Co-ruminating increases stress hormones in women

Abstract

Same-sex friendships are an important source of social support and typically contribute to positive adjustment. However, there can be adjustment trade-offs if the friends co-ruminate (i.e., talk excessively about problems) in that co-rumination is related to having close friendships but also to increased internalizing symptoms. The current study utilized an experimental manipulation that elicited co-rumination in young women and thus mirrored an everyday response to stress. Observed co-rumination was associated with a significant increase in the stress hormone, cortisol (after controlling for self-reported co-rumination and for cortisol levels assessed before the discussion of problems). These findings suggest that co-rumination can amplify, rather than mitigate, the hormonal stress response to personal life stressors.

Full text

Co-ruminating increases stress hormone levels in women
Jennifer Byrd-Craven ⁎, David C. Geary, Amanda J. Rose, Davide Ponzi
University of Missouri-Columbia, USA
Received 4 September 2007; revised 17 October 2007; accepted 3 December 2007
Available online 14 December 2007
Abstract
Same-sex friendships are an important source of social support and typically contribute to positive adjustment. However, there can be
adjustment trade-offs if the friends co-ruminate (i.e., talk excessively about problems) in that co-rumination is related to having close friendships
but also to increased internalizing symptoms. The current study utilized an experimental manipulation that elicited co-rumination in young women
and thus mirrored an everyday response to stress. Observed co-rumination was associated with a significant increase in the stress hormone, cortisol
(after controlling for self-reported co-rumination and for cortisol levels assessed before the discussion of problems). These findings suggest that
co-rumination can amplify, rather than mitigate, the hormonal stress response to personal life stressors.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Friendships; Stress; Cortisol; Evolution; Sex differences; Adolescent development
Introduction
Same-sex friendships are core sources of social support for
girls and women during adolescence and through the early
twenties (Carbery and Buhrmester, 1998; Grabill and Kerns,
2000; Rubin et al., 2006). Although disclosure and social sup-
port processes generally are adaptive and protective (Uchino
et al., 1999), recent research identifies a support process with
adjustment trade-offs. Co-rumination is characterized by exten-
sive discussion of problems and involves mutual encouragement
of problem talk, rehashing the details of problems, speculating
about problems, and dwelling on negative affect (Rose, 2002).
Co-rumination in friendships is not only related to feelings of
closeness between the friends but also to depression and anxiety
(Rose, 2002). Co-rumination is more common among girls than
boys in childhood and adolescence (Rose, 2002; Rose et al.,
2007) and predicts increased depression and anxiety over time
for girls but not boys at this age (Rose et al., 2007).
The current study considers co-rumination between women
friends in their late teens and early twenties. The primary aim
is to examine one potential biological underpinning of co-
rumination. Specifically, co-rumination observed in the labora-
tory is predicted to be related to increases in the stress hormone,
cortisol. Co-rumination's persistent focus on problems is ex-
pected to trigger a biological stress response and elevate cortisol.
This may be especially true if co-rumination tends to focus on
social issues because socially evaluative and conflicted situa-
tions are linked with cortisol increases (Dickerson and Kemeny,
2004; Flinn, 2006; Heuther, 1998; McEwen, 1998).
The research is also designed to validate a laboratory method
for studying naturally occurring stress and coping processes in
friendships. The research utilizes a condition designed to elicit co-
rumination about personal problems (Rose et al., 2005)anda
control condition in which friends are not required to discuss
personal problems. Co-rumination is expected to elicit the greatest
increases in cortisol in the condition that prompts women to talk
about problems that are self-identified as stressful to them. Such
findings would indicate that this condition not only elicits social
behavior that mimics natural social behavior but also elicits bio-
logical responses that are consistent with real-world stress
responses.
Method
Participants
Participants were 48 undergraduate women (24 friendship dyads;
Mage = 19.9 years, SD = 2.1) recruited from a larger study (n= 30) and from
A
vailable online at www.sciencedirect.com
Hormones and Behavior 53 (2008) 489 –492
www.elsevier.com/locate/yhbeh
⁎Corresponding author. Department of Psychology, Oklahoma State
University, 116 North Murray, Stillwater, OK 74078, USA. Fax +405 744 8067.
E-mail address: jennifer.byrd.craven@okstate.edu (J. Byrd-Craven).
0018-506X/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.yhbeh.2007.12.002

fliers (n= 18). Women who participated in the larger study were invited to
participate based on their self-reported co-rumination scores; the goal was to
obtain a varied sample in terms of self-reported co-rumination. All dyads
indicated they were “best”or “close”friends.
Procedure
In the first of two laboratory sessions, participants completed the Co-Rumi-
nation questionnaire and a Problem Generation Questionnaire (see Measures
section; women who participated in the larger study completed the co-rumination
questionnaire as part of that study instead of the laboratory assessment). During
the second session, participants were randomly assigned to the Problem Talk
condition (14 dyads; adapted from Rose et al., 2005) or the Control condition
(10 dyads). Saliva samples were collected from both friends (see Salivary
Cortisol) and the friends were video recorded in a five-minute warm-up task
(i.e., planning a menu). Before the warm-up task, each participant in the Problem
Talk condition selected one problem from the Problem Generation questionnaire
for discussion. After the warm-up task, participants in the Problem Talkcondition
were asked to discuss the problem as they normally would and were told that they
could discuss either friend's problem or both. Participants in the Control condi-
tion were asked to design a recreation center. Participants in both conditions were
video recorded during 17-min discussion periods.
The friends were then separated and asked to look at home, garden, travel,
furniture, or architecture magazines. Saliva was collected again after this 15 min
delay when cortisol levels should be close to their post-stressor peak (Sapolsky
et al., 2000).
Salivary cortisol
Saliva was obtained by having participants saturate cotton rolls in their
mouths. The rolls were placed in salivettes® (Salimetrics, State College, PA) and
stored at −20 °C until assayed. Samples for two participants (one dyad) from the
control group were lost and two participants (one from the problem talk group,
one from the control group) produced insufficient volumes for assay (b50 μl).
Samples were assayed using Enzymatic Immunoassay (EIA) (Salimetrics,
Inc.), following standard procedures outlined by Salimetrics, State College, PA.
The procedure is designed to capture the full range of salivary cortisol (0.003 to
3.00 μg/dL). Each EIA kit includes a microtitre plate coated with monoclonal
antibodies to cortisol. Standard cortisol levels (supplied by Salimetrics) and
unknown cortisol levels (study samples) compete with cortisol linked to horse-
radish peroxidase (the enzyme conjugate) for the antibody binding sites. Twenty-
five mL of the standards, controls, and unknowns are pipetted into appropriate
wells on the microtitre plate. Next, 25 mL of assay diluent were pipetted into two
wells to serve as the zero for comparison purposes and into each of the other
wells. A dilution (1:1600) of the enzyme conjugate was made by adding 15 μLof
the conjugate to the 24 mL of assay diluent. Two hundred μL of this solution was
immediately pipetted into each well using a multichannel pipette. The plate was
then mixed on a rotator for 5 min at 500 rpm and incubated at room temperature
for 55 min.
After incubating, unbound components were washed out 4 times with a wash
buffer. Next, 200 μL of the substrate tetramethylbenzidine (TMB) was added to
each well with a multichannel pipette. The solution was mixed on a plate rotator
for 5 min at 500 rpm and the plate placed in the dark at room temperature for an
additional 25 min. Finally, 50 μL of stop solution was added to each well via a
multichannel pipette to stop the enzymatic reaction. This was mixed on a plate
rotator for 3 min. The plate was placed on a plate reader within 10 min of adding
the stop solution. Bound cortisol peroxidase was then measured by the reaction
of the peroxidase enzyme on the substrate tetramethylbenzidine, which produces
color differences. These differences in optical density were read on a standard
plate reader.
All samples from an individual were assayed in duplicate in the same assay
batch. Further, samples from each dyad were assayed in the same batch. Because
all samples were assayed in duplicate, the mean intracoefficient of variation
provides a measure of the average variability for each assay from the same
sample. The mean intracoefficient of variation was 6.55%. The mean inter-
coefficient of variation provides a measure of the average variation from the
controls provided in the assay kits. In other words, it is the average difference
from expected values for the control samples. Mean intercoefficient of variation
for this study was 15.95%.
Measures
Co-rumination questionnaire (Rose, 2002)
Participants rated 27 items assessing co-rumination with same-sex friends on
a 5-point scale. Items assessed discussing problems extensively, rehashing
problems, mutual encouragement of discussion of problems, speculating about
problems, and focusing on negative affect. Items assessed a more extreme form
of problem discussion than items typically used to assess normative self-
disclosure (e.g., “When we talk about a problem that one of us has, we usually
talk about that problem every day even if nothing new has happened”). Scores
were the mean rating across the items (α=.96).
Problem Generation questionnaire (Rose et al., 2005)
Participants generated three current problems and wrote short descriptions of
them.
Coding
The system for coding co-rumination of Rose et al. (2005) was adapted for
this study. The 17 min interactions in both conditions were coded on four
dimensions of co-rumination: mutual encouragement of problem talk, rehashing
problems, speculating about causes and consequences of problems, and focusing
on negative affect. Coders rated each dyad on a 5-point Likert scale in terms of
the degree to which the dyad was characterized by the dimension. Two or three
coders scored each interaction. For each dimension, correlations across coders
were high (range= 0.82 to 0.94). Total observed co-rumination scores were the
sum of the scores across the four dimensions.
Results
Validating random assignment
One-way ANOVAs indicated that participants in the Problem
Talk and Control conditions did not differ on self-reported co-
rumination, F(1,42) b1, or pre-task cortisol level, F(1, 42) b1.
Experimental manipulation
Separate one-way ANOVAs confirmed that the Problem
Talk group scored higher than the Control group on: total
observed co-rumination, F(1, 46) = 60.45, pb0.0001, d=2.26,
mutual encouragement, F(1, 46) = 53.97, pb0.0001, d=2.19,
rehashing, F(1, 46)= 43.68, pb0.0001, d= 1.94, speculating,
F(1, 46) = 18.24, pb0.0001, d= 1.24, and dwelling on negative
affect, F(1, 46) = 51.26, pb0.0001 d=2.10; see Tabl e 1.
Table 1
Means and standard deviations of observed co-rumination and component scores
Variable Problem talk Control
M SD Range M SD Range
Observed co-rumination 13.68 2.56 8.5–17 7.65 2.77 4.5–13
Mutual encouragement 3.61 .84 2–5 1.90 .72 1–3
Rehashing 3.82 .92 2–5 2.10 .85 1–3
Speculating 3.39 .89 2–5 2.25 .95 1–4
Dwell negative affect 2.86 .71 2–4 1.40 .68 1–3
Note. Mutual Encouragement, Rehashing, Speculating, and Dwelling on
Negative Affect were coded on a 1 (low) to 5 (high) scale based on the entire
17 min discussion. Observed Co-Rumination is the total of these four categories.
490 J. Byrd-Craven et al. / Hormones and Behavior 53 (2008) 489–492

Co-rumination and post-task cortisol levels
Regression analyses tested whether the effect of co-rumina-
tion on post-task cortisol varied by condition (Problem Talk
versus Control). Post-task cortisol was predicted from condition,
observed co-rumination, and their interaction (pre-task cortisol
and self-reported co-rumination were covariates). The effect
for condition was significant, β= 0.39, F(1,38) = 4.89, pb0.05.
The effect of observed co-rumination was not significant,
β= 0.17, F(1, 38) b1, but the interaction between condition and
observed co-rumination was, β= 0.45, F(1, 38) = 5.69, pb0.05.
Regression analyses conducted by condition indicated that
observed co-rumination predicted post-task cortisol levels in the
Problem Talk condition, β= .44, F(1, 23) = 3.24, pb0.05, but
not in the Control condition, β=−0.23, F(1, 13) = 1.93, pN0.15
(controlling for pre-task cortisol and self-reported co-rumina-
tion); see Table 2.
Next, separate regression analyses were performed for each
aspect of co-rumination to test whether the effect of that aspect
varied by condition. Post-task cortisol was predicted from con-
dition, each aspect, and the interaction between condition and
the aspect (pre-task cortisol and self-reported co-rumination
were covariates). The interaction between dwelling on negative
affect and condition was significant, β= 0.37, F(1, 38)= 4.18,
pb0.05, although the main effects in this analysis were not
[dwelling on negative affect, β= 0.11, F(1,38) = 1.10, pN0.25;
condition, β= 0.16, F(1,38) = 3.01, p= 0.09]. Analyses con-
ducted by condition indicated that dwelling on negative affect
predicted post-task cortisol in the Problem Talk condition,
β= 0.37, F(1,23) = 3.89, pb0.05, but not in the Control con-
dition, β=−0.06, F(1,13) b1 (controlling for pre-task cortisol
and self-reported co-rumination). Neither significant main ef-
fects nor interactions were found for the other aspects (e.g.,
rehashing).
Discussion
The current study was successful in eliciting co-rumination
and identifying a biological correlate of this social dynamic. As
expected, friends who were in the Problem Talk condition co-
ruminated more than friends in the Control condition. Most
importantly, observed co-rumination was linked with increased
cortisol immediately (within 15 min) following the problem
discussion, suggesting that co-ruminating amplified women's
stress response to these problems. Dwelling on negative affect in
particular contributed to cortisol elevations. Focusing on ne-
gative feelings associated with problems may be more stressful
than engaging in other aspects of co-rumination, such as
speculating about causes and consequences of problems,
which may be more analytical. However, a limitation of this
study is that it is unable to differentiate whether this cortisol
increase is due to increased negative affect in general (Jacobs
et al., 2007) or whether it is due to the process of dwelling on
negative affect during co-rumination. Future studies should
examine the difference between self rumination and co-rumi-
nation to determine whether cortisol increases are amplified by
the social nature of co-rumination or whether dwelling on ne-
gative affect alone is responsible for cortisol elevation.
Note too that co-rumination, specifically, dwelling on nega-
tive affect, predicted increased cortisol only in the Problem Talk
condition. The Control condition women who co-ruminated
often dwelled on problems associated with the task of creating
the recreation center. Co-rumination may need to be about per-
sonal problems to be meaningful enough to elicit a biological
stress response. Also, given that most of the discussions in the
Problem Talk condition focused on social problems, the current
results fit with other findings indicating social evaluation and
social stressors are likely to elicit cortisol release (e.g., Dickerson
and Kemeny, 2004; Heuther, 1998, Flinn, 2006).
Although the current study does not speak to longer-term
effects, it may have implications for understanding links bet-
ween co-rumination and emotional adjustment. Studies indicate
that administering moderate amounts of cortisol to individuals
increases recall of details of events but that higher amounts
impair memory (Abercrombie et al., 2003). On the positive side,
if co-ruminating results in mild to moderate increases in cortisol,
a more accurate memory of the details of the problems could
contribute to increased problem solving efforts. However, more
precise memories of worries and concerns could contribute to
the internalizing symptoms of depression and anxiety that often
occur with co-rumination (Rose, 2002; Rose et al., 2007). This
might be particularly true if the co-rumination creates antici-
patory anxiety and if the ambiguity of the social stressors
prevents habituation of the cortisol response (Gunnar, 2001).
The implications for the well-being of the friendships are not
clear. It could be that friendships are strained as a result of
engaging in behaviors that create a biological stress response.
Because women who co-ruminate often do so frequently, their
relationship may be characterized by repeated spikes in cortisol,
the long-term effects of which are not currently known. How-
ever, previous findings indicate that co-rumination is linked with
the quality of friendships improving over time (Rose et al.,
2007). Perhaps friends perceive one another as being supportive
even though their behavior actually intensifies at least one
component of their biological stress responses to problems.
Finally, the current study makes a methodological contribu-
tion. Previous research examining cortisol stress responses typi-
cally involved administering doses of cortisol (e.g., Abercrombie
et al., 2003) or situations that participants are unlikely to regularly
encounter (e.g., giving a speech, viewing accidents; Abercrombie
et al., 2006; Dickerson and Kemeny, 2004). In contrast, talking
about problems is a common activity among young women. Thus,
the current study employed an experimental paradigm for study-
ing biological stress responses in a relatively naturalistic manner.
The validation of this paradigm is an important methodological
Table 2
Means and standard deviations of pre-task and post-task cortisol levels
Variable Problem talk mean
(n= 26)
Standard
deviation
Control mean
(n= 18)
Standard
deviation
Pre-task 0.41 0.31 0.43 0.32
Post-
task
0.43 0.32 0.32 0.22
Note. Values shown are μg/dL.
491J. Byrd-Craven et al. / Hormones and Behavior 53 (2008) 489–492

contribution that will facilitate the study of other biological
correlates of co-rumination as well as the study of biological
correlates of other support strategies that friends use to help one
another deal with personal and salient life stressors.
Acknowledgments
The authors would like to thank Debi Bell, Gary Brase,
Kendra Cerveny, Rachel Christensen, Brian Hampel, Mary
Hoard, Mary Lemp, Cy Nadler, Cheryl Neiman, Lara Nugent
and Amanda Shocklee for their comments on previous drafts
and their help with various aspects of the project.
References
Abercrombie, H.C., Speck, N.S., Monticelli, R.M., 2006. Endogenous cortisol
elevations are related to memory facilitation only in individuals who are
emotionally aroused. Psychoneuroendocrinology, 31, 187–196.
Abercrombie, H.C., Kalin, N.H., Thurow, M.E., Rosenkranz, M.A., Davidson,
R.J., 2003. Cortisol variation in humans affects memory for emotionally
laden and neutral information. Behavioral Neuroscience, 117, 505–516.
Carbery, J., Buhrmester, D., 1998. Friendship and need fulfillment during three
phases of young adulthood. Journal of Social and Personal Relationships, 15,
393–409.
Dickerson, S.S., Kemeny, M.E., 2004. Acute stressors and cortisol responses: a
theoretical integration and synthesis of laboratory research. Psychological
Bulletin, 130, 355–391.
Flinn, M.V., 2006. Evolution and ontogeny of the stress response to social
challenges in the human child. Developmental Review, 26, 138–174.
Grabill, C., Kerns, K.A., 2000. Attachment style and intimacy in friendship.
Personal Relationships 7, 363–378.
Gunnar, M.R., 2001. The role of glucocorticoids in anxiety disorders: a
critical analysis. In: Vasey, M.W., Dadds, M.R. (Eds.), The develop-
mental psychopathology of anxiety. Oxford University Press, New York,
pp. 143–159.
Huether, G., 1998. Stress and the adaptive self-organization of neural
connectivity during childhood. International Journal of Developmental
Neuroscience, 16, 297–306.
Jacobs, N., Myin-Germeys, I., Derom, C., Delespaul, P., van Os, J., Nicolson, N.
A., 2007. A momentary assessment study of the relationship between
affective and adrenocortical stress responses in daily life. Biological
Psychology, 74, 60–66.
McEwen, B.S., 1998. Protective and damaging effects of stress mediators. New
England Journal of Medicine, 338 (3), 171–179.
Rose, A.J., 2002. Co-rumination in the friendships of girls and boys. Child
Development, 73, 1830–1843.
Rose, A.J., Carlson, W., Waller, E.M., 2007. Prospective associations of co-
rumination with friendship and emotional adjustment: considering the
socioemotional trade-offs of co-rumination. Developmental Psychology, 43,
1019–1031.
Rose, A.J., Schwartz, R.A., Carlson, W., 2005, April. An observational assessment
of co-rumination in the friendships of girlsand boys. In: Coie, J.D., Putallaz, M.
(Eds.), The costs and benefits of interpersonal processes underlying girls'
friendships. Symposium conducted at the biennial meeting of the Society for
Research in Child Development.
Rubin, K.H., Bukowski, W., Parker, J.G., 1998. Peer interactions, relationships,
and groups. In: Damon, W., Eisenberg, N. (Eds.), Handbook of child
psychology: Vol. 3. Social, emotional, and personality development. Wiley,
New York, pp. 619–700.
Sapolsky, R.M., Romero, L.M., Munck, A.U., 2000. How do glucocorticoids
influence stress responses? Integrating permissive, suppressive, stimulatory,
and preparative actions. Endocrine Reviews, 21, 55–89.
Uchino, B.N., Uno, D., Holt-Lunstad, J., 1999. Social support, physiological
processes, and health. Current Directions in Psychological Sciences, 8,
145–148.
492 J. Byrd-Craven et al. / Hormones and Behavior 53 (2008) 489–492