and Women
Terry R. Hartley, PhD, William R. Lovallo, PhD, and Thomas L. Whitsett, MD
Caffeine increases blood pressure (BP). In men, acute BP men (4.1 and 3.8 mm Hg, respectively). Men given
elevations after caffeine intake are due to an increase in caffeine versus placebo showed the expected elevation
vascular resistance, with no change in cardiac output. in vascular resistance throughout the remainder of the
The hemodynamic effects of caffeine have not been protocol (p <0.001), with no difference in cardiac out-
studied in women. Accordingly, BP and hemodynamic put. In contrast, women responded to caffeine with in-
responses to caffeine were measured in a double-blind creases in stroke volume (p <0.001) and cardiac output
trial comparing age-matched men and women at rest (p <0.001), with no difference in vascular resistance
and during mental stress. Caffeine (3.3 mg/kg, equiva- from women taking placebo. Men and women have
lent to 2 to 3 cups of brewed coffee) or placebo was similar BP responses to caffeine, but the BP responses
given to separate groups of women (n ⴝ 21 and 21) and may arise from different hemodynamic mechanisms.
men (n ⴝ 16 and 19) (mean ages 29 and 27 years,
Women who consume a dietary dose of caffeine showed
respectively). BP, cardiac output, and vascular resistance
an increase in cardiac output, whereas men showed
were observed at rest, during a stressful public-speaking
increased vascular resistance. 䊚2004 by Excerpta
simulation, reading aloud, and recovery. Caffeine
Medica, Inc.
caused nearly identical systolic and diastolic BP eleva-
tions in women (4.5 and 3.3 mm Hg, respectively) and (Am J Cardiol 2004;93:1022–1026)
1022 ©2004 by Excerpta Medica, Inc. All rights reserved. 0002-9149/04/$–see front matter
The American Journal of Cardiology Vol. 93 April 15, 2004 doi:10.1016/j.amjcard.2003.12.057
(Waveshell, Center for Biomedical Engineering, Re-
TABLE 1 Subject Characteristics*
search Triangle Institute, Research Triangle Park,
Variables Women (n ⫽ 42) Men (n ⫽ 35) North Carolina).
Age (yrs) 29 (0.98) 27 (0.79) Cardiovascular variables were systolic BP, dia-
Height (cm) 167 (4.8) 181 (9.0)† stolic BP, mean arterial pressure, and pulse pressure
Weight (kg) 64 (1.7) 80 (1.7)† (systolic BP ⫺ diastolic BP) in millimeters of mer-
Quetelet index (g/cm2) 2.44 (0.042) 2.28 (0.054)‡ cury; heart rate in beats per minute; stroke volume in
Caffeine intake (mg/d) 154 (10.6) 167 (9.9)
milliliters; cardiac output (stroke volume ⫻ heart rate)
*Entries show mean (SEM); †p ⬍0.001; ‡p ⬍0.05. in liters per minute; total peripheral
resistance (mean arterial pressure ⫻
80/cardiac output) in dynes per sec-
TABLE 2 Baseline Cardiovascular Values* ond per centimeters to the fifth
power and a vascular compliance in-
Caffeine Placebo Gender Caffeine
dex (stroke volume/pulse pressure)
Heart rate (beats/min) in milliliters per millimeter of mer-
Women 66 (1.6) 67 (1.7) NS NS cury.
Men 67 (2.6) 62 (1.9)
Systolic BP (mm Hg) BP was measured every 2 min-
Women 104 (1.3) 107 (1.6) .001 NS utes throughout the study. Imped-
Men 114 (2.0) 116 (1.6) ance data were recorded continu-
Diastolic BP (mm Hg) ously and then averaged for 12 time
Women 62 (1.9) 66 (1.6) NS NS
Men 64 (1.4) 65 (1.7)
periods: baseline (10 minutes), caf-
Stroke volume (ml) feine or placebo response (15, 30,
Women 69 (4.3) 71 (5.0) NS NS and 45 minutes after taking the
Men 68 (5.8) 80 (7.7) drug), task I (preparation and task)
Cardiac output (L/min) and recovery periods 1 and 2 (15
Women 4.6 (1.1) 4.8 (1.5) NS NS
Men 4.6 (1.5) 5.1 (2.4) minutes each), and task II (prepara-
Peripheral resistance (dyne · s⫺1 · cm⫺5) tion and task) and recovery periods 1
Women 1,424 (84) 1,518 (161) NS NS and 2 (15 minutes each). Responses
Men 1,637 (143) 1,612 (187) to caffeine and the tasks were tested
*Entries show means (SE); n ⫽ 42 women, 21 in caffeine and 21 in placebo groups, and n ⫽ 35 men, as changes from baseline in the sub-
16 in caffeine and 19 in placebo groups. Comparisons are based on gender ⫻ drug group ANOVAs. sequent 11 periods.
Baseline activity during the pre-
drug period was tested using 2 gen-
lasted 3 hours. The protocol included instrumentation der ⫻ 2 drug groups analyses of variance (ANOVAs).
(20 minutes), adaptation (30 minutes), baseline (10 To characterize gender differences in response to caf-
minutes), caffeine or placebo drink (5 minutes), drug feine versus placebo, we performed a multivariate
absorption (45 minutes), task I (reading or speaking, 6 ANOVA on each dependent variable comparing 2
minutes), recovery (30 minutes), task II (alternate gender ⫻ 2 drug groups ⫻ 11 periods after taking the
task, 6 minutes), and recovery (30 minutes). drug. Significant main effects or interactions were
Tasks included reading aloud versus public speak- followed by univariate ANOVAs and specific con-
ing, and task order was counterbalanced across sub- trasts as indicated.
jects. Public speaking causes anxiety and increases
BP, heart rate, and stress hormones.17 The subject was
given a topic and spent 3 minutes preparing and 3 RESULTS
minutes delivering a speech to a video camera in front Table 1 presents anthropometric and screening
of 2 experimenters wearing white coats. The control data. Men weighed more and had a higher Quetelet
task consisted of 3 minutes studying and 3 minutes index than women (F [1,75] ⫽ 41.35 and 5.86, re-
reading aloud a neutral passage from Readers’ Digest spectively; p ⬍0.02). Table 2 provides cardiovascular
while alone. data before drug administration. Men had higher sys-
All cardiovascular measurements were made as the tolic BP at rest than women (F [1,73] ⫽ 29.20, p
subject sat semirecumbent in a recliner chair. BP was ⬍0.0001).
measured with a Dinamap monitor (Critikon, Tampa, The primary gender ⫻ drug ⫻ periods multivariate
Florida). Stroke volume and systolic time intervals ANOVAs showed that the caffeine group had higher
were recorded by an impedance cardiograph (model systolic and diastolic BPs than the placebo group
304B, Minnesota Impedance Cardiograph, Minneap- (main effects of drug, F [11,59] ⫽ 2.30 and 3.28 and
olis, Minnesota) according to previously described p ⬍0.02 and 0.001, respectively). Women had higher
methods.18 Impedance cardiography is a noninvasive levels of stroke volume and cardiac output than men
technique appropriate for behavioral research, and it is (main effects of gender, F [11,59] ⫽ 2.04 and 2.16 and
reliable for within-day and between-day measure- p ⬍0.05 and 0.03, respectively). Men in the caffeine
ments if electrode placement is consistent.9,19 Imped- group had the highest levels of total peripheral resis-
ance signals and electrocardiograms were ensemble tance (F [11,56] ⫽ 2.03, p ⬍0.05). These analyses
averaged20 and analyzed by proprietary software were followed by separate univariate ANOVAs on