, distribution, and elimination of caeine: one
focused on the time points 15 min through 90 min post-dosing
(early phase), another on time points 120 min through 240 min
(middle phase), and one on 360 min through 720 min (late phase).
Each of these three correlations involved four data pairs per sub-
ject. A P value less than 0.05 was accepted as signi®cant for all
statistical tests.
Results
Caeine concentrations
It was previously shown by Kamimori et al. [6] that the
three doses used in this study were in fact divergent
enough throughout the time course to cause dose-de-
pendent dierences in caeine pharmacokinetics in these
severely sleep-deprived subjects. Mean serum concen-
trations for each group as a function of time post-dosing
are presented in Table 2.
Catecholamine concentrations
Figure 1 shows the mean epinephrine concentration for
each group just prior to and for 12 h following drug ad-
ministration. At 49 h of prolonged wakefulness (time 0),
there was no signi®cant dierence in epinephrine between
placebo and any of the three drug groups. The overall
analysis of variance for the 12-h dosing period showed a
signi®cant dose group-by-time interaction. Simple main
eects were signi®cant for the six time points between
60 min and 240 min post-dosing. With the exception of
the 150-min and 180-min time points, the ordering of the
four dose groups showed a dose±response pattern. A
consistent ®nding for the six time points from the multiple
comparison tests was that the high dose group was sig-
ni®cantly higher than the placebo group; also the high
dose group's epinephrine was higher than that of the low
dose group for time points 60 min through 150 min.
The correlation of epinephrine with caeine over all
time points was 0.36 (Table 3); for the early phase (time
points 15±90) the value was 0.49; for the middle phase
(time points 120±240), 0.28, and for the late phase (time
points 360±720), 0.11. All but the last correlation were
statistically signi®cant.
The noradrenaline data are illustrated in Fig. 2. Al-
though plasma levels appeared to increase with caeine,
the dose group-by-time interaction was not signi®cant,
and no post-hoc analysis of variance procedures were
performed.
Alertness measures: descriptive statistics
and correlations
The SL measures for the four dose groups closely fol-
lowed each other during the ®rst two days of the study.
Table 2 Serum caeine concentrations (lgml
)1
) as a function of dose group and time post-dosing. Data from Kamimori et al. [6]
Dose
group
Time post-dosing (min)
15 30 60 90 120 150 180 240 360 480 600 720
Low 1.20 2.22 2.60 2.64 2.49 2.36 2.25 2.09 1.69 1.20 0.83 0.66
Medium 2.12 3.78 5.38 5.90 5.46 4.98 4.92 4.52 3.52 2.82 2.27 1.59
High 5.53 8.34 11.75 12.29 11.96 11.57 10.82 9.92 8.72 7.00 5.48 4.21
539

On day 1, averaged values across all subjects (i.e., over
the four dose groups) ranged from 16.3 min to
19.9 min, then dropped on day 2 to a range of 5.6±
7 min (the last overall average before dosing, given at
2030 hours on day 2 was 7.0). After dosing, as shown
in Table 4, the times to reach stage-2 sleep showed
dierences as a function of dose group and time post-
dosing, and indicate a dose±response relationship,
especially during the ®rst 4 h post-dosing. During this
interval, placebo and low dose group latencies contin-
ued the decline from days 1 and 2, whereas the laten-
cies for the medium and high groups showed a
dierential increase; the latencies for the high group
were 2.0±3.4 times higher than those for placebo, and
medium group latencies were some 1.3±2.6 times higher
than placebo.
Fig. 1 Mean adrenaline con-
centration just prior to and for
12 h following drug administra-
tion. *Signi®cantly dierent
from placebo, P £ 0.05
ADR NORADR
All points
(409±421)
Early phase
(138±144)
Middle phase
(135±141)
Late phase
(136±143)
All points
(409±421)
Early phase
(138±144)
Middle phase
(135±141)
Late phase
(136±143)
CAF 0.36* 0.49* 0.28* 0.11 )0.06 )0.05 )0.14 0.03
ADR ± ± ± ± )0.02 )0.06 )0.04 )0.12
*Correlation is signi®cantly dierent from zero at the 0.05 level or less
Table 3 Correlations of catecholamines with caeine serum
concentrations and with each other. All correlations involve three
dose groups only (placebo group did not receive caeine). Early
phase 15±90 min; middle phase 120±240 min; late phase 360±
720 min; (number of data pairs per correlation); CAF caeine;
ADR adrenaline; NORADR noradrenaline
Fig. 2 Mean noradrenaline
concentration just prior to and
for 12 h following drug admin-
istration
540

The SSS values likewise were similar for the four
groups at each time point during day 1 and day 2. Av-
erages over all subjects on day 1 ranged from 1.6 to 4,
and on day 2 from 3.1 to 4.8. The last value pre-dosing,
at 0600 hours, was 4.8. During the interval 60±240 min
post-dosing, the averages ranged from 0.5 to 0.8 of the
placebo groups.
Mean throughput values on the CRT tests at each
time point were similar for the four groups during day 1
and day 2. On day 1, the means ranged from 77 to 84,
and on day 2 from 72 to 85; the last average before
dosing, gathered at 0600 hours on day 3, was 62. The
post-dosing averages for each group are shown in
Table 4. Within each of the eight administrations, the
average value of throughput in each of the three dose
groups is greater than the placebo value. The highest
value at each time point is in the high-dose group;
these values range from 1.1 to 1.4 times greater than
placebo value.
Correlation coecients were used as a convenient
summary statistic to compare alertness measures both
with each other as well as with caeine and catechol-
amine serum concentration levels during the post-dosing
period. The results are listed in Table 5 [1].
For the correlations involving caeine over all the
time points, all three alertness measures are signi®cantly
associated with this stimulant in the hypothesized di-
rection, although SL and the performance-based reac-
tion time measure appear to predict somewhat better
than the subjective SSS measure. Techniques for statis-
tically comparing two correlations for the present type
of design are not well developed, and thus descriptive
rather than analytical comparisons are made. When
looked at as a function of early versus late phase, the
association for CRT holds up over both phases, whereas
SL's relationship is in the early phase only [2].
For the catecholamines, signi®cant relationships are
found with epinephrine correlating with SSS and SL; a
signi®cant association of SSS with adrenaline was found
over both phases of the dosing period. No predictability
is apparent with noradrenaline and the subjective SSS
measure, but some relationship is seen with the physio-
Table 5 Correlations of alertness measures with catecholamines
and caeine serum concentrations. All correlations involve three
dose groups only. A minus sign for a correlation of SSS with an-
other variable implies that subjective alertness increased as serum
concentration or sleep latency increased (values at the low end of
the SSS indicate higher alertness). CAF caeine; ADR adrenaline;
NORADR noradrenaline; SSS Stanford sleep scale; SL sleep la-
tency; CRT choice reaction time; all time points 60±720 min post-
dosing; early phase 60±240 min post-dosing; late phase 360±
720 min post-dosing
All time points (275±304 data
pairs/correlation)
Early phase (136±152 data
pairs/correlation)
Late phase (139±152 data
pairs/correlation)
SSS SL CRT SSS SL CRT SSS SL CRT
CAF )0.18* 0.30* 0.30* )0.08 0.34* 0.29* )0.15 0.08 0.28*
ADR )0.25* 0.18* )0.01 )0.19* 0.16 0.02 )0.27* 0.09 )0.10
NORADR 0.05 0.13* )0.12* 0.01 0.21* )0.19* 0.06 0.05 )0.05
SSS ± )0.15* )0.29* ± )0.11 )0.29* ± )0.12 )0.26*
SL ± ± )0.02 ± ± ) 0.07 ± ± 0.02
*Correlation is signi®cantly dierent from zero at the 0.05 level or less
Dose group Time post-dosing (min)
60 120 180 240 360 480 600 720
Sleep latency (min)
Placebo 5.0  3.8 4.0  2.4 3.9  2.3 3.0  1.6 4.5  4.6 4.6  3.4 4.0  2.5 7.6  4.8
Low 4.8  4.7 5.0  2.8 4.1  2.1 3.6  1.6 3.6  1.6 3.4  2.1 3.9  2.4 4.6  3.8
Medium 6.6  3.0 7.4  4.4 8.3  5.1 7.9  4.6 5.8  3.1 5.8  3.8 6.1  3.8 9.3  5.9
High 10.2  6.9 10.0  6.2 9.9  6.6 10.2  6.2 5.9  3.9 4.7  2.5 6.6  3.6 8.3  6.0
Stanford sleepiness scale
Placebo 3.5  1.6 3.8  1.3 3.7  1.2 3.2  1.5 2.9  0.7 2.8  0.8 2.9  0.9 2.8  0.8
Low 2.2  1.2 2.2  1.4 2.5  1.1 2.8  1.3 3.0  1.2 3.3  1.4 3.5  1.1 3.1  1.0
Medium 2.4  1.4 2.7  1.5 2.4  1.4 3.2  1.7 3.3  1.5 3.1  1.5 3.0  1.3 3.3  1.6
High 1.9  0.8 2.2  0.9 2.5  1.2 2.5  1.1 2.5  1.0 2.8  1.2 2.6  0.8 2.3  0.8
Choice reaction time
Placebo 63.1  14.0 65.3  26.1 68.0  22.0 63.1  21.7 70.6  22.3 64.3  22.1 68.6  24.2 75.9  22.6
Low 81.2  24.0 84.4  25.8 84.3  19.2 78.9  22.5 78.2  23.7 69.4  22.1 72.4  24.7 79.9  19.6
Medium 77.5  19.3 82.5  18.2 80.4  19.4 78.3  22.5 75.9  24.1 75.4  19.3 81.8  16.6 78.1  17.1
High 88.5  17.0 87.7  18.6 90.7  17.6 89.5  13.1 87.3  16.0 85.1  16.3 85.2  17.1 87.1  16.0
Table 4 Average alertness measure values (SD) by dose group
and time post-dosing. The number of values per average was either
12 or 13. Low dose 2.1 mg kg
)1
; medium dose 4.3 mg kg
)1
; high
dose 8.6 mg kg
)1
; Stanford sleepiness scale rating from 1 ``feeling
alert, wide awake'' to 7 ``almost in reverie''; choice reaction time
throughput ± a measure of both accuracy and speed
541

logical alertness measure SL in the early phase [3]. For
the alertness measures correlated with each other, SL
and CRT show a signi®cant correlation with SSS over
all the time points. The CRT association with SSS is
subjectively greater than SL, and occurs in both phases.
Discussion
Many segments of society put in extended workdays.
Whether the tasks involve such endeavors as driving,
studying for exams, projects at the oce, or in extreme
cases performing long-term emergency disaster relief or
military operations, a stimulant such as caeine is often
taken to maintain alertness. The results of this study
demonstrate that following 49 h of prolonged wakeful-
ness, ingestion of moderate and high doses of caeine
signi®cantly increase plasma adrenaline and alertness in
a dose-related fashion.
To our knowledge, this is the only study in which
relatively high doses of caeine (8.6 mg kg
)1
or 600 mg
70 kg
)1
) have been administered in conjunction with an
extended period of prolonged wakefulness (49 h baseline
plus 12 h post-dosing). The duration of prolonged
wakefulness chosen for this study was based on our
previous studies, in which we have reported signi®cant
decreases in both cognitive performance and alertness
following 48 h of prolonged wakefulness [2±4]. The de-
sign allowed for both charting of the time course of the
catecholamines plus correlating their values with our
three separate alertness measures (subjective, behavioral,
and physiological).
There were no signi®cant dierences between groups
in the control concentrations (at the 49-h mark) of either
noradrenaline or adrenaline, indicating that prolonged
wakefulness per se had no signi®cant aect on the basal
circulating catecholamine levels. The caeine adminis-
tration, given as a bolus at the 49-h mark, resulted in
dose-dependent serum levels (Table 2) and stimulated an
adrenaline response (best de®ned during the 15- to 90-
min block; Fig. 1, Table 3). As caeine levels decreased
in the late phase, their in¯uence on adrenaline stimula-
tion greatly decreased. Although mean noradrenaline
levels appear to be increased in the caeine groups, there
was no signi®cant dierence in comparison with placebo
concentrations. These results are consistent with other
caeine studies that have reported a signi®cant increase
in the resting levels of adrenaline, but observed no sig-
ni®cant eect on noradrenaline [17±19]. These ®ndings
support the contention that caeine acts predominantly
at the adrenomedullary level rather than through un-
dierentiated sympathetic stimulation (e.g., postural
changes).
The more meaningful associations were seen during
the ®rst 4 h post-dosing. For example, in the case of SL,
the high dose resulted in a time-to-stage-2 SL twice that
of the placebo group at the 60-min mark and maintained
at least this dierence throughout the 240-min mark.
The highest value of 10.2 min, occurring in the high dose
group, is similar to the mean SL value found by Walsh
et al. [20] after approximately 20 h of wakefulness in
subjects given 4.0 mg kg
)1
caeine (essentially our me-
dium dose). In comparison with a study using the same
design but
D
-amphetamine instead of caeine, New-
house et al. [4] were able to reverse the eects of pro-
longed wakefulness to almost 100% of that seen in
rested conditions; these eects began 2 h post-dosing
and signi®cant dierences in SL compared with control
groups continued for an additional 5 h.
It is important to note the short SLs in the low-dose
group. Although plasma adrenaline was increased in the
low group (Fig. 1), the SLs suggest that the low dose
had no eect on alertness. These data can be explained
by the threshold eect, which has been previously as-
sociated with caeine. In this type of a relationship, a
minimum blood concentration of a drug is required to
elicit a signi®cant pharmacodynamic response, in this
case from the adrenal medulla (i.e., adrenaline). A sec-
ond possible explanation may be related to each indi-
vidual's tolerance to caeine. Subjects were screened for
the consumption of 4.3 mg kg
)1
of caeine or less per
day, so it is possible that participants habitually ingested
more than 2.1 mg kg
)1
. It is well established that
chronic caeine use will result in an increase in tolerance
to both its physiological and psychological eects [21].
The ineectiveness of the low dose on alertness could
also be explained by a subject's habitual use of caeine,
which would lead to an increase in tolerance to the drug.
Lastly, the high individual variability common with
epinephrine response may also aect SL.
At the end of the study, most dose groups' scores
suggested a slight increase in alertness (Table 4). This
may have been partially caused by an end-spurt phe-
nomenon, as subjects were aware of the ensuing release
from the study.
Because of the time-locked nature of the individual
biochemical and alertness measures over the 49-h post-
dosing period, a correlation approach was used to ex-
amine the relationships. Those that were statistically
signi®cant formed the basis for the discussion above. It
may also be asked whether the correlations are practi-
cally signi®cant, that is, are the strengths of the rela-
tionships meaningful within the intended applied
context of our research program. A useful concept that
conveys practical signi®cance is that of eect size, the
degree to which the null hypothesis is false, or the degree
to which the phenomenon is present in the population
[22]. Cohen has established guidelines for categorizing
the eects as small, medium, or large. For product-
moment correlations, a value between 0.10 and 0.29 is
considered a small eect; if between 0.30 and 0.49, me-
dium, and if 0.50 or greater, large. Using this scheme, of
the 21 statistically signi®cant correlations presented in
Table 3 and Table 5, the eect sizes of ®ve are medium
and the rest are small. It is noteworthy that trends did
appear in spite of the within-subject and between-subject
variability inherent in our six measures. Depending
on the measure, the variability is a by-product of
542

pharmacokinetics, fatigue state (both physiological and
perceived), time of day (diurnal rhythm eect), and
motivation.
This study is the ®rst phase of a research program to
investigate caeine's eects on alertness, and the rela-
tionships we did ®nd are suggestive enough to help guide
the course of study. One experiment is currently in
progress and another is planned to look at an alternative
means of administering caeine, namely through caf-
feine-containing chewing gum as a distribution medium.
We will be investigating dierent dosages, their time
course, their initial alertness eect in sleep-deprived
subjects, as well as the ecacy of ``maintenance dosing'',
and will compare these measures with the eects from
caeine pills.
As caeine is a stimulant, it is important to note the
possible side eects associated with its use. The eects of
moderate doses of caeine (300±600 mg) on the car-
diovascular system have been extensively examined. The
administration of doses of 250±350 mg may produce a
small decrease in heart rate and modest increases in both
the systolic and diastolic blood pressure, but this re-
sponse may not be seen in habitual caeine users [23]. At
higher concentrations (greater than 300 mg) caeine will
result in a signi®cant tachycardia and sensitive individ-
uals may experience other arrhythmias. However, in a
study of 22 patients with a history of symptomatic
nonsustained ventricular tachycardia, ventricular
tachycardia, or ventricular ®brillation, Chelsky et al.
[24] administered 275 mg caeine and found no signi®-
cant alteration in the inducibility or severity of ar-
rhythmias. Newcombe et al. [25] administered
approximately 500 mg to 34 normal subjects and found
no increase in ventricular arrhythmias on a 24-h Holter
recording. Myers and Harris [26] reported similar results
in 35 patients who had had a myocardial infarction and
received 450 mg caeine. They reported a very low
likelihood of caeine being associated with an increase
in ventricular ectopy. These studies suggest that al-
though a rare patient may possess a particular suscep-
tibility to caeine, which could lead to the development
of a cardiac arrhythmia, in general, the ingestion of a
moderate amount of caeine (about 500 mg) is unlikely
to cause an arrhythmia. The administration of high
doses of caeine has also been associated with a number
of side eects, including nausea, dizziness, muscle
tremors, nervousness, restlessness, and insomnia [5]. The
possibility that manual dexterity would be aected by an
increase in muscle tremor did not appear to be a factor
in this study as performance was improved, as opposed
to degraded, with the administration of the high dose of
caeine.
In conclusion, this study quanti®es the eects of
caeine upon three measures of alertness in severely
sleep-deprived subjects. The high dose of caeine
(8.6 mg kg
)1
) was eective in stimulating the release of
adrenaline and improving alertness, as de®ned for ex-
ample by the physiological measure of SL, whose values
were 2.0±3.4 times that of a baseline condition (i.e., no
caeine) over the ®rst 4 h post-dosing. Although these
eects were not as eective as
D
-amphetamine, caeine
is still considered to be the stimulant of choice as it is a
universally available, legal, socially accepted, and widely
used stimulant with a low toxicity and abuse potential.
Acknowledgements We would like to acknowledge the assistance
of the members of the Department of Neurobiology and Behavior
for their assistance in the collection of these data. Use of trade
names does not constitute endorsement of product. The opinions
or assertions contained herein are the private views of the authors
and are not to be construed as ocial or re¯ecting the opinions of
the Department of the Army or the Department of Defence.
References
1. Hartmann E, Orzack MH, Branconnier R (1977) Sleep depri-
vation de®cits and their reversal by
D
- and l-amphetamine.
Psychopharmacology 53: 185±189
2. McCann UD, Penetar DM, Shaham Y, Thorne DR, Gillin JC,
Sing HC, Thomas MA, Belenky G (1992) Sleep deprivation
and impaired cognition possible role of brain catecholamines.
Biol Psychiatry 31: 1082±1097
3. Penetar D, McCann U, Thorne D, Kamimori G, Galinski C,
Sing H, Thomas M, Belenky G (1993) Caeine reversal of sleep
deprivation eects on alertness and mood. Psychopharmacol-
ogy 112: 359±365
4. Newhouse PA, Belenky G, Thomas M, Thorne D, Sing HC,
Fertig J (1989) The eects of
D
-amphetamine on arousal,
cognition, and mood after prolonged total sleep deprivation.
Neuropsychopharmacology 2: 153±164
5. Rall TW (1985) Central nervous system stimulants: the meth-
ylxanthines. In: Gilman AG, Goodman LS, Rall TW, Murad F
(eds) Goodman and Gilman's the pharmacological basis of
therapeutics, 7th edn. Macmillan Publishing Co., New York
6. Kamimori G, Lugo SI, Penetar DM, Chamberlain AC, Brun-
hart GE, Brunhart AE, Eddington ND (1995) Dose-dependent
caeine pharmacokinetics during severe sleep deprivation in
humans. Int J Clin Pharmacol Ther 32: 182±186
7. Nehlig A, Duval J-L, Debry G (1992) Caeine and the central
nervous system: mechanisms of action, biochemical, metabolic,
and psychostimulant eects. Brain Res Brain Res Rev 17:
139±170
8. Rechtschaen A, Kales A (1968) A manual of standardized
terminology techniques and scoring system for sleep stages of
human subjects. National Institutes of Health Publication 204,
Government Printing Oce, Washington, DC
9. Penetar DM, McCann U, Thorne D, Schelling A, Galinski C,
Sing H, Thomas M, Belenky G (1994) Eects of caeine on
cognitive performance, mood, and alertness in sleep-deprived
humans. In: BM Marriott (ed) Food components to enhance
performance. National Academy Press, Washington, pp
407±431
10. Schneider DA, Kamimori GH, Wu SY, McEniery MT, Solo-
mon C (1995) Plasma catecholamine and ventilatory responses
to cycling after propranolol treatment. Med Sci Sports Exerc
27: 1616±1620
11. Hoddes E, Zarcone V, Smythe H, Phillips R, Dement WC
(1973) Quanti®cation of sleepiness: a new approach. Psycho-
physiology 10: 431±436
12. Thorne DR, Genser SG, Sing HC, Hegge F (1985) The Walter
Reed performance assessment battery. Neurotoxicol Teratol 7:
415±418
13. Thorne DR, Genser SG, Sing HC, Hegge FW (1983) Plumbing
human performance limits during 72 hours of high task load.
In: Proceedings of the 24th Defense Research Group Seminar
on the human as a limiting element in military systems. Defense
and Civil Institutes of Environmental Medicine, Toronto,
Canada, pp 17±40
543

14. Carskadon NA, Dement WC (1979) Eects of total sleep loss
on sleep tendency. Percept Mot Skills 48: 495±506
15. Carskadon NA, Dement WC (1981) Cumulative eects of
sleep restriction on daytime sleepiness. Psychophysiology 18:
107±113
16. Kirk RE (1982) Experimental design: procedures for the
behavioral sciences, 2nd edn. Brooks Cole, Paci®c Grove
17. MacNaughton KW, Sathasivam P, Graham TE (1990) The
in¯uence of caeine on metabolic responses of men at rest in 28
and 5 °C. J Appl Physiol 68: 1889±1896
18. Robertson D, Ju
È
rgen C, Fro
È
lich MD, Carr RK, Watson JT,
Holli®eld JW, Shand DG, Oates JA (1978) Eects of caeine
on plasma renin activity, catecholamines and blood pressure.
N Engl J Med 298: 181±186
19. Van Soeren MH, Sathasivam P, Spriet LL, Graham TE (1993)
Caeine metabolism and epinephrine responses during exercise
in users and nonusers. J Appl Physiol 75: 805±812
20. Walsh JK, Muehlbach MJ, Humm TM, Stokes Dickins Q,
Sugerman JL, Schweitzer PK (1990) Eect of caeine on
physiological sleep tendency and ability to sustain wakefulness
at night. Psychopharmacology 101: 271±273
21. Griths RR, Mumford GK (1995) Caeine-A drug of Abuse?
In: Bloom FE, Kupfer DJ (eds) Psychopharmacology: the
fourth generation of progress. Raven, New York
22. Cohen J (1988) Statistical power analysis for the behavioral
sciences, 2nd edn. Lawrence Erlabaum, Hillsdale
23. Sera®n WE (1996) Drugs used in the treatment of asthma In:
Gilman AG, Goodman LS, Rall TW, Murad F (eds) Goodman
and Gilman's the pharmacological basis of therapeutics, 9th
edn. McGraw-Hill, New York
24. Chelsky LB, Cutler JE, Grith K, Kron J, McClelland JH,
McAnulty JH (1990) Caeine and ventricular arrhythmias.
JAMA 264: 2236±2240
25. Newcombe PF, Renton KW, Rautaharju PM, Spencer CA,
Montague TJ (1988) High-dose caeine and cardiac rate and
rhythm in normal subjects. Chest 94: 90±94
26. Myers MG, Harris L (1990) Caeine and ventricular arrhyth-
mias. Can J Cardiol 6: 95±98
544