Changes in Cortisol and Dehydroepiandrosterone in
Women Victims of Physical and Psychological Intimate
Partner Violence
Maria A. Pico-Alfonso, M. Isabel Garcia-Linares, Nuria Celda-Navarro, Joe Herbert, and
Manuela Martinez
Background: Although intimate partner violence (IPV) has a great impact on women’s health, few studies have assessed the
consequences on physiologic responses.
Methods: Women abused by their intimate male partners either physically (n ⫽ 70) or psychologically (n ⫽ 46) were compared with
nonabused control women (n ⫽ 46). Information about sociodemographic characteristics, smoking, pharmacologic treatment,
lifetime history of victimization (childhood and adulthood), and mental health status (depression, anxiety, and posttraumatic stress
disorder, PTSD) was obtained through structured interviews. Saliva samples were collected at 8 am and 8 pm for 4 consecutive days
to determine morning and evening basal levels of cortisol and dehydroepiandrosterone (DHEA).
Results: Women who were victims of IPV had more severe symptoms of depression, anxiety, and incidence of PTSD and higher levels
of evening cortisol and morning and evening DHEA compared with control women. Intimate partner violence was the main factor
predicting the alterations in hormonal levels after controlling for age, smoking, pharmacologic treatment, and lifetime history of
victimization. Mental health status did not have a mediating effect on the impact of IPV on hormonal levels.
Conclusions: This study shows that both physical and psychological IPV have a significant impact on the endocrine systems of women.
Key Words: Anxiety, cortisol, dehydroepiandrosterone, depression, intimate partner violence, posttraumatic stress disorder
M
ost studies on the impact of intimate partner violence
(IPV) on women’s health focus on mental health, with a
low proportion assessing its impact on physical health
(reviewed in Martinez et al 2003). Regarding mental state, many
studies show that abused women have increased prevalence of
depression, posttraumatic stress disorder (PTSD), and chronic
anxiety states (Campbell 2002; Campbell et al 1996; Golding
1999; Martinez et al 2003; Woods 2000). Studies on physical
health show increased psychosomatic symptoms (e.g., back
pain) and higher incidence of chronic illnesses (e.g., gastrointestinal and other complaints) (Campbell et al 2002; Coker et al
2000; Hamberg et al 1999; Lown and Vega 2001).
To date, there are only two studies on the consequences of
IPV on physiologic responses such as the endocrine and
immune systems, even though these may have considerable
implications for the health status of women (Constantino et al
2000; Seedat et al 2003). The only study on endocrine changes
in women victims of IPV reports alterations in the hypothalamic-pituitary-adrenal function, with these women having
lower morning levels of cortisol than control women (Seedat
et al 2003). Other victims of violence also show alterations in
cortisol levels (De Bellis et al 1994). For example, sexually or
physically abused children have altered (higher or lower)
basal cortisol levels in the morning (Cicchetti and Rogosch
2001; King et al 2001), and women with a prior history of rape
From the Department of Psychobiology (MAPA, MIGL, NCN, MM), University
of Valencia, Valencia, Spain; and Department of Anatomy and
Cambridge Centre for Brain Repair (JH), University of Cambridge, Cambridge, England.
Address reprint requests to Dr. Manuela Martinez, Department of Psychobiology, Faculty of Psychology, University of Valencia, Avda Blasco Ibañez,
21, 46010 Valencia, Spain; E-mail: [email protected].
Received December 3, 2003; revised February 18, 2004; revised May 11,
2004; accepted June 1, 2004.
0006-3223/04/$30.00
doi:10.1016/j.biopsych.2004.06.001
may respond to a further rape with a lower cortisol response
than those experiencing rape for the first time (Resnick et al
1995; Yehuda et al 1998).
The assessment of the impact of IPV on women’s health has
focused on physical and sexual violence, with few studies paying
attention to psychological violence alone despite the finding that
this type of violence can be as devastating as, if not more so, than
physical violence (Aguilar and Nightingale 1994; Coker et al
2000; Follingstad et al 1990; O’Leary 1999).
Major depression (MDD) is associated with alterations in both
cortisol and dehydroepiandrosterone (DHEA; Goodyer et al
1996; Michael et al 2000; Parker et al 2003; Sachar et al 1985).
Many studies, carried out primarily on inpatients, demonstrate
increased evening levels of cortisol in MDD (Gold et al 1988;
Parker et al 2003; Rubin et al 1996). Altered steroid levels may not
only be state-dependent but may also represent a risk factor for
subsequent onset of MDD. For example, higher cortisol levels in
the morning in nondepressed adult women increased the risk of
MDD following an adverse life event (Harris et al 2000). On the
other hand, PTSD has been associated with low levels of cortisol
in men (Yehuda 2002; Yehuda et al 1996), and comorbid
depression and PTSD has been associated with lower levels of
cortisol than MDD alone (Oquendo et al 2003). Studies have
found DHEA to be both lower and higher in depressed compared with control subjects (Goodyer et al 1996; Heuser et al
1998; Michael et al 2000); however, evidence from both adults
and adolescents suggests that the current psychosocial environment has no discernible effect on either cortisol or DHEA
(Goodyer et al 1996; Harris et al 2000). Whether this applies to
other, less normative situations, such as recent or current IPV,
has not been extensively studied, yet such changes, if they occur,
might have considerable implications for the current or future
health of women victims.
The aim of this study was twofold: to assess the impact of IPV
on basal cortisol and DHEA levels in adult women and to
determine whether there was any distinction between those
women exposed to both physical and psychological violence
and those experiencing only the latter.
BIOL PSYCHIATRY 2004;56:233–240
© 2004 Society of Biological Psychiatry
234 BIOL PSYCHIATRY 2004;56:233–240
Methods and Materials
Subjects
This study is part of a larger research project on the impact
of IPV on women’s health, carried out between 2000 and 2002
in a sample of 182 women from the Valencian community of
Spain (Garcia-Linares et al, unpublished data). Women victims
of IPV were recruited from 24-hour centers for helping
women, located in the three provinces of the community
(Alicante, Castellon, and Valencia). Control women, who lived
in a nonviolent intimate partner relationship, were contacted
through women’s clubs. All participants were of Spanish
nationality. The study was approved by the University of
Valencia research ethics committee, and informed written
consent was obtained from all participants at the outset.
Procedure
The study consisted of a structured interview in which four
trained female psychologists asked women about their lives and
health. In general, each woman was interviewed 4 – 6 times by
the same psychologist, with each session taking 1.5 hours.
Additionally, women were asked to provide saliva samples for
hormonal assays.
Questionnaires
A comprehensive questionnaire was designed for a face-toface interview. The majority of questions were devised to collect
objective reports of facts. The questionnaires from which information for the study was obtained are described as follows.
Control Variables. Information about the age, level of education, incidence of smoking, and pharmacologic treatment
(antidepressants, benzodiazepines, estrogens and progestagens,
glucocorticoids, and lithium) was obtained.
Violence Perpetrated by an Intimate Male Partner. A questionnaire was constructed to obtain detailed information about
the different types of violence (physical, sexual, and psychological) perpetrated by the batterer. Each type of violence consisted
of one or more of the acts described in the following paragraphs.
Women were asked to answer yes or no to the incidence of each
act.
1. Physical violence included punches, kicks, slaps, pushes,
bites, and strangling.
2. Sexual violence included forced heterosexual sex (vaginal
or anal penetration, oral sex from female to male partner or
from male to female partner, objects inserted in vagina or
anus); forced homosexual sex, sex with animals, prostitution, or public sex; physical violence during sexual intercourse (bites, kicks, blows, and slaps); threats to hit the
woman or children if rejecting sex, including threats with
knives, guns, or other weapons; involvement of children in
forced sex or children’s witnessing of sexual attacks; and
use of pornographic films and photographs.
3. Psychological violence included verbal attacks (insults,
humiliations), control and power (isolation from family and
friends; impeded decision making, economic abandonment), pursuit and harassment, verbal threats (woman and
family’s life threatened, threats regarding the custody of
children, intimidating phone calls), and blackmail (economic or emotional).
Control women were also asked the questions to ensure that
they had had no experience of violence in any intimate partner
relationship.
www.elsevier.com/locate/biopsych
M.A. Pico-Alfonso et al
Endorsement or lack thereof of any of the acts of physical,
sexual, or psychological violence was used as criteria to designate women as abused or nonabused. The presence or absence
of any of the acts of physical violence was used to distribute
abused women into two groups: physically abused and psychologically abused.
Lifetime History of Victimization
Childhood Abuse. Women were asked about the incidence
of physical, sexual, or psychological abuse during their childhood. Physical abuse was defined as noted earlier. Sexual abuse
included one or more of the following acts: forced sex, forced to
touch male sexual organs or being touched, forced exposure to
the display of sexual organs, and threats of forced sex. Psychological abuse was defined as noted earlier, but without the
factors regarding child custody and impeded decision making.
Adulthood Victimization. Women were asked about their
experience of violence during adulthood independently of their
being battered. Physical, sexual, and psychological violence was
defined as described for childhood abuse.
Mental Health
The severity of depression and anxiety and the incidence of
PTSD at the time of the study was assessed. Severity of depression
was measured with Beck’s Depression Inventory (BDI; Beck et al
1961), anxiety levels (both state and trait) were measured with
Spielberger’s State–Trait Anxiety Inventory (STAI; Spielberger et al
1970), and the severity of PTSD symptoms was determined using a
structured interview based on DSM-IV criteria (Echeburua et al
1997).
Saliva Samples
Saliva samples for the analysis of cortisol and DHEA levels
were also collected. The participants provided a minimum of .5
mL of saliva in a plastic tube twice a day (between 8 and 9 AM and
between 8 and 9 PM) for 4 consecutive days, starting the fourth
day after the beginning of menstruation. Saliva samples were
frozen in women’s freezers and brought to the Department of
Psychobiology in a mobile freezer, where they were kept frozen
at ⫺21°C. All samples were sent to the Department of Anatomy
of the University of Cambridge, England, for analysis.
Hormone Assays
Cortisol was measured by validated ELISA on 20 ␮L samples
of saliva (antibody, Cambio, Cambridge, United Kingdom) without extraction (intra-assay variation: 4.1%; interassay: 7.6%);
DHEA was measured by validated radioimmunoassay on 333 ␮L
samples after extraction into hexane/ether (4:1; antibody Bioclin
[Westmeath, Ireland]; intra-assay variation: 5.1%; interassay:
11.2%). Details are given in Harris et al (2000). Data from the four
daily morning and evening collections were summed to give a
mean value at each time point.
Statistical Analyses
The three groups of women (nonabused, physically abused,
and psychologically abused) were compared with respect to age
and depression and anxiety (trait and state) scores using one-way
analysis of variance (ANOVA) and with respect to the level of
education, the incidence of childhood abuse, adulthood victimization, and the incidence of PTSD by using Pearson’s chi-square
tests. Comparisons among groups with respect to the levels of
BIOL PSYCHIATRY 2004;56:233–240 235
M.A. Pico-Alfonso et al
Table 1. Characteristics of the Nonabused, Physically Abused, and Psychologically Abused Women
Variable
Age (Mean and SD)
Education Level (%)
Illiterate
Able to read and write
Incomplete primary school
Primary school
Secondary school
University studies: 3–4 years
University studies: 5–6 years
Cohabitation with Partner (%)
During the last year
At saliva collection
Smoking (%)
Number of cigarettes (mean and SD)
Pharmacological Treatment (%)
Antidepressants
Benzodiazepines
Estrogens and progestagens
Glucocorticoids
Lithium
Childhood Abuse (%)
No abuse
Physical
Psychological
Sexual
Adulthood Victimization (%)
No victimization
Physical
Psychological
Sexual
Intimate Partner Violence (%)
Physical
Psychological
Sexual
Nonabused
(n ⫽ 46)
Physically
Abused
(n ⫽ 70)
Psychologically
Abused
(n ⫽ 46)
p Value
47.57 ⫾ 12.79
44.77 ⫾ 10.76
45.39 ⫾ 10.09
ns
0
2.2
15.2
39.1
34.8
4.3
4.3
1.4
10.0
28.6
30.0
21.4
2.9
5.7
2.2
10.9
13.0
41.3
28.3
2.2
2.2
ns
97.8
97.8
41.3
13.37 ⫾ 7.64
81.4
61.4
41.4
19 ⫾ 13.47
87.0
71.7
37.0
13.53 ⫾ 9.47
—
—
13.0
10.9
19.6
4.3
0
22.9
41.4
10.0
5.7
2.9
26.1
26.1
10.9
4.3
0
—
—
—
—
—
50.0
28.3
26.1
13.0
35.7
47.1
42.9
38.6
37.0
54.3
39.1
32.6
ns
.05
ns
.01
60.9
10.9
23.9
15.2
41.4
30.0
45.7
34.3
43.5
17.4
30.4
37.0
ns
.05
.05
.05
0
0
0
100
100
33.0
0
100
15.2
—
—
—
cortisol and DHEA were carried out after logarithmic transformation of the data by one-way analysis of variance and Scheffé post
hoc comparisons and two-way ANOVA with the Time of Day and
Group as factors. Additionally, one-way ANOVA comparison was
carried out between women who were cohabiting with the
batterer and those who were not at the time of the saliva
collection. Correlation between hormonal levels and between
mood states and hormonal variables were assessed using Pearson’s linear correlation coefficient.
To determine the effect of IPV on cortisol and DHEA levels
after controlling for age, smoking, pharmacologic treatment
(antidepressants, benzodiazepines, estrogens, progestagens, glucocorticoids, and lithium), childhood abuse (physical, psychological, and sexual), and adulthood victimization (physical, psychological, and sexual), hierarchical multiple regression analyses
were carried out entering control variables (step 1) and incidence
of IPV (step 2). To test the hypothesis of the mediating effect of
the severity of depression and anxiety and the incidence of PTSD
on the effect of battering on cortisol and DHEA levels, hierarchical multiple regression analyses were performed entering control
variables (step 1), incidence of IPV (step 2), and score of
depression or anxiety or the incidence of PTSD (step 3). The
level of significance for all analyses was set at p ⬍ .05.
Results
Subjects
One hundred and sixty-two women participated in this study.
They were distributed into three groups: nonabused (n ⫽ 46),
physically abused (n ⫽ 70), and psychologically abused (n ⫽ 46) by
their intimate male partners. There were no differences between
groups in age [F(2,159) ⫽ .9, p ⫽ ns] or educational level [␹2(12, N
⫽ 162) ⫽ 12.0, p ⫽ ns]; see Table 1. Nearly all the nonabused
women (97.8%) had been cohabiting with their partners for the
previous year, and the same percentage were still cohabiting during
the collection of saliva samples. The percentages for the physically
abused and the psychologically abused women were 81.4%, 61.4%,
and 87.0%, 71.7%, respectively.
Violence Perpetrated by the Intimate Male Partner
All women (100%) exposed to physical violence (physically
abused group) had also suffered psychological violence by their
intimate male partners. Furthermore, 33% of physically abused
and 15.2% of psychologically abused women had also been
sexually abused.
Most women of the physically abused group experienced
violence by the batterer during the 12 months before the start
www.elsevier.com/locate/biopsych
236 BIOL PSYCHIATRY 2004;56:233–240
M.A. Pico-Alfonso et al
Table 2. Severity of Self Rated Depression-anxiety, and PTSD in Nonabused, Physically Abused, and Psychologically
Abused Women
Variable
BDI (Mean Score and SD)
STAI (Mean Score and SD)
State
Trait
PTSD
Incidence (%)
Total Score (Mean and SD)
Nonabused
(n ⫽ 46)
Physically Abused
(n ⫽ 70)
a
Psychologically Abused
(n ⫽ 46)
a
p Value
14.24 ⫾ 8.5
.001
52.7 ⫾ 33.8a
56.43 ⫾ 32.9a
48.03 ⫾ 36.8a
55.15 ⫾ 31.6a
.001
.001
21.4
13.3 ⫾ 11.6a
32.6
13.0 ⫾ 9.7a
.001
.001
5.65 ⫾ 5.5
16.1 ⫾ 12.2
18.28 ⫾ 18.1
24.49 ⫾ 27.6
0
1.9 ⫾ 2.6
BDI, Beck Depression Inventory; PTSD, posttraumatic stress disorder; STAI, Spielberger’s State–Trait Anxiety
Inventory.
a
Differs from nonabused group p ⬍ .001.
of their participation in the study: 90.8% experienced physical
abuse, 97% psychological abuse, and 18.5% sexual abuse.
Similarly, 84.8% of the psychologically abused women experienced continued psychological abuse during the last year,
and 2.2% were also sexually abused.
Lifetime History of Victimization
Childhood Abuse. There was a history of childhood abuse in
all three groups. Whereas 28.3%, 26.1%, and 13% of the nonabused women had a history of childhood physical, psychological, or sexual abuse, respectively, the percentages were 47.1%,
42.9%, and 38.6% for the physically abused group and 54.3%,
39.1%, and 32.6% for the psychologically abused group. These
differences were statistically significant between the groups for
childhood physical [␹2(2, N ⫽ 162) ⫽ 6.9, p ⫽ .05] and sexual
[␹2(2, N ⫽ 162) ⫽ 8.9, p ⫽ .01], but not for psychological abuse
[␹2(2, N ⫽ 162) ⫽ 3.5, p ⫽ ns]. The incidence of physical and
sexual childhood abuse was higher than that expected in the two
abused groups but not in the nonabused group.
Adulthood Victimization. Violence perpetrated toward
women by others than an intimate male partner during adulthood occurred in all three groups, with significant differences
between groups in the incidence of physical [␹2(2, N ⫽ 162) ⫽
6.6, p ⫽ .05], psychological [␹2(2, N ⫽ 162) ⫽ 6.4, p ⫽ .05], and
sexual violence [␹2(2, N ⫽ 162) ⫽ 6.5, p ⫽ .05]. In the physically
abused group, the incidence of the three types of violence
observed was higher than that expected, but this was not the
case for the nonabused group. In the psychologically abused
group, only the incidence of sexual violence was higher than that
expected.
Current Mental Health Status
There were highly significant differences between groups in
the severity of self-rated depression [BDI: F (2,159) ⫽ 17.1, p ⬍
.001], state anxiety [F (2,159) ⫽ 18.3; p ⬍ .001], and trait anxiety
[F (2,159) ⫽ 16.8; p ⬍ .001], and in the incidence of PTSD [␹2(2,
N ⫽ 162) ⫽ 16.9; p ⬍ .001]; see Table 2. Both physically and
psychologically abused groups had a higher score of severity of
depression and of state and trait anxiety and a higher incidence
of PTSD than the nonabused group; there were no differences
between the two abused groups. Almost all women (90%) with
PTSD had depressive symptomatology, ranging from mild
(33.3%), moderate (26.7%), to severe (30%) depression. Only
10% of physically abused and 2.2% of psychologically abused
women had a history of depression before the IPV experience.
www.elsevier.com/locate/biopsych
Differences in Salivary Cortisol Between Abused and
Nonabused Women
Repeated-measures ANOVA showed significant effect of Time
of Day [F (1,159) ⫽ 560.1, p ⬍ .001) and a Group ⫻ Time
interaction (F (1,159) ⫽ 6.3, p ⬍ .002) on salivary cortisol. There
were no differences between groups in the mean morning levels
of cortisol [F (2,159) ⫽ 1.2, p ⫽ ns]; see Figure 1. Mean evening
cortisol, however, was significantly different between groups
[F (2,159) ⫽ 6.9, p ⬍ .001]. Post hoc comparison showed that
while the physically abused group had significantly higher levels
of cortisol than the nonabused (p ⬍ .001), the psychologically
abused group fell short of significance (p ⬍ .065); however, there
were no differences between the two abused groups. A comparison of evening cortisol levels within the two abused groups
between those women who were or were not depressed showed
no significant difference (data not shown).
With regard to the increase in evening cortisol levels, hierarchical multiple regression analyses were carried out to control for
the following variables: age, smoking, pharmacologic treatment,
childhood abuse, and adulthood victimization. The analyses
showed that these controlled variables, in total, did not significantly predict the increase in evening cortisol level [F (13,148) ⫽
1.2; R2 ⫽ .095; p ⫽ ns]; however, the intake of benzodiazepines
by itself was identified as a significant predictor of this increase
(t ⫽ 2.6, p ⬍ .01). The overall effect of IPV (nonabused vs.
abused-groups) on the increase in the evening cortisol level
remained highly significant even after such variables had been
controlled [F (1,147) ⫽ 11.73; R2 ⫽ .067; p ⬍ .001].
Figure 1. Morning and evening cortisol saliva levels of nonabused, physically abused, and psychologically abused women. ⫹ Differs from nonabused group, p ⬍ .065. *** Differs from nonabused group, p ⬍ .001.
BIOL PSYCHIATRY 2004;56:233–240 237
M.A. Pico-Alfonso et al
Table 3. Correlations Between Severity of Self Rated Depression (BDI),
Anxiety (STAI), and PTSD and Morning and Evening Cortisol and DHEA
saliva levels in abused and non-abused women (n ⫽ 162).
STAI
Hormone
BDI Rho
State Rho
Trait Rho
PTSD Total
Score Rho
⫺.24b
.17a
⫺.21b
.07
⫺.30c
.22b
⫺.07
.13
.06
.14
.04
.07
.07
.14
Cortisol
AM
PM
DHEA
AM
PM
Figure 2. Morning and evening dehydroepiandrosterone (DHEA) level in
saliva of nonabused, physically abused, and psychologically abused
women. ⫹ Differs from nonabused group, p ⬍ .065. * Differs from nonabused group, p ⬍ .05. ** Differs from nonabused group, p ⬍ .01.
From the four daily morning and evening samples, the
maximum and minimum value for each woman at each time
point was determined. This showed that maximum levels of
evening cortisol were significantly greater in the physically
abused than in the nonabused group (p ⬍ .001); this difference
did not reach significance for the psychologically abused group
(p ⬍ .06); however, the two abused groups did not differ from
each other. On the contrary, there were no significant differences
in the minimum levels of evening cortisol (F ⫽ 1.2; p ⫽ ns).
There was, however, no correlation between morning and
evening cortisol levels (Pearson, r ⫽ ⫺.072, p ⫽ ns).
One-way ANOVA comparison of abused women who cohabited with the batterer at the time of saliva collection with those
who did not showed no significant differences in either morning
or evening cortisol levels [morning: F (1,115) ⫽ .37, p ⫽ ns;
evening: F (1,115) ⫽ .16, p ⫽ ns]. More restricted analysis of
physically abused women with respect to cohabitation also
showed no differences [F (1,69) ⫽ .008 –1.64, p ⫽ ns]. Similar
results were found in the psychologically abused group [F (1,45)
⫽ 1.64 –1.82, p ⫽ ns].
One-way ANOVA comparison of control women with childhood abuse (n ⫽ 23) with those without (n ⫽ 23) showed no
differences in cortisol [morning: F (1,45) ⫽ 2.12, p ⫽ ns; evening:
F (1,45) ⫽ .82, p ⫽ ns].
Differences in Salivary DHEA Between Abused and
Nonabused Women
Repeated-measures ANOVA showed a significant effect of
Time of Day on DHEA levels [F (1,159) ⫽ 181.8, p ⬍ .001) but no
Group ⫻ Time interaction [F (1,159) ⫽ .56, p ⫽ ns]. Both morning
and evening DHEA were different between groups [F (2,159) ⫽
5.2, p ⬍ .006; F (2,159) ⫽ 6.1, p ⬍ .003, respectively]; see Figure
2. Overall, there was a high positive correlation between morning and evening DHEA (Pearson, r ⫽ .82, p ⬍ .001). Scheffé
comparisons showed that morning DHEA was higher in psychologically abused women than in the nonabused group (p ⬍ .01),
although differences did not reach significant levels for the
physically abused group (p ⫽ .065); however, there were no
differences between the two abused groups. Evening DHEA was
significantly higher in both abused groups than in the nonabused
one (physically abused: p ⫽ .03, psychologically abused: p ⫽
.004). The two abused groups did not differ. Analysis of daily
variations in the four morning and evening DHEA samples
.10
.09
BDI, Beck Depression Inventory; DHEA, Dehydroepiandrosterone; STAI,
Spielberger’s State-Trait Anxiety Inventory; PTSD, posttraumatic stress
disorder.
a
p ⬍ .05.
b
p ⬍ .01.
c
p ⬍ .001.
(maximum and minimum values) showed no significant differences between groups.
Hierarchical regression analysis showed that the total controlled variables significantly predicted the increase in morning
DHEA [F (13,148) ⫽ 3.575; R2 ⫽ .239, p ⬍ .001] with age (t ⫽
⫺5.022, p ⬍ .001) and adulthood psychological victimization (t
⫽ ⫺2.048, p ⬍ .05) having an independent effect. The controlled
variables also had an effect on evening DHEA [F (13,148) ⫽
2.179, R2 ⫽ .161; p ⬍ .02] with age (t ⫽ ⫺3.014, p ⬍ .003) and
intake of benzodiazepines (t ⫽ 2.231, p ⬍ .027) having an
independent effect. None of the other controlled variables were
identified as individual predictors. The effect of IPV (nonabused
vs. abused women) on both morning and evening DHEA levels
remained highly significant even after these variables had been
controlled [morning: F (1,147) ⫽ 7.132, R2 ⫽ .035, p ⬍ .008;
evening: F(1,147) ⫽ 9.865, R2 ⫽ .053, p ⬍ .002].
One-way ANOVA comparison of control women with childhood abuse (n ⫽ 23) with those without (n ⫽ 23) showed no
differences in DHEA levels [morning: F (1,45) ⫽ .05, p ⫽ ns;
evening: F (1,45) ⫽ 2.13, p ⫽ ns]. As for cortisol, there was no
effect of cohabitation at the time of saliva collection on DHEA
levels [morning: F (1,115) ⫽ .26, p ⫽ ns; evening: F (1,115) ⫽
1.08, p ⫽ ns].
Correlations Between Cortisol and DHEA Levels
The analysis of correlation between cortisol and DHEA levels
showed a positive correlation between mean evening cortisol
and both morning and evening mean DHEA levels (morning
DHEA r ⫽ .28, p ⬍ .001; evening DHEA r ⫽ .48, p ⬍ .001).
Correlations Between Mental Health Status and Hormonal
Levels
There was a negative correlation between morning cortisol
and severity of depression (r ⫽ ⫺.24, p ⬍ .002) but a positive
correlation with evening cortisol (r ⫽ .17, p ⬍ .03); see Table 3.
State anxiety was negatively correlated with morning cortisol (r
⫽ ⫺.21, p ⬍ .007), and trait anxiety was negatively correlated
with morning cortisol (r ⫽ ⫺.30, p ⬍ .001) but positively
correlated with evening cortisol (r ⫽ .22, p ⬍ .005). There were
no significant correlations between depression, anxiety, and
DHEA levels. Finally, there was no correlation between total
PTSD score and hormonal levels.
www.elsevier.com/locate/biopsych
238 BIOL PSYCHIATRY 2004;56:233–240
Testing the Mediating Effect of Depression, Anxiety, and
PTSD on Cortisol and DHEA Levels
Hierarchical multiple regression analysis to test the mediating
effect of mood states on evening cortisol levels showed no effect
of either depression [F (1,146) ⫽ .12; R2 ⫽ .001; p ⫽ ns], state
anxiety [F (1,146) ⫽ .97; R2 ⫽ .006; p ⫽ ns], trait anxiety [F (1,146)
⫽ 1.85; R2 ⫽ .01; p ⫽ ns], or PTSD [F (1,146) ⫽ 1.38; R2 ⫽ .008;
p ⫽ ns]. Similarly, there was no mediating effect for depression,
anxiety, or PTSD on DHEA levels: depression [morning DHEA:
F (1,146) ⫽ .001; R2 ⫽ .001; p ⫽ ns; evening DHEA: F (1,146) ⫽
.37; R2 ⫽ .002; p ⫽ ns], state anxiety [morning DHEA: F (1,146) ⫽
.72; R2 ⫽ .004; p ⫽ ns; evening DHEA: F (1,146) ⫽ .67; R2 ⫽ .004;
p ⫽ ns], trait anxiety [morning DHEA: F (1,146) ⫽ .2; R2 ⫽ .001;
p ⫽ ns; evening DHEA: F (1,146) ⫽ .24; R2 ⫽ .001; p ⫽ ns], or
PTSD [morning DHEA: F (1,146) ⫽ .004; R2 ⫽ .001; p ⫽ ns;
evening DHEA: F (1,146) ⫽ .21; R2 ⫽ .001; p ⫽ ns].
Discussion
This study shows that women victims of IPV (battered
women) have higher levels of evening cortisol and higher levels
of both morning and evening DHEA than control women.
Furthermore, they have a markedly higher score of severity of
depression and state and trait anxiety and a higher incidence of
PTSD than control subjects. On the other hand, although battered
women and control subjects were well-matched for age and
education level, the incidence of childhood abuse and adulthood
victimization was higher in the former. Thus, this made it
necessary to consider whether the lifetime history of victimization or the mental state was responsible for the endocrine
changes observed in the battered women.
Increased evening cortisol is characteristic of current depression (Dalh et al 1991; Plotsky et al 1998). Therefore, although
depression was more frequent in both the physically and psychologically abused women, this factor did not seem to account
for the evening hypercortisolemia. As expected, however, there
was a positive correlation between the severity of current
depressive symptoms and evening cortisol, but there was no
significant difference in evening cortisol between those battered
women who were or were not depressed. This suggests that
either the experience of IPV (particularly physical violence) or
some genetic feature distinguishing this group of women (which
seems intrinsically unlikely) was responsible for this endocrine
feature of battered women. Elevated evening cortisol is present
in other conditions besides depression, including food deprivation and anorexia, which are also associated with this endocrine
pattern (Murphy 1991). Our previous studies have shown no
consistent associations between state or trait anxiety and salivary
cortisol (Goodyer et al 2000, 2001; Harris et al 2000), and these
mood states did not contribute to the endocrine differences
found in the study between battered and control women. We did
find a positive correlation between severity of trait anxiety and
evening cortisol, however, even though this could not account
for the differences between abused and control groups. Posttraumatic stress disorder has been associated with lowered cortisol in
men (Yehuda et al 1996), and there are reports that PTSD or
similar stressor states increase cortisol in women (for example,
Rasmusson et al 2001). There was no evidence from our results
that its presence played a part in determining evening cortisol
levels. Almost all women (90%) with PTSD in our group had
depressive symptomatology, which made it difficult to assess the
relation between PTSD alone and cortisol levels. It did not
appear, however, that current mental state was associated with
www.elsevier.com/locate/biopsych
M.A. Pico-Alfonso et al
the differences we found in evening cortisol between women
victims of IPV and control subjects.
On the other hand, although battered women had higher
incidence of lifetime history of victimization than control subjects, both during childhood and as adults, these experiences did
not seem to account for the increase in evening cortisol. It should
be noted that some degree of what is now defined as physical or
psychological abuse was a common childrearing practice in
Spain at the time when our subjects were children. This explains
the high levels (50%) in the control group; however, the nature
of the IPV that women had experienced had a moderate effect.
Elevated evening cortisol was most marked in physically abused
women, whereas the psychologically abused group was not
significantly different from control subjects; the two abused
groups did not differ from each other, however. A finer-grained
analysis of the cortisol data in our study showed that the evening
hypercortisolemia was associated with increased maximum
(“peak”) values; that is, battered women (particularly those
suffering physical violence) experienced intermittently higher
levels of cortisol in the evening, which accounted for the overall
higher mean value. By contrast, minimum values were unchanged, showing that the daily variation in evening cortisol was
increased in battered women.
It is noteworthy that morning cortisol was unchanged in
battered women. These results differ from those of Seedat et al
(2003), which were based on a much smaller sample (22 abused
women and 16 control subjects; 10 of the abused women had
PTSD and 12 had never had it). Furthermore, abused women
were free from the abusive relationships for more than 4 months
and less than 2 years; cortisol was measured only in the morning
(9 –12 AM) from blood samples that were not collected in the
women’s normal environment.
Elevated levels of salivary cortisol in the morning is a risk
factor for subsequent MDD in both adult women and adolescents
(Goodyer et al 2001; Harris et al 2000). This factor was not found
in our sample of battered women, suggesting that morning
cortisol did not contribute to the increased incidence of depression, although whether individual differences in morning cortisol
earlier in life— before they had been exposed to the battering
experience—might play a part in the affective response to
subsequent IPV remains unknown and would require a different
type of study. There is some evidence that morning cortisol
increases during the 30-60 min after wakening (Federenko et al
2004; Pruessner et al 1999). It is unlikely that sleep patterns and
“endogenous” morning cortisol cancelled each other out in the
three groups, although we had no data on wake times. Similar to
our findings with evening cortisol, trait (as well as state) anxiety
seem to be related to morning cortisol. In the light of results from
other studies (described earlier in the article), the relation
between current mood state and cortisol requires further exploration.
We also observed elevated DHEA in both morning and
evening salivary samples in physically and psychologically
abused women. This is not characteristic of depression, which is
associated with lower levels of DHEA in the morning (Michael et
al 2000). This finding reinforces the supposition that the endocrine patterns we observed in battered women are not simply the
result of associated and consequent affective (e.g., mood) state.
Unlike cortisol, DHEA declines markedly with age (Orentreich et
al 1992), but battered and control women were well-matched for
age, and controlling for age did not alter the association between
higher DHEA and both physical and psychological IPV. Unlike
cortisol, however, there was a significant association between
M.A. Pico-Alfonso et al
morning DHEA levels and adulthood psychological victimization. This analysis does suggest a rather different relationship
between psychological variables and DHEA than exists for
cortisol. Benzodiazepine intake was also associated with higher
evening DHEA, although whether this is a causal relationship,
and, if so, in what direction, remains uncertain. Elevated DHEA
has also been found to be associated with increased risk of
depression in adolescents but not in adult women (Goodyer et al
2001), so whether this endocrine factor plays a part in the
increased incidence of depression in our battered women is also
uncertain. We did find, however, that DHEA was not associated
with the presence of depression, anxiety, or PTSD, and thus the
pattern of raised DHEA is unlikely to reflect current affective
state.
The increased incidence of depression, anxiety, and PTSD we
observed in the battered women who participated in our study
confirms previous findings and suggests that our sample was
similar to those of other studies. Dysphoria and depression is
common in battered women across ethnic groups (Campbell et al
1996; Clements and Sawhney 2000; Torres and Han 2000).
Psychological violence can be as damaging as physical violence
(Street and Arias 2001), although our findings suggest that the
endocrine consequences of the latter may be somewhat greater.
Similar to other reports, we found a high incidence of sexual
violence associated with both physical and psychological IPV
(Marshal and Holtzworth-Munroe 2002). Posttraumatic stress
disorder is common in battered women, particularly those suffering sexual violence (Bennice et al 2003). Because a high
proportion of our battered women had experienced either recent
or current abuse, we cannot know whether the endocrine
features we describe represent a stress reaction to this abuse or
a more persistent state of stress that might outlast active physical
or psychological violence. What our findings show is that both
physical and psychological IPV apparently results in changes in
cortisol and DHEA in women victims, both hormones having
been implicated in mental health and neural functioning. For
example, raised cortisol can sensitize the brain to other noxious
or adverse events. Brain injury is relatively common as part of the
IPV (Valera and Berenbaum 2003). Whether the endocrine
features of battered women we describe add to this injury or
have longer-term consequences for either the physical integrity
of the brain or affective or cognitive function is now a question
of considerable interest. Attempts to establish the impact that
IPV—a form of chronic, often inescapable, social stress— has on
women victims need to take into account the whole spectrum of
the organism functioning from physiologic alterations to mental
and physical health.
This research was supported by the Institute of the Woman,
Ministry of Work and Social Affairs (Grant Nos. 53/98), FEDER,
and the Ministry of Science and Technology (Grant No.
BSO2001-3134). The authors thank Helen Shiers and Sarah
Cleary for the assays, Miriam Phillips for English editorial
assistance, and Dr. Vicente Gonzalez-Roma for statistical help.
We also thank the Conselleria of Social Welfare and the 24-hour
centers for helping women of the Valencian community, Spain,
for their assistance in contacting the battered women.
Aguilar RJ, Nightingale NN (1994): The impact of specific battering experiences on the self-esteem of abused women. J Fam Violence 9:35– 45.
Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961): An inventory for
measuring depression. Arch Gen Psychiatry 4:561–571.
BIOL PSYCHIATRY 2004;56:233–240 239
Bennice JA, Resick PA, Mechanic M, Astin M (2003): The relative effects of
intimate partner physical and sexual violence on post-traumatic stress
disorder symptomatology. Violence Vict 18:87–94.
Campbell J, Jones AS, Dienemann J, Kub J, Schollenberger J, O’Campo P, et al
(2002): Intimate partner violence and physical health consequences.
Arch Intern Med 162:1157–1163.
Campbell J, Kub JE, Rose L (1996): Depression in battered women. J Am Med
Womens Assoc 51:106 –110.
Campbell JC (2002): Health consequences of intimate partner violence.
Lancet 359:1331–1336.
Cicchetti D, Rogosch FA (2001): Diverse patterns of neuroendocrine activity
in maltreated children. Dev Psychopathol 13:677– 693.
Clements CM, Sawhney DK (2000): Coping with domestic violence: Control
attributions, dysphoria, and hopelessness. J Trauma Stress 13:219 –240.
Coker AL, Smith PH, Bethea L, King MR, McKeown RE (2000): Physical health
consequences of physical and psychological intimate partner violence.
Arch Fam Med 9:451– 457.
Constantino RE, Sekula LK, Rabin B, Stone C (2000): Negative life experiences, depression, and immune function in abused and nonabused
women. Biol Res Nurs 1:190 –198.
Dalh RE, Ryan ND, Puig-Antich J, Nguyen NA, al-Shabbout M, Meyer VA, et al
(1991): 24-hour cortisol measures in adolescents with major depression:
A controlled study. Biol Psychiatry 30:25–36.
De Bellis MD, Chrousos GP, Dorn LD, Burke L, Helmers K, Kling M, et al (1994):
Hypothalamic-pitutary-adrenal axis dysregulation in sexually abused
girls. J Clin Endocrinol Metab 78:249 –255.
Echeburua E, Corral P, Amor PJ, Zubizarreta I, Sarasua B (1997): Escala de
gravedad de Sı́ntomas de Trastorno de Estrés Postraumático:
Propiedades psicométricas. Análisis y Modificación de Conducta 23:503–
526.
Federenko I, Wust S, Hellhammer DH, Dechoux R, Kumsta R, Kirschbaum C
(2004): Free cortisol awakening responses are influenced by awakening
time. Psychoneuroendocrinology 29:174 –184.
Follingstad DR, Rutledge LL, Berg BJ, Hause ES, Polek DS (1990): The role of
emotional abuse in physically abusive relationships. J Fam Violence
5:107–120.
Gold PW, Goodwin FK, Chrousos GP (1988): Clinical and biochemical manifestations of depression. Relation to neurobiology of stress. N Engl Med
319:413– 420.
Golding JM (1999): Intimate partner violence as a risk factor for mental
disorders: A meta-analysis. J Fam Violence 14:99 –132.
Goodyer IM, Herbert J, Altham PM, Secher S, Shiers HM (1996): Adrenal
secretion during major depression in 8 to 16 year olds. I. Altered diurnal
rhythms in salivary cortisol and dehydroepiandrosterone (DHEA): At
presentation. Psychol Med 26:245–256.
Goodyer IM, Herbert J, Tamplin A, Altham PM (2000): First-episode major
depression in adolescents. Affective, cognitive and endocrine characteristics of risk status and predictors of onset. Br J Psychiat 176:142–149.
Goodyer IM, Park RJ, Netherton CM, Herbert J (2001): Possible role of cortisol
and dehydroepiandrosterone in human development and psychopathology. Br J Psychiatr 179:243–249.
Hamberg K, Johansson EE, Lindgren G (1999): “I was always on guard”—an
exploration of woman abuse in a group of women with musculoskeletal
pain. Fam Pract 16:238 –244.
Harris TO, Borsanyi S, Messari S, Stanford K, Cleary SE, Shiers HM, et al (2000):
Morning cortisol as a risk factor for subsequent major depressive disorder in adult women. Br J Psychiat 177:505–510.
Heuser I, Deuschle M, Luppa P, Schweigher U, Standhardt H, Weber B (1998):
Increased diurnal plasma concentrations of dehydroepiandrosterone in
depressed patients. J Clin Endocrinol Metab 83:3130 –3133.
King JA, Mandansky D, King S, Fletcher KE, Brewer J (2001): Early sexual
abuse and low cortisol. Psychiatry Clin Neurosci 55:71–74.
Lown EA, Vega WA (2001): Intimate partner violence and health: Self-assessed health, chronic health, and somatic symptoms among Mexican
American women. Psychosom Med 63:352–360.
Marshal AD, Holtzworth-Munroe A (2002): Varying forms of husband sexual
aggression: Predictors and subgroup differences. J Fam Psychol 16:286 –
296.
Martinez M, Garcı́a-Linares MI, Pico-Alfonso MA (2003): Women victims of
domestic violence: Consequences for their health and the role of the
health system. In: Renate CA, Klein RCA, Wallner B, editors. Gender, Conflict, and Violence. Vienna: Studien-Verlag, 53–71.
www.elsevier.com/locate/biopsych
240 BIOL PSYCHIATRY 2004;56:233–240
Michael A, Jenaway A, Paykel ES, Herbert J (2000): Altered salivary dehydroepiandrosterone levels in major depression in adults. Biol Psychiatry
48:989 –995.
Murphy BE (1991): Steroids and depression. J Steroid Biochem Mol Biol 38:
537–559.
O’Leary KD (1999): Psychological abuse: A variable deserving critical attention in domestic violence. Violence Vict 14:3–23.
Oquendo MA, Echevarria G, Galfalvy HC, Grunebaum MF, Burke A, Barrera A,
et al (2003): Lower cortisol levels in depressed patients with comorbid
post-traumatic stress disorder. Neuropsychopharmacology 28:591–598.
Orentreich N, Brind JL, Vogelman JH, Andres R, Baldwin H (1992): Long-term
longitudinal measurements of plasma dehydroepiandrosterone sulfate
in normal men. J Clin Endocrinol Metab 75:1002–1004.
Parker KJ, Schatzberg AF, Lyons DM (2003): Neuroendocrine aspects of
hypercortisolism in major depression. Horm Behav 43:60 – 66.
Plotsky PM, Owens MJ, Nemeroff CB (1998): Psychoneuroendocrinology of
depression. Hypothalamic-pituitary-adrenal axis. Psychiatr Clin North Am
21:293–307.
Pruessner JC, Hellhammer DH, Kirschbaum C (1999): Burnout, perceived
stress, and cortisol responses to awakening. Psychosom Med 61:197–
204.
Rasmusson AM, Lipshitz DS, Wang S, Hu S, Vojvoda D, Bremner JD, et al
(2001): Increased pituitary and adrenal reactivity in premenopausal
women with posttraumatic stress disorder. Biol Psychiatry 50:965–
977.
Resnick HS, Yehuda R, Pitman RK, Foy DW (1995): Effect of previous trauma on
acute plasma cortisol level following rape. Am J Psychiatry 152:1675–1677.
Rubin RT, Phillips JJ, McCracken JT, Sadow TF (1996): Adrenal gland volume
www.elsevier.com/locate/biopsych
M.A. Pico-Alfonso et al
in major depression: Relationship to basal and stimulated pituitaryadrenal cortical axis function. Biol Psychiatry 40:89 –97.
Sachar EJ, Puig-Antich J, Ryan ND, Asnis GM, Rabinovich H, Davies M, et al
(1985): Three tests of cortisol secretion in adult endogenous depressives.
Acta Psychiatr Scand 71:1– 8.
Seedat S, Stein MB, Kennedy CM, Hauger RL (2003): Plasma cortisol and
neuropeptide Y in female victims of intimate partner violence. Psychoneuroendocrinology 28:796 – 808.
Spielberger CD, Gorsuch RL, Lushene RE (1970): The State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists’ Press.
Street AE, Arias I (2001): Psychological abuse and posttraumatic stress disorder in battered women: Examining the role of shame and guilt. Violence Vict 16:65–78.
Torres S, Han HR (2000): Psychological distress in non-Hispanic white and
Hispanic abused women. Arch Psychiatr Nurs 14:19 –29.
Valera EM, Berenbaum (2003): Brain injury in battered women. J Consult Clin
Psychol 71:797– 804.
Woods SJ (2000): Prevalence and patterns of posttraumatic stress disorder in abused and postabused women. Issues Ment Health Nurs
3:309 –324.
Yehuda R (2002): Current status of cortisol findings in post-traumatic stress
disorder [review]. Psychiatr Clin North Am 25:341–368, vii.
Yehuda R, Resnick HS, Schmeidler J, Yang RK, Pitman RK (1998): Predictors of
cortisol and 3-Methoxy-4-Hydroxy-phenylglycol responses in the acute
aftermath of rape. Biol Psychiatry 43:855– 859.
Yehuda R, Teicher MH, Trestman RL, Levengood RA, Siever LJ (1996): Cortisol
regulation of posttraumatic stress disorder and major depression: A
chronobiological analysis. Biol Psychiatry 40:79 – 88.