Stress

The International Journal on the Biology of Stress

ISSN: 1025-3890 (Print) 1607-8888 (Online) Journal homepage: http://www.tandfonline.com/loi/ists20

Anxious attachment style predicts an enhanced

cortisol response to group psychosocial stress

Nina Smyth, Lisa Thorn, Andrea Oskis, Frank Hucklebridge, Phil Evans &
Angela Clow

To cite this article: Nina Smyth, Lisa Thorn, Andrea Oskis, Frank Hucklebridge, Phil Evans &
Angela Clow (2015) Anxious attachment style predicts an enhanced cortisol response to group
psychosocial stress, Stress, 18:2, 143-148, DOI: 10.3109/10253890.2015.1021676

To link to this article: http://dx.doi.org/10.3109/10253890.2015.1021676

Published online: 11 Mar 2015.

Submit your article to this journal

Article views: 76

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=ists20

Download by: [University of California Santa Barbara] Date: 06 November 2015, At: 08:11
 http://informahealthcare.com/sts
ISSN: 1025-3890 (print), 1607-8888 (electronic)

Stress, 2015; 18(2): 143–148

! 2015 Informa UK Ltd. DOI: 10.3109/10253890.2015.1021676

ORIGINAL RESEARCH REPORT

Anxious attachment style predicts an enhanced cortisol response to

group psychosocial stress
Nina Smyth1, Lisa Thorn1, Andrea Oskis2, Frank Hucklebridge1, Phil Evans1, and Angela Clow1,2
1
Psychophysiology and Stress Research Group, Department of Psychology, University of Westminster, London, UK and 2Department of Psychology,
Middlesex University, London, UK
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

Abstract Keywords
Insecure attachment style is associated with poor health outcomes. A proposed pathway Group stressor, HPA axis, healthy females,
implicates the hypothalamic-pituitary-adrenal axis (HPA-axis), dysregulation of which is stress reactivity, saliva, trier social stress
associated with a wide range of mental and physical ill-health. However, data on stress test
reactivity in relation to attachment style is contradictory. This relationship was examined using
the novel Trier Social Stress Test for groups (TSST-G): a group-based acute psychosocial stressor. History
Each participant, in the presence of other group members, individually performed public
speaking and mental arithmetic tasks. Seventy-eight healthy young females (20.2 ± 3.2 years), Received 14 November 2014
in groups of up to six participants completed demographic information and the Vulnerable Revised 26 January 2015
Attachment Style Questionnaire (VASQ), and were then exposed to the TSST-G. Physiological Accepted 16 February 2015
stress reactivity was assessed using salivary cortisol concentrations, measured on seven Published online 11 March 2015
occasions at 10-min intervals. Vulnerable attachment predicted greater cortisol reactivity
independent of age, smoking status, menstrual phase and body mass index. Supplementary
analysis indicated that insecure anxious attachment style (high scores on the insecurity and
proximity-seeking sub-scales of the VASQ) showed greater cortisol reactivity than participants
with secure attachment style. Avoidant attachment style (high scores for insecurity and low
scores for proximity seeking) was not significantly different from the secure attachment style.
Attachment style was not associated with the timing of the cortisol peak or post-stress recovery
in cortisol concentrations. These findings in healthy young females indicate subtle underlying
changes in HPA axis function in relation to attachment style and may be important for future
mental health and well-being.

Introduction et al., 2013; Jinyao et al., 2012; Puig et al., 2013). The
biologic underpinnings of these links however are not clear.
Attachment style is suggested to be important for regulating
One of the proposed pathways implicates the hypothal-
threat appraisal, stress response and recovery from stress,
amic-pituitary-adrenal (HPA) axis (Repetti et al., 2002),
although the mechanisms underlying this complex interplay
dysregulation of which is associated with a wide range of
are not well-understood (Diamond, 2001). Within the adult
mental and physical ill-health (McEwen, 2000). A flattened
attachment literature, insecure attachment style is generally
diurnal rhythm of cortisol secretion has been reported in
conceptualized along two dimensions, namely attachment
anxious attachment style (Oskis et al., 2011; Quirin et al.,
anxiety and avoidance (Brennan et al., 1998). A securely
2008); however, there are mixed findings from studies
attached individual is considered to be an individual with low
examining reactivity of the HPA axis in relation to attachment
levels of both (Brennan et al., 1998). High attachment
style. For example, in one study, avoidant (but not anxious)
anxiety is associated with preoccupation with the availability
attachment predicted enhanced stress-induced cortisol
and responsiveness of the other, maximization of negative
responding in females (Powers et al., 2006) while the opposite
experiences and hyper-vigilance to potential threat.
was found in another study (Quirin et al., 2008). Other studies
Attachment avoidance is associated with a tendency to
have reported insecure dismissing attachment style to predict
devalue intimacy and dependency and maximize autonomous
enhanced cortisol reactivity (Pierrehumbert et al., 2012;
behavior strategies when faced with potential threat. Insecure
Rifkin-Graboi, 2008) whereas, secure and dismissive attach-
attachment style is known to predict a range of poor physical
ment styles have been reported as similar elsewhere
and mental health outcomes (Bifulco et al., 2002a,b; Carr
(Kidd et al., 2011). Further studies show no relationship
between cortisol responding to a stressor and attachment style
Correspondence: Dr Angela Clow, Psychophysiology and Stress
(Ditzen et al., 2008; Smeets, 2010).
Research Group, Department of Psychology, University of
Westminster, 115 New Cavendish Street, London W1W 6UW, UK. Whether attachment style predicts acute stress responding
Tel: +44-0207-911-5000. E-mail: clowa@wmin.ac.uk remains unclear. The disparity in the literature may in part be
 144 N. Smyth et al. Stress, 2015; 18(2): 143–148

related to the wide array of methodologies that have been used purposes, and another participant did not complete the
to investigate this issue. Stressors have ranged from the Trier attachment questionnaire. A single participant was removed
Social Stress test (TSST) for individuals (Ditzen et al., 2008; from the data set on the basis that their cortisol data were >5
Smeets, 2010) to experimental conflict negotiation (Powers standard deviations (SD) above the mean for each sample, and
et al., 2006), visualization of hypothetical distressing situ- their data remained as outliers following square root trans-
ations (Rifkin-Graboi, 2008) and behavioral interference tasks formation. Analyses were performed on 78 participants, age
(Kidd et al., 2011). The most commonly used tool to assess ranging from 18 to 33 (mean ± SD: 20.1 ± 3.1) years.
attachment style in adult stress reactivity studies is the Participants were ethnically diverse; of those who disclosed
Experiences in Close Relationships Scale (Brennan & Shaver, their ethnicity, 26 were Asian (Indian, Chinese, Pakistani,
1995), which assesses attachment in romantic relationships. Bangladeshi and Arabic), 31 were white European, 13 were
The Vulnerable Attachment Style Questionnaire (VASQ; African Caribbean and 4 were mixed race.
Bifulco et al., 2003) is arguably a more appropriate measure To control for gender differences in cortisol reactivity, only
for use in research investigating attachment and HPA axis females were recruited. To reduce the impact of variables
activity as, rather than romantic attachment, it focuses on how known to influence cortisol reactivity, exclusion criteria
individuals generally relate to others. Furthermore, it per- included medication, illness and history of psychiatric illness.
forms somewhat better in predicting depression than other As cortisol reactivity is influenced by the menstrual cycle and
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

self-report measures of attachment (Bifulco et al., 2003), as body mass index (BMI) the number of days since last period
well as predicting negative psychosocial well-being and was recorded, as was height and weight (Dockray et al., 2009;
mental health in university students (Carr et al., 2013). The Smyth et al., 2013). Two participants used oral contracep-
VASQ was developed and validated in relation to an in-depth tives. The majority of participants (86%) were non-smokers.
interview procedure (Attachment Style Interview; Bifulco
et al., 2002a,b), which has been used in previous research Vulnerable Attachment Style Questionnaire
examining HPA axis activity and attachment style (Oskis The VASQ (Bifulco et al., 2003) is a brief self-report tool,
et al., 2011, 2014). which assesses general adult attachment. It is designed to
The TSST (Kirschbaum et al., 1993) comprises uncon- assess overall attachment vulnerability as well as two
trollability and socio-evaluative threat known to reliably dimensions of attachment: a global dimension of attachment
activate HPA axis function (Dickerson & Kemeny, 2004) and insecurity common to all insecure subtypes (representing a
has recently been adapted for use in group settings: the TSST- deep-rooted mistrust of others and their motives) and a
G (von Dawans et al., 2011). A primary motivation was to proximity-seeking dimension reflecting the strategy individ-
increase the rate of participant exposure to the TSST but it uals use to manage their insecurity (i.e. some individuals with
provides the opportunity to examine the impact of social high insecurity develop excessive neediness and vigilance of
dynamics on stress reactivity (Häusser et al., 2012). In the others, while other individuals develop an aversion to
present, study we adapted the TSST-G to maximize closeness with others). The scale comprises 22 items
opportunities for group interaction which may attenuate or measured on a 5-point scale, ranging from strongly disagree
increase stress reactivity depending on the characteristics of to strongly agree. Low scores on the 12-item insecurity
the individual within the group. Given that, those with high subscale (e.g. ‘‘I find it hard to trust others’’ and ‘‘People let
attachment insecurity easily perceive threats in their environ- me down a lot’’) represent secure attachment and high scores
ment, frequently experience social interactions as stressful reflect insecure attachment. The proximity-seeking subscale
and excessively ruminate about psychologically distressing consists of 10 items (e.g. ‘‘I get anxious when people close to
experiences (Burnette et al., 2009; Shaver & Mikulincer, me are away’’ and ‘‘I look forward to spending time on my
2002) they might find the group version of the TSST own’’). Low scores represent propensity for avoidant behavior
particularly stressful. We chose to investigate an all-female and high scores reflect a need for closeness with others.
sample since gender is known to moderate the link between Cronbach’s a was 0.81 for the insecurity scale and 0.74 for the
attachment style and HPA reactivity (Kiecolt-Glaser et al., proximity scale. A total attachment vulnerability measure can
1996; Kirschbaum et al., 1995; Stroud et al., 2002). Young, be derived by summing items on both scales. The VASQ can
healthy participants were recruited to explore whether also be used to categorize participants according to secure,
attachment style might be a pre-clinical indicator of vulner- insecure anxious or insecure avoidant attachment styles. The
ability rather than a consequence of concurrent poor health. insecure anxious attachment style category is derived
Due to discrepancies in the stress reactivity and attachment from high scores on both insecure and proximity-seeking
literature, the aim of this study was to examine self-reported measures. The insecure avoidant type category is derived
attachment and physiological stress responding to a group from high scores on the insecurity scale and low scores on the
psychosocial stressor. proximity scale.

Methods
Procedure
Participants
The study was approved by the University of Westminster
Eighty-one female undergraduate student participants were Ethics Committee. Following recruitment, groups of partici-
recruited. They did not receive financial incentives but did pants were invited to attend a test session at a set time and
receive course credits. Cortisol data was missing for one place. In line with best practice, guidelines (Kirschbaum
participant as the salivary volume was insufficient for assay et al., 1993; Smyth et al., 2013) testing commenced in the
 DOI: 10.3109/10253890.2015.1021676 Stress reactivity and attachment style 145

afternoon between 13:00 and 15:00 h, to control for changes and S7, at 62 min). This cortisol profile allowed us to capture
in basal cortisol secretion in the morning and following the the rise in cortisol, the cortisol peak and the decline of cortisol
post-prandial period. Participants were asked to refrain from (i.e. the recovery period) (Dickerson & Kemeny, 2004; Smyth
food, caffeine, alcohol, exercise and smoking 30 min prior to et al., 2013). Saliva samples were frozen at 20  C until
the research session. The TSST-G (von Dawans et al., 2011) assayed at the University of Westminster. Samples were
included three main phases: the group preparatory period thawed and centrifuged for 10 min at 3500 rpm. Cortisol
(30 min); the group stress task period (22 min); and a group concentrations were determined by enzyme-linked immuno-
resting and debriefing period (40 min). During the preparatory sorbent assay developed by Salimetrics LLC (State College,
period, groups of up to six participants met in Room 1 where PA). The standard range in the assay was 0.33–82.77 nmol/l.
they were informally seated around a single table and Intra- and inter-assay variations were both <10%.
introduced to the experimenter, they were free to talk to
each other at this time. Statistical analysis
Following informed written consent, participants com- Cortisol data were moderately skewed and, therefore, a square
pleted in silence demographic questions, the date of their last root transformation was applied which normalized distribu-
menstruation and the VASQ, if they had not already tions, although cortisol concentrations shown in figures and
completed it on-line (it had been available since the beginning tables are representative of original units. Descriptive statis-
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

of the recruitment period). Each participant then received a tics were explored for each cortisol sample measured
large sticker with a number between 1 and 6. They were throughout the TSST-G procedure, and a one-way within-
informed that they would be identified with this number subjects analysis of variance (ANOVA) was conducted to
during the task period and that the numbers would be called in examine differences in cortisol over time. Within-subjects
a random order. Participants were then introduced to the contrasts were used to assess the pattern of cortisol secretion.
saliva-sampling method. Following this, participants were As participants performed the TSST-G tasks at slightly
given 10-min quiet time to prepare notes for a mock job different time-points, cortisol reactivity was computed for
interview. They were asked to prepare a free 2-min speech as each individual as their peak sample minus baseline. Cortisol
if applying for a job of their choice and to introduce recovery was computed as individual peak sample minus
themselves to the committee. They were asked to convince the sample 7 (recovery). Pearson’s correlation coefficients were
committee that they were the most suitable candidates for the used to examine relationships between these cortisol indices,
position. After this preparatory period, the baseline saliva VASQ attachment measures and demographics variables.
sample was collected immediately prior to leaving Room 1. Significant relationships between cortisol and attachment
Participants were taken into Room 2 (a short distance away) measures were examined in a multiple regression analysis
and instructed to stand in a straight line in front of the already controlling for variables known to affect cortisol stress
seated committee, comprising one woman and one man. The reactivity. Participants were categorized according to the
committee were wearing white laboratory coats and there VASQ attachment style: secure (n ¼ 20), insecure anxious
were two conspicuous video cameras pointing at the partici- (n ¼ 37) and insecure avoidant (n ¼ 21). A one-way between-
pants. A committee member called the number of each subjects ANOVA explored group differences in cortisol stress
participant in turn in a random order to make a 2-min speech reactivity and Bonferroni post hoc tests were applied. 2
as if applying for a job. After all participants gave their speech square test was used to examine the association between
(a total of up to 12 min), the committee asked the participants, participants’ peak cortisol time and attachment style group.
in the same order, to serially subtract the number 17 from a
given number (e.g. 4878) as fast and accurately as possible for
Results
80 s. Each participant received an individual starting number
to avoid learning effects. Standard responses from the Results indicated that the TSST-G induced an overall cortisol
committee were followed where participants ended their response in this sample [F(6,462) ¼ 7.623, p < 0.001), illu-
speech before the 2-min duration (e.g. ‘‘you still have time, strated in Figure 1. Within-subjects contrasts revealed a
please continue’’) or failed in the subtraction task (e.g. ‘‘you significant quadratic effect [F(1,77) ¼ 23.807, p < 0.001], such
made a mistake please start again from the number. . .’’) (von that on average cortisol increased from baseline peaked at the
Dawans et al., 2011). Immediately after all participants had fourth sampling point (10 min after the completion of the
completed the TSST-G, participants were returned to Room 1, TSST-G) and subsequently declined.
where they collected saliva samples every 10 min up to 40 min Relationships between cortisol data and attachment vari-
following the TSST-G period. During this time they were ables measured by the VASQ were examined using focused
debriefed. composite cortisol indices: individual peak sample minus
baseline (cortisol reactivity) and individual peak sample
minus sample 7 (recovery). Descriptive statistic and intercor-
Saliva-sampling collection
relations for all variables are presented in Table 1.
Cortisol was measured in saliva samples collected using There was a significant positive relationship between
Salivettes (saliva-sampling devices; Sarstedt Ltd., Leicester, cortisol reactivity and VASQ vulnerability score (r ¼ 0.289,
UK) at baseline (immediately before the TSST-G: S1, at 0 min) p ¼ 0.010) in that participants with a higher level of
immediately after the public speaking task (S2, at 12 min), vulnerable attachment exhibited a greater increase in corti-
after the mental arithmetic task (S3, at 22 min), and every sol from baseline to peak value. In other terms, participants
10 min up to 62 min (S4, at 32 min S5, at 42 min, S6, at 52 min who demonstrated an insecure anxious attachment style
 146 N. Smyth et al. Stress, 2015; 18(2): 143–148

(those scoring highly on both VASQ subscales) displayed known to affect cortisol reactivity were included in the model
greater cortisol reactivity. With regards to the dimensions of (Table 2). Vulnerable attachment and age remained signifi-
the VASQ, insecurity was significantly positively correlated cant independent predictors of cortisol reactivity.
with cortisol reactivity, whereas proximity was not. There The VASQ can be used to group participants according to
were no relationships between attachment measures and secure, anxious or avoidant attachment styles. The insecure
cortisol recovery. There were also no relationships between anxious attachment style category is derived from high scores
cortisol measures and pertinent demographic characteristics, on both insecure and proximity-seeking measures. In a
apart from age, which was positively related with both supplementary analysis, a one-way between subjects
cortisol reactivity and recovery. In terms of VASQ attachment ANOVA was performed to examine the difference in cortisol
measures, vulnerability was unrelated to demographic vari- reactivity between the three groups. There was a significant
ables; however, insecurity was positively related with age, and effect of attachment style group on cortisol reactivity
proximity was positively correlated with smoking status. [F(2,5) ¼ 5.300, p ¼ 0.007] (Figure 2). Bonferroni post hoc
A multiple regression analysis was conducted to examine tests indicated that the insecure anxious group was signifi-
whether the relationship between cortisol reactivity and cantly different from the secure group (p ¼ 0.011). There was
vulnerable attachment remained significant when variables no association between when participants peaked and their
attachment style group (2 ¼ 16.405, p ¼ 0.173).
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

Discussion
Vulnerable attachment, determined by the VASQ, predicted
greater cortisol reactivity to a group psychosocial stressor
independent of age, smoking status, menstrual phase and
BMI. While there were no differences in the timing of the
cortisol peak, supplementary analysis revealed that partici-
pants with an insecure anxious attachment style (a combin-
ation of high scores on the insecurity and proximity-seeking
sub-scales of the VASQ) showed greater stress-induced

Table 2. Prediction of cortisol reactivity.

Predictors b t p Value
VASQ vulnerability 0.241 2.205 0.031
Age 0.322 2.902 0.005
Menstrual cycle phase 0.118 1.132 0.261
Figure 1. Mean (±SEM) salivary free cortisol concentrations (nano- Smoking status 0.182 1.706 0.092
moles per litre) for all participants (n ¼ 78). (A) Immediately before BMI 0.051 0.468 0.641
onset of the TSST-G; (B) mid-way through the TSST-G and (C) R2 0.162 (p ¼ 0.003)
immediately after the end of the TSST-G.

Table 1. Descriptive statistics and intercorrelations between cortisol, VASQ attachment measures and demographic data (n ¼ 78).

Correlations
Cortisol Menstrual Smoking
Variables Descriptives recovery Vulnerability Insecurity Proximity Age cycle phase status BMI
Cortisol reactivity 4.47 (6.50) 0.185 0.289* 0.269* 0.177 0.349** 0.141 0.103 0.011
nmol/l M (SD)
Cortisol recovery 4.20 (4.50) 0.020 0.047 0.016 0.276* 0.176 0.015 0.017
M (SD)
VASQ Vulnerability 60.98 (9.81) 0.775** 0.770** 0.209 0.037 0.163 0.119
nmol/l M (SD)
VASQ Insecurity 33.37 (6.39) 0.193 0.247* 0.051 0.001 0.035
M (SD)
VASQ Proximity 27.61 (6.32) 0.075 0.006 0.254* 0.150
M (SD)
Age (years) 20.22 (3.21) 0.054 0.099 0.218
M (SD)
Menstrual cycle phase % 38.5 0.017 0.007
(luteal)
Smoking status 85.9 0.111
% (non-smoker)
BMI 21.46 (3.76)
M (SD)

*p < 0.05, **p < 0.001.

 DOI: 10.3109/10253890.2015.1021676 Stress reactivity and attachment style 147

Another distinguishing feature of this study was the use of

the VASQ for the assessment of attachment style. We chose to
use this self-report tool as it has been shown to perform
somewhat better in predicting depression than other self-
report measures of attachment (Bifulco et al., 2003), as well
as predicting negative psychosocial well-being and mental
health in university students (Carr et al., 2013). The study has
several other strengths in that self-reported menstrual phase,
age, smoking and BMI were all accounted for in the modeling
of cortisol reactivity. The study also controlled for time of day
and collected multiple saliva samples at 10-min intervals for
>60 min, providing a full neuroendocrine response profile,
enabling accurate examination of individual cortisol reactivity
and recovery.
The findings may reflect subtle underlying changes in HPA
axis function linked to attachment style that are important for
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

Figure 2. Attachment style differences in cortisol reactivity. Reactivity

is significantly greater for the insecure anxious attachment style group in
future mental health and well-being. For example, an
comparison to the securely attached group (p ¼ 0.011). enhanced cortisol response to the TSST has been shown to
predict depressive symptoms in young adults (Morris et al.,
2012) and suicidal ideation in female adolescents with a
history of mental health concerns (Giletta et al., 2014). The
cortisol reactivity than participants with secure attachment results are also consistent with evidence showing greater
style. Individuals with avoidant attachment style (high scores cortisol reactivity to the TSST in older adults subjected to
for insecurity and low scores for proximity seeking) did not separation from both parents during childhood (Pesonen et al.,
differ from the secure attachment style group in terms of their 2010) and in young adults exposed to severe pre-natal stress
cortisol reactivity. Attachment style was not associated with (Entringer et al., 2009).
the post-stress recovery in cortisol concentrations. These data The findings may also inform basal HPA axis function as it
provide evidence for an association between attachment style has been shown that the cortisol response to laboratory stress
and increased reactivity of the HPA axis in response to a is positively associated with average cortisol concentrations
standardized group psychosocial stressor in healthy young over the day (Kidd et al., 2014). This may provide an eventual
female participants. route to allostatic overload and negative physical and mental
The results are consistent with the work of Quirin et al. health outcomes (McEwen, 2000; Morris et al., 2012). It may
(2008), which also showed that anxious attachment style also underpin observed aberrant diurnal profiles of cortisol
predicted greater cortisol responding in females with no effect secretion in anxious attachment style (Oskis et al., 2011;
for avoidant attachment. However, the results are in contrast Quirin et al., 2008). The study also provided supportive
with other work showing that avoidant (but not anxious) evidence that avoidant insecure attachment style is somewhat
attachment predicted enhanced stress-induced cortisol similar to secure attachment in terms of neuroendocrine
responding in females (Powers et al., 2006). The results are function, consistent with evidence concerning the cortisol
also inconsistent with other studies showing no relationship awakening response (Oskis et al., 2011), as well as the
between attachment style and cortisol reactivity (Ditzen et al., findings that there are lower health risks in insecure avoidant
2008; Smeets, 2010). individuals compared to those anxiously attached (Bifulco
These discrepancies in findings may in part be attributable et al., 2002a; Fraley & Bonanno, 2004; Sbarra et al., 2013).
to the choice of stressor. We chose to use an adapted form of The current results are limited to healthy young females so
the TSST for use with groups. The TSST is a reliable activator it would be interesting to repeat the study in healthy young
of HPA axis function comprising the key elements of males and with different age ranges. Another limitation is the
uncontrollability and socio-evaluative threat (Dickerson & reliance upon self-reported menstrual phase, not hormonal
Kemeny, 2004). It is not possible to compare the size of the assessment. Also the cross-sectional design, with no long-
cortisol response described here to those that would be term follow-up in relation to health outcomes, means we are
elicited by the individual TSST; hence, whether the group unable to draw any conclusions about whether the observed
nature of the stressor represented a particularity potent results are implicated in future health outcomes.
stimulus for this group is undecided. It would be interesting
to repeat this study using the individual TSST to explore
Conclusions
further this possibility. However, it is noteworthy that the
modifications to the TSST-G employed here (i.e. free group In conclusion, the study used an adapted version of the newly
interactions in the preparatory phase and open visibility developed TSST-G to explore the impact of attachment style
during the stressor) produced a statistically significant cortisol on acute stress responding within a group setting. Healthy
stress reactivity response. This provides opportunities for young females with anxious attachment style showed a more
exploration of other social interventions in stress responding marked cortisol response to the stressor. Results obtained
such as reported by Häusser et al. (2012) in terms of group were not related to age, self-reported menstrual phase,
social identity. smoking status or BMI. The results indicate that differences
 148 N. Smyth et al. Stress, 2015; 18(2): 143–148

in HPA axis activation may provide a pre-clinical indication Kidd T, Carvalho LA, Steptoe A. (2014). The relationship between
cortisol responses to laboratory stress and cortisol profiles in daily
of ill-health vulnerability. life. Biol Psychol 99:34–40.
Kidd T, Hamer M, Steptoe A. (2011). Examining the association
Acknowledgements between adult attachment style and cortisol responses to acute stress.
Psychoneuroendocrinology 36(6):771–9.
We would like to acknowledge the Bial Foundation (grant 72/ Kiecolt-Glaser JK, Newton T, Cacioppo JT, MacCallum RC, Glaser R,
12), who funded this study. Malarkey WB. (1996). Marital conflict and endocrine function: are
men really more physiologically affected than women? J Consult Clin
Declaration of interest Psychol 64(2):324–32.
Kirschbaum C, Klauer T, Filipp S-H, Hellhammer DH. (1995). Sex-
The authors report no conflicts of interest. specific effects of social support on cortisol and subjective responses
to acute psychological stress. Psychosom Med 57(1):23–31.
Kirschbaum C, Pirke KM, Hellhammer DH. (1993). The ‘‘Trier Social
References Stress Test’’–a tool for investigating psychobiological stress responses
Bifulco A, Mahon J, Kwon JH, Moran PM, Jacobs C. (2003). The in a laboratory setting. Neuropsychobiology 28(1-2):76–81.
vulnerable attachment style questionnaire (VASQ): an interview-based McEwen BS. (2000). Allostasis and allostatic load: implications for
measure of attachment styles that predict depressive disorder. Psychol neuropsychopharmacology. Neuropsychopharmacolo 22(2):108–24.
Med 33(6):1099–110. Morris MC, Rao U, Garber J. (2012). Cortisol responses to psychosocial
Bifulco A, Moran PM, Ball C, Bernazzani O. (2002a). Adult attachment stress predict depression trajectories: social-evaluative threat and prior
Downloaded by [University of California Santa Barbara] at 08:11 06 November 2015

style. I. Its relationship to clinical depression. Soc Psych Psych Epid depressive episodes as moderators. J Affect Disord 143(1):223–30.
37(2):50–9. Oskis A, Clow A, Loveday C, Hucklebridge F, Sbarra D. (2014).
Bifulco A, Moran PM, Ball C, Lillie A. (2002b). Adult attachment style. Biological stress regulation in female adolescents: a key role for
II. Its relationship to psychosocial depressive-vulnerability. Soc Psych confiding. J Youth Adolescence 1–12. [Epub ahead of print]. doi:
Psych Epid 37(2):60–7. 10.1007/s10964-014-0182-z.
Brennan KA, Clark CL, Shaver PR. (1998). Self-report measurement of Oskis A, Loveday C, Hucklebridge F, Thorn L, Clow A. (2011). Anxious
adult attachment theory and close relationships. New York: Guilford attachment style and salivary cortisol dysregulation in healthy female
Press. children and adolescents. J Child Psychol Psyc 52(2):111–18.
Brennan KA, Shaver PR. (1995). Dimensions of adult attachment, affect Pesonen A-K, Räikkönen K, Feldt K, Heinonen K, Osmond C, Phillips
regulation, and romantic relationshop functioning. Pers Soc Psychol B DIW, Barker DJP, et al. (2010). Childhood separation experience
21(3):267–83. predicts HPA axis hormonal responses in late adulthood: a nat-
Burnette JL, Davis DE, Green JD, Worthington Jr EL, Bradfield E. ural experiment of World War II. Psychoneuroendocrinology 35(5):
(2009). Insecure attachment and depressive symptoms: the mediating 758–67.
role of rumination, empathy, and forgiveness. Pers Individ Dif 46(3): Pierrehumbert B, Torrisi R, Ansermet F, Borghini A, Halfon O. (2012).
276–80. Adult attachment representations predict cortisol and oxytocin
Carr S, Colthurst K, Coyle M, Elliott D. (2013). Attachment responses to stress. Attach Hum Dev 14(5):453–76.
dimensions as predictors of mental health and psychosocial Powers SI, Pietromonaco PR, Gunlicks M, Sayer A. (2006). Dating
well-being in the transition to university. Eur J Psychol Educ 28(2): couples’ attachment styles and patterns of cortisol reactivity and
157–72. recovery in response to a relationship conflict. J Pers Soc Psychol
Diamond LM. (2001). Contributions of psychophysiology to research on 90(4):613–28.
adult attachment: review and recommendations. Pers Soc Psychol Rev Puig J, Englund MM, Simpson JA, Collins WA. (2013). Predicting adult
5(4):276–95. physical illness from infant attachment: a prospective longitudinal
Dickerson SS, Kemeny ME. (2004). Acute stressors and cortisol study. Health Psychol 32(4):409–17.
responses: a theoretical integration and synthesis of laboratory Quirin M, Pruessner JC, Kuhl J. (2008). HPA system regulation and
research. Psychol Bull 130(3):355–91. adult attachment anxiety: individual differences in reactive and
Ditzen B, Schmidt S, Strauss B, Nater UM, Ehlert U, Heinrichs M. awakening cortisol. Psychoneuroendocrinology 33(5):581–90.
(2008). Adult attachment and social support interact to reduce Repetti RL, Taylor SE, Seeman TE. (2002). Risky families: family social
psychological but not cortisol responses to stress. J Psychosom Res environments and the mental and physical health of offspring. Psychol
64(5):479–86. Bull 128(2):330–66.
Dockray S, Susman EJ, Dorn LD. (2009). Depression, cortisol reactivity, Rifkin-Graboi A. (2008). Attachment status and salivary cortisol in a
and obesity in childhood and adolescence. J Adolesc Health 45(4): normal day and during simulated interpersonal stress in young men.
344–50. Stress 11(3):210–24.
Entringer S, Kumsta R, Hellhammer DH, Wadhwa PD, Wüst S. (2009). Sbarra DA, Borelli JL. (2013). Heart rate variability moderates
Prenatal exposure to maternal psychosocial stress and HPA axis the association between attachment avoidance and self-concept
regulation in young adults. Horm Behav 55(2):292–98. reorganization following marital separation. Int J Psychophysiol
Fraley RC, Bonanno GA. (2004). Attachment and loss: a test of three 88(3):253–60.
competing models on the association between attachment-related Shaver PR, Mikulincer M. (2002). Attachment-related psychodynamics.
avoidance and adaptation to bereavement. Pers Soc Psychol B 30(7): Attach Hum Dev 4(2):133–61.
878–90. Smeets T. (2010). Autonomic and hypothalamic-pituitary-adrenal stress
Giletta M, Calhoun C, Hastings P, Rudolph K, Nock M, Prinstein M. resilience: impact of cardiac vagal tone. Biol Psychol 84(2):290–5.
(2014). Multi-level risk factors for suicidal ideation among at-risk Smyth N, Hucklebridge F, Thorn L, Evans P, Clow A. (2013). Salivary
adolescent females: the role of hypothalamic-pituitary-adrenal axis cortisol as a biomarker in social science research. Soc Personal
responses to stress. J Abnorm Child Psychol 1–14. [Epub ahead of Psychol Compass 7(9):605–25.
print]. doi: 10.1007/s10802-014-9897-2. Stroud LR, Salovey P, Epel ES. (2002). Sex differences in stress
Häusser JA, Kattenstroth M, van Dick R, Mojzisch A. (2012). ‘‘We’’ are responses: social rejection versus achievement stress. Biol Psychiat
not stressed: social identity in groups buffers neuroendocrine stress 52(4):318–27.
reactions. J Exp Soc Psychol 48(4):973–77. von Dawans B, Kirschbaum C, Heinrichs M. (2011). The Trier Social
Jinyao Y, Xiongzhao Z, Auerbach RP, Gardiner CK, Lin C, Yuping W, Stress Test for Groups (TSST-G): a new research tool for controlled
Shuqiao Y. (2012). Insecure attachment as a predictor of depressive simultaneous social stress exposure in a group format.
and anxious symptomology. Depress Anxiety 29(9):789–96. Psychoneuroendocrinology 36(4):514–22.