Health Psychology Review: Click For Updates

This article was downloaded by: [New York University]
On: 24 April 2015, At: 08:24

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Health Psychology Review

Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/rhpr20
Expressive writing interventions in

cancer patients: a systematic review
a
Erin L. Merz , Rina S. Fox & Vanessa L. Malcarne
SDSU/UCSD Joint Doctoral Program in Clinical Psychology, 6363

Alvarado Court, Suite 103, San Diego, CA 92120-4913, USA
Accepted author version posted online: 05 Feb 2014.Published
online: 18 Feb 2014.
Click for updates

To cite this article: Erin L. Merz, Rina S. Fox & Vanessa L. Malcarne (2014) Expressive writing
interventions in cancer patients: a systematic review, Health Psychology Review, 8:3, 339-361, DOI:
10.1080/17437199.2014.882007
To link to this article: http://dx.doi.org/10.1080/17437199.2014.882007
PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
Content) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Downloaded by [New York University] at 08:24 24 April 2015
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Health Psychology Review, 2014

Vol. 8, No. 3, 339361, http://dx.doi.org/10.1080/17437199.2014.882007
Expressive writing interventions in cancer patients:

a systematic review
Erin L. Merz*, Rina S. Fox and Vanessa L. Malcarne
SDSU/UCSD Joint Doctoral Program in Clinical Psychology, 6363 Alvarado Court, Suite 103,
San Diego, CA 92120-4913, USA
(Received 16 May 2013; accepted 8 January 2014)

Decades of research have suggested that expressive writing produces physical and
psychological benefits in controlled laboratory experiments among healthy college
students. This work has been extended to clinical and medical populations, including
cancer patients. Although expressive writing could be a promising and inexpensive
intervention for this population, the effects have not been systematically examined in
oncology samples. A systematic review using PRISMA guidelines was conducted for
experimental trials of cancer patients who participated in an expressive writing
intervention. PsycINFO and PubMed/Medline were searched for peer-reviewed
studies. Thirteen articles met the inclusion/exclusion criteria. Although the majority
of the intervention effects were null, there were several main effects for expressive
writing on sleep, pain, and general physical and psychological symptoms. Several
moderators were identified, suggesting that expressive writing may be more or less
beneficial based on individual characteristics such as social constraints. The reviewed
studies were limited due to representativeness of the samples, performance, detection
and patient-reported outcomes biases, and heterogeneity of the intervention protocol
and writing prompts. Future studies with rigorous designs are needed to determine
whether expressive writing is therapeutically effective in cancer patients.
Keywords: expressive writing; cancer; adjustment; coping; emotional disclosure;
oncology
I write about illness to work out some terms in which it can be acceptedSeizing the
opportunity means experiencing it fully, then letting go and moving on.
- Sociologist Arthur W. Frank (1991, p. 3) on writing about his cancer
Cancer can be an overwhelming and traumatic event in a persons life, impacting not only
physical health but also emotional, social and spiritual functioning. Over the past several
decades, much attention has been paid to how cancer affects the whole person, and
particularly how to promote psychosocial adjustment to the disease. Interventions
including psychotherapy (Hart et al., 2012), peer groups (Hoey, Ieropoli, White, &
Jefford, 2008) and Internet-based support (Hybye et al., 2010) have been used in cancer
patients with some benefits. Another complementary intervention that has gained
substantial interest among psycho-oncology clinicians is expressive writing (EW), an
*Corresponding author. Email: [email protected]

2014 Taylor & Francis
340
E.L. Merz et al.
emotional disclosure technique wherein a person writes about a major life event or
trauma.
The first standardised EW prompt, developed in 1986 by Pennebaker and Beall,
instructed respondents to write for approximately 20 minutes over four consecutive days
about their deepest emotions and thoughts regarding traumatic/upsetting experiences.
Early studies reported that, compared to control participants who wrote factual essays
about superficial (i.e., non-traumatic) topics, those who wrote expressively gained
physical and psychological health benefits over time (e.g., Pennebaker, 1993), suggesting
that a simple writing exercise can improve health. These findings have been supported by
Frattarolis (2006) meta-analysis, which reported a small health-enhancing effect of
written or spoken emotional disclosure across 10,994 people (unweighted mean effect
size = .075; d = .15).
How does expressive writing work?
Several theoretical frameworks have been proposed to explain the mechanisms
underlying the benefits of EW. There is likely no single explanatory route, but rather a
complex combination of interconnected and mutually influential processes. It is possible
that EW serves a cathartic function, allowing the writer to express and release their
emotions, which in turn improves their well-being (Pennebaker & Chung, 2007).
Cognitive restructuring of a trauma and habituation to the emotions associated with that
trauma may also facilitate the effectiveness of EW (Pennebaker & Chung, 2011). That is,
repeatedly writing about ones experience with cancer may serve an exposure function.
Writing may lead to initial distress, but it also enables the writer to process, reflect and
reframe their experience, extinguishing the connection between the event and their
emotional reaction to it, and assimilating it into their self-schema (Low, Stanton, &
Danoff-Burg, 2006; Pennebaker & Chung, 2011). Another explanation is that EW offers
an opportunity for personal mastery and self-regulation (Creswell et al., 2007; Lepore,
Greenberg, Bruno, & Smyth, 2002). EW may enable people to step back and observe
how they handled their cancer experience, increasing their confidence to manage stress
and regulate their emotions, thoughts and behaviours. It is also possible that EW serves to
restore a sense of perceived control, which can be lost or diminished in the context of a
serious illness (Andersson & Conley, 2008).
Written content
The linguistic characteristics of EW passages provide information regarding whether the
intervention was followed (i.e., manipulation check) and the emotional processing of the
writer, which can provide insight into the mechanisms underlying the impact of EW.
However, reliably and validly evaluating written content is challenging. Computer text
analysis programs have been developed, including the Linguistic Inquiry Word Count
program (LIWC; Pennebaker, Booth, & Francis, 2007), which calculates the frequency
and percentage of words in 74 content categories. Using LIWC, researchers have
demonstrated that emotional (e.g., happy and depressed) and cognitive mechanism (e.g.,
because and realise) words predict changes in physical and psychological health after
EW (Pennebaker, Mayne, & Francis, 1997). Although such programs offer quick,
inexpensive and reliable assessment, they are not sensitive to contextual aspects of
language (e.g., tone, cultural jargon and abstraction; Tausczik & Pennebaker, 2010).
341
Trained independent judges and self-ratings have also been widely adopted to
appraise aspects of written passages (e.g., degree of emotionality, personal meaning and
narrative structure) that are not detected by computers. Both methods have advantages
and shortcomings. Judges are more objective than self-rating, but they are time
consuming and costly, and inter-rater reliability can be problematic (Tausczik &
Pennebaker, 2010). Although self-rating is easier to implement, it assumes a sophisticated
level of evaluation on behalf of the writer, and ratings may be influenced by recognition
of condition assignment (i.e., expectation/performance bias from the informed consent
process and knowledge of the prompt). However, both methods have been shown to
sufficiently analyse text characteristics (e.g., Danoff-Burg, Mosher, Seawell, & Agee,
2010; Smyth, True, & Souto, 2001).
Expressive writing and cancer

Given that EW has been shown to promote health in relatively healthy people, there has
been an increasing interest in its therapeutic potential. As such, Pennebaker and Bealls
(1986) prompt has been tailored to accommodate clinical populations. EW has been
investigated as a potentially useful intervention for psychosocial adjustment to medical
illness in numerous disease groups, including rheumatoid arthritis (Broderick, Stone,
Smyth, & Kaell, 2004), asthma (Harris, Thoresen, Humphreys, & Faul, 2005), HIV
(Rivkin, Gustafson, Weingarten, & Chin, 2006) and cardiovascular disease (Manzoni,
Castelnuovo, & Molinari, 2011). There have also been a number of recent studies on EW
in cancer (e.g., Kallay & Baban, 2008; Laccetti, 2007; Lu, Zheng, Young, KagawaSinger, & Loh, 2012; Owen et al., 2006; Smith, Anderson-Hanley, Langrock, & Compas,
2005; Zakowski, Herzer, Barrett, Milligan, & Beckman, 2011). Because many cancer
patients report feeling emotionally inhibited (Servaes, Vingerhoets, Vreugdenhil, Keuning, & Broekhuijsen, 1999; Zakowski et al., 2003), and because emotional inhibition has
been linked with poorer psychological functioning (Tamagawa et al., 2013), EW may
offer an outlet for patients to express their feelings related to cancer.
EW is time- and cost-effective, can be easily self-administered, and people generally
report a high level of satisfaction with it (Frattaroli, 2006). Given this, it is not surprising
that EW has already been employed at cancer centres around the world. However, the
cumulative evidence of EW in cancer has not yet been systematically examined,
suggesting that such wide dissemination could be premature. Therefore, the primary
objective of the present study was to examine the efficacy of EW on physical and mental
health in cancer patients. A second, exploratory aim was to identify the text analysis
methods used in each study, determine the purpose of this analysis and evaluate the
related findings.
Methods
Search strategy
The search was limited to peer-reviewed articles1 published in English after 1986, as this
was the year that the first EW study using Pennebakers prompt was published. Articles
were identified via a search of PsycINFO (1986 to November 2012) and PubMed/
Medline (1986 to November 2012) using combinations of expressive writing (expressive
writing OR emotional disclosure OR written emotional disclosure OR emotional
expression OR creative writing OR Pennebaker OR self-disclosure OR cancer narrative
342
E.L. Merz et al.
OR illness narrative OR reflective writing OR writing cure) and cancer (cancer OR

oncology OR breast cancer OR prostate cancer OR neoplasm OR metastatic OR
leukaemia OR lymphoma OR lung cancer OR cancer clinic OR tumour OR oncology OR
oncologist OR malignant OR malignancy OR chemotherapy OR radiation) terms found
anywhere in the text. Backward (examining references of articles found in the keyword
search) and forward (examining articles cited by those in the keyword/backward
searches) searches were subsequently conducted.
Study selection
Supplemental Figure 1 illustrates the selection process. One-hundred sixty articles were
screened for inclusion, with four studies added from the backward/forward searches.
Studies were independently screened by the first (E.M.) and second (R.F.) authors.
Disagreements were resolved by collaborative reanalysis of the studies in question. To be
included, a study was required to have an experimental design with an EW group
[expressive writing for at least a single 20-minute session using the Pennebaker and Beall
(1986) paradigm] and a control group for comparison. Studies needed to measure a
physical and/or mental health outcome. Participants needed to (i) have any cancer
diagnosis; and (ii) be undergoing or have undergone medical treatment for their cancer
(e.g., surgery, radiation and chemotherapy). To avoid drawing omnibus conclusions from
multiple publications that reported findings from the same sample, articles that presented
results from a previously published study were excluded. Studies without quantitative
analysis were also excluded.
Data extraction
Information from the 13 included articles was recorded in an electronic spreadsheet
designed by the first author. Data were collected on (i) participant characteristics (i.e.,
sample size, demographics and disease); (ii) study characteristics (i.e., design, and
experimental and control group details); (iii) session dosing (i.e., number, spacing and
duration); (iv) text analysis details; (v) health outcomes; (vi) statistical values for EW
effects; (vii) participant feedback; and (viii) risk of bias. Standardised effect sizes
(Cohens d) were calculated for the standardised mean difference between the
intervention and control groups based on statistical values and descriptive statistics
provided in the original manuscripts. For mediation and moderation analyses, p-values
were listed in the final table.
The Cochrane Collaborations tool for assessing risk of bias was used. Random
sequence generation (i.e., selection bias) was rated as low risk if participants had equal
possibility of being assigned to either the treatment or control group, and allocation for
each participant could not be predicted by the investigator (e.g., computer-generated
randomisation schedule); high risk if group assignment could be predicted by the
investigator (e.g., sequential assignment); and unclear risk if no details regarding group
assignment or the method of random assignment were reported. Allocation concealment
(i.e., selection bias) was rated as low risk if authors reported that allocations were
concealed such that investigators or participants could not possibly foresee group
assignment (e.g., central allocation or sequentially numbered, opaque and sealed
envelopes); high risk if allocation could be foreseen (e.g., based on case record number
and sequential by admission into the study); and unclear risk if authors did not report
concealment or describe the method used. Blinding of participants and personnel
343
(i.e., performance bias) is problematic in all EW studies because a true double-blind

design is not possible given the nature of EW tasks. Although all EW studies are high risk
because participants will be aware of group assignment upon receiving their instructions,
blindness and/or awareness of study hypotheses of personnel who interacted with
participants during the intervention was noted. Blinding of outcome assessment (i.e.,
detection bias and patient-reported outcomes) was rated as low risk under the following
conditions: for interviews, personnel were blind to group assignment and efforts were
made to mask the purpose of the study for the participants; for written self-report
assessments, efforts were made to mask the purpose of the study for the participants; for
medical/biological assessments, personnel were blind to group assignment or assessments
were made via chart review. Blinding of outcome assessment was rated as high risk if
personnel were not blind to group assignment or there was evidence that participants were
aware of the study purpose; and unclear risk if there were insufficient details to determine
whether there were efforts to mask participant knowledge of the studys purpose.
Incomplete outcome data (i.e., attrition bias) was rated as low risk if 30% of patients
were lost to follow-up, and rates and reasons for attrition were similar across study
groups; high risk if > 30% of patients were lost to follow-up, and rates and reasons for
attrition across study groups were different or were not reported; and unclear risk if no
details regarding loss to follow-up were reported. Selective reporting (i.e., reporting bias)
was rated as low risk if all outcomes listed in the Methods section were reported on, and
there was no other suggestion of selective outcome reporting; and high risk if there was
evidence that not all outcomes were reported or suggestion of selective outcome
reporting.
Results
Study characteristics
Descriptions of the samples and studies are presented in Table 1. Sample sizes ranged
from 30 to 507. A variety of cancers of all stages were represented, with breast cancer
being the most common (9 of the 13 samples). Average age ranged from 49.5 to 70.4
years, with the majority of studies reporting a mean age in the 50s. Most samples were
predominantly female given the cancer diagnoses being studied (e.g., breast and
gynaecological). Five studies did not describe the ethnic composition of the sample.
Although Cepeda et al. (2008) did not report ethnicity, the patients were recruited from
oncology clinics in Bogot, Colombia, and were presumably of Latino background. In the
eight studies that reported ethnicity, participants were predominantly White, with the
exception of the study by deMoor et al. (2008), which was 17% Latina, 10% African
American and 10% other.
The EW interventions were heterogeneous. Dosing varied from one to four writing
sessions. Gaps between sessions ranged from consecutive days to several weeks. The
emotional prompt specific to cancer (in which participants write about their deepest
thoughts and feelings/emotions) was the most widely used, although variations were
noted. One study (Jensen-Johansen et al., 2012) used the emotional prompt but allowed
participants to write about either their cancer or any trauma [56% of patients wrote about
their cancer at least once (n = 108), 44% of patients always wrote about another trauma
(n = 85)]. The authors also reported varying the instructions slightly between sessions to
avoid uniformity (Jensen-Johansen et al., 2012, p. 1494). Craft, Davis, and Paulson
(2013) used the emotional prompt for two experimental groups: one group wrote
Task and follow-up
Results; (Cohens d) for main effects
Cepeda et al. (2008)

N = 198 patients
Any cancer
stages 14
Race: NR
Mage = 48.5 12.4
64.0% female
EW: STORY (n = 65)

Control: SURVEY (n = 67)
USUAL (n = 66)
Three sessions over 3 weeks
Follow-up at weeks 18; first 3 follow-ups
took place during the 3 EW weeks
Over 8 weeks, EW did not directly improve pain or well-being (insufficient information
to calculate ds; ps > .05).
Emotional disclosure interacted with EW to predict pain and well-being over time.
Respondents with at least one essay rated as very much disclosure (vs. respondents
with no essays rated as very much disclosure) reported improved pain (2.5
difference; 95% CI: 4.2 to 0.9) and well-being (+1.37 difference; 95% CI: 1 to 1.7).
Craft et al. (2013)

N = 97
Breast cancer
stages 03
Race: 92.8% White
Mage = 56 10.5
All female
EW: EMO (n =26)

EMO (n = 19)
Control: NWc (n = 22)
USUAL (n = 30)
Four sessions, consecutive days
Follow-up at 1, 6 months
At 1 and 6 months, EMO did not significantly improve quality of life compared to NWc
(ds = .40, .06) or USUAL (ds = .62, .56). At 1 and 6 months, EMO did not
significantly improve quality of life compared to NWc (ds = .62, .48) or USUAL (ds =
.34, .25).
For EMO, quality of life increased from baseline to 1 and 6 months (ps < .05). For NWc,
quality of life increased from baseline to 1 month (p < .05).
deMoor et al. (2002)

N = 42
Renal cell carcinoma
stage 4
Race: NR
Mage = 56.4 NR
85.7% male
EW: EMO (n = 21)

Control: NWb (n = 21)
Four sessions over 4 weeks
Follow-up immediately after last writing
session, 4, 6, 8, 10 weeks
Scores were pooled across the five follow-ups. Over time, EW did not significantly
improve distress (d = .35), perceived stress (d = .17) or overall mood disturbance (d =
.18) excepting the vigour subscale (d = .64). EW significantly improved overall sleep
disturbance (d = .64), sleep quality (d = 1.3), sleep duration (d = .87), daytime
dysfunction (d = .87); however, sleep latency (d = .65), sleep efficiency (d = .11),
sleep disturbances (d = .44), sleep medications (d = .09) not significantly changed.
deMoor et al. (2008)

N = 49
Breast cancer
stages 2 and 3
Race: 63% White
Mage = 53.5 10.4
All female
EW: EMO (n = 24)

Control: NWb (n = 25)
Four sessions over 7 days, 3 weeks prior to
surgery
Follow-up at 3 days pre-surgery and 2
weeks post-surgery
At 3 days pre-surgery and 2 weeks post-surgery, EW did not directly improve distress (ds
= .07, .01), perceived stress (ds = .24, .16), sleep disturbance (ds = .13, .33), worst
pain (ds = .10, .67), least pain (ds = .28, .45), average pain (ds = .05, .40) or pain
interference (ds = .23, .54). EW reported increased sleep medication use 3 days before
surgery (d = .57); group differences at baseline were not reported.
Social constraints interacted with EW to predict pre-surgery (p = .05) and post-surgery (p
= .05) average pain. High social constraints in EW associated with less pain; low
social constraints in EW associated with more pain.
E.L. Merz et al.
Study and sample
344
Table 1. Studies (N = 13) reporting experimental trials of expressive writing interventions in cancer patients.
Table 1 (Continued)
Task and follow-up
EW: MULTIPLE (n = 38)

Control: USUAL (n = 42)
Four sessions, consecutive days (different
prompt each day)
Follow-up at 1, 3, 6 months
At 1, 3 and 6 months, EW did not directly improve quality of life (ds = .16, .00, .07),
mood disturbance (ds = .15, .07, .10), healthcare utilisation or satisfaction with
perceived practical support (insufficient information to calculate latter ds; ps > .05).
EW directly improved perceived emotional support (ds = .36, .48, .50; overall d over
time = .47) at all three follow-ups.
Henry et al. (2010)

N = 80
Breast cancer
stages 14
Race: NR
Mage = 58.9 NR
All female
EW: BEN (n = 40)

One session
At 3 and 9 months, EW did not directly improve somatisation or retarded activity

(insufficient information to calculate ds; ps > .05). EW directly improved depressive
symptoms (d = .76), physical symptoms (d = .76), mood disturbance (d = .58) at 3
months, but not 9 months (depressive d = .18, physical d = .04, mood disturbance d =
.21).
Location (rural vs. urban) did not interact with EW (p > .05).
Jensen-Johansen
et al. (2012)
N = 507
Breast cancer
stages 1-2
Race: NR
Mage = 53.6 9.1
All female
EW: EMO (n = 253)

Control: NW (n = 254)
Three sessions over 3 weeks
At 3 and 9 months, EW did not directly improve distress (ds = .15, .04), intrusive
thoughts (ds =.07, .03), avoidance (ds = .18, .08), depression (ds = .12, .07), negative
mood (ds = .04, .10) or positive mood (ds =.06, .01).
Externally oriented thinking interacted with EW to predict distress at 3 months (p =
.021). In EW (vs. NW), lower external orientation predicted less distress. Difficulties
describing feelings interacted with EW to predict positive mood at 3 months (p =
.049). In NW (vs. EW), greater difficulties describing feelings predicted better mood.
Writing topic (cancer vs. another trauma) moderated the effects of EW to predict
avoidance (p < .05), depression (p = .013), negative mood (p = .052), positive mood (p
= .014). Writing about another trauma (vs. cancer) predicted less avoidance. Writing
about cancer (vs. NW) improved depression and mood. Number of essays written,
social constraints, other alexithymia variables did not interact with EW (ps > .05).
345
Gellaitry et al.
(2010)
N = 80
Breast cancer
stages 12
Race: NR
Mage = 57.9 9.9
All female
Study and sample
Task and follow-up
Low et al. (2010)

N = 62
Breast cancer
stage 4
Race: 87% White
Mage = 53.8 10.3
All female
EW: EMO (n = 31)

Follow-up at 3 months
At 3 months, EW did not directly improve depression (d = .05), intrusive thoughts (d =

.26), somatic symptoms (d = .09) or sleep disturbances (d = .18).
Emotional support interacted with EW to predict intrusive thoughts (p = .001). Low (vs.
high) support in EW had fewer intrusions. Emotional support interaction with EW to
predict somatic symptoms (trend; p = .08); same pattern. Time since diagnosis
interacted with EW to predict somatic symptoms (p = .034), sleep quality (p = .023).
Diagnosis < 20 months ago (vs. > 20 months) in EW reported fewer symptoms (p =
.048). Diagnosis > 20 months ago (vs. < 20 months) in EW reported more sleep
disturbances (p = .025).
Mosher et al. (2012)

N = 86
Breast cancer
stage 4
Race: 81% White
Mage = 57.9 12.1
All female
EW: EMO (n = 44)

Four sessions over 4-7 weeks
Follow-up at 8 weeks
At 8 weeks, EW did not directly improve existential well-being (meaning/peace d = .25,

demoralisation d = .14), distress (d = .07), depressive symptoms (d = .01), anxiety (d =
.23), sleep (d = .23) or fatigue (d = .28). EW directly increased uptake of mental health
services (d = 1.01), as compared to NW, at 8 weeks.
Rosenberg et al.
(2002)
N = 30
Prostate cancer
stage NR
Race: 97% White
Mage = 70.4 5.4
All male
EW: EMO (n = 16)

Four sessions, consecutive days
At 6 months, EW did not directly improve immune-directed cancer activity markers

[PSA-specific CD4 helper cells (d = .00), PSA-specific CD8 cytolytic cells (d = .36),
IL-10 (d = .52), IL-4, TNF (insufficient information to calculate ds)], psychological
symptoms, quality of life or coping (insufficient information to calculate ds; ps > .05).
At 3 and 6 months, EW directly improved pain (ds = .57, .98); significant trends for
EW on decreased healthcare utilisation (ds = .13, .51), decreased medication use (ds =
.62, .29).
Stanton et al. (2002)

N = 60
Breast cancer
EW: EMO (n = 21)

BEN (n = 21)
Neither EW group directly improved distress, vigour or quality of life (insufficient

information to calculate ds; ps > .05) at 1 or 3 months, or somatic symptoms at 1
month (EMO d = .58, BEN d = .10). EW directly improved somatic symptoms at 3
E.L. Merz et al.
Study and sample
346
Table 1 (Continued)
Table 1 (Continued)
Task and follow-up
stages 12
Race: 93% White
Mage = 49.5 12.2
All female

months (EMO d = .89, BEN d = .55). EW predicted fewer cancer-related (EMO d =

.94, BEN d = .70) but not other medical appointments (EMO d =.19, BEN d =.16) at
3 months.
Cancer-related avoidance interacted with EW to predict distress (1 month p = .0015, 3
month p = .0148). EMO with low (vs. high) avoidance and BEN with high (vs. low)
avoidance predicted less distress. No interaction for vigour, quality of life, somatic
symptoms, medical appointments (ps > .05).
Walker et al. (1999)

N = 39
Breast cancer
stages 12
Race: 95% White
Mage = 53.6, range:
2976
All female
EW: 1 dose EMO (n = 11)

3 dose EMO (n = 14)
Control: ATT (n = 14)
One or three sessions over 14 days
Follow-up at 1, 4-6, 16, 28 weeks
At the four follow-ups, EW (1 and 3 doses) did not directly improve (p > .0125) negative
affect (1 dose ds = .22, .53, .64, .49; 3 dose ds = .11, .32, .48, .49), positive affect (1
dose ds = .79, .31, .49, .66; 3 dose ds = .40, .43, .05, .25), avoidance of cancer
reminders (1 dose ds = .04, .48, .35, .15; 3 dose ds = .00, .20, .01, .02) or intrusiveness
of thoughts (1 dose ds = .15, .13, .39, .60; 3 dose ds = .01, .18, .04, .34).
Zakowski et al.
(2004)
N = 104
Prostate,
gynaecological
stages 14
Race: 95.2% White
Mage = 59.8 11.1
51.9% female
EW: EMO (n = 62)

Three sessions, consecutive days
Follow-up at 6 months
At 6 months, EW did not directly improve distress (d = .03), avoidance (d = .08) or

intrusive thoughts (d = .23).
Social constraints interacted with EW to predict distress (p < .04) and avoidance (p <
.02). High social constraints in EW (vs. NW) associated with less distress, less
avoidance. Avoidance partly mediated the interaction as it predicted distress. Social
constraints did not interact with EW to predict intrusive thoughts (p > .05).
347
ATT = non-cancer attention; BEN = benefit-finding (cancer) prompt; EMO = emotional (cancer) prompt; EMO = emotional (any trauma) prompt; MULTIPLE= emotional, benefitfinding, cognitive-appraisal, coping prompts; NR = not reported; NW = neutral writing (trivial); NWc = neutral writing (cancer); NWb = neutral writing (health behaviour); STORY =
cancer story; SURVEY = took survey; USUAL = usual care
Study and sample
348
E.L. Merz et al.
emotionally about cancer, the other wrote emotionally about any trauma (the authors did
not report whether any patients in the latter group chose to write about cancer). Stanton
et al. (2002) used two experimental groups: one wrote emotionally about cancer, the other
wrote about their positive thoughts and feelings related to cancer (i.e., benefit-finding).
Henry, Schlegel, Talley, Molix, and Bettencourt (2010) used the benefit-finding prompt
only. Alternatively, the study by Gellaitry, Peters, Bloomfield, and Home (2010) used a
different topic for each of the four writing sessions (emotional, benefit-finding, cognitive
appraisals of cancer and coping strategies); and the study by Cepeda et al. (2008)
prompted patients to write a story about how cancer affected their lives (p. 625).
The majority of control conditions instructed participants to engage in writing. The
studies by Low, Stanton, Bower, and Gyllenhammer (2010) and Stanton et al. (2002)
used a neutral prompt regarding the facts of their cancer (e.g., number of chemotherapy
sessions, oncologists name and medication). Alternatively, deMoor et al.s (2002, 2008)
neutral prompt asked participants to write factually about health behaviours (e.g., diet and
sleep). Three studies used a neutral prompt to write about the facts of a trivial/everyday
topic (e.g., daily activities; Jensen-Johansen et al., 2012; Mosher et al., 2012; Zakowski,
Ramati, Morton, Johnson, & Flanigan, 2004). Craft et al. (2013) had two control
conditions: neutral writing about the facts of their cancer2 and usual care (with no
writing). Cepeda et al. (2008) had two conditions: usual care and completing a pain
questionnaire. Usual care was the only control condition in three studies (Gellaitry et al.,
2010; Henry et al., 2010; Rosenberg et al., 2002). One study used an attentional control
wherein patients met with a researcher to discuss non-cancer life events (Walker, Nail, &
Croyle, 1999).
Effects of expressive writing on physical and mental health
The effects of EW on physical and mental health outcomes are presented in Table 1. All
13 studies reported some null effects of experimental condition on the outcomes of
general or existential well-being, quality of life, social support satisfaction, coping,
distress, stress, affect, mood, depression, intrusive thoughts, avoidance, somatisation,
vigour, pain, sleep characteristics, healthcare utilisation and cancer-related immune
functioning. Six studies reported some significant benefits for EW. Standardised effect
sizes for statistically significant main effects indicating that EW improved physical and/or
mental health were small (d = .20) to large (d = 1.21) in magnitude, with the majority
being in the moderate range (ds = .54 to .76). Improvements were noted in energy and
sleep characteristics (deMoor et al., 2002), depressive (Henry et al., 2010) and physical
symptoms (Henry et al., 2010; Stanton et al., 2002), emotional support (Gellaitry et al.,
2010), pain (Rosenberg et al., 2002), uptake of mental health services (Mosher et al.,
2012) and healthcare utilisation (Stanton et al., 2002).
Six studies reported significant moderators of EW. The extent of perceived social
constraints (i.e., inhibition from expressing thoughts/feelings due to social restrictions or
norms) affected the impact of the intervention in two studies (deMoor et al., 2008;
Zakowski et al., 2004). deMoor et al. (2008) found that patients with high levels of social
constraints who wrote expressively had lower average daily pain before and after surgery
compared to high-constraint patients who wrote about health behaviours. The pain
reported by the EW group actually paralleled all patients who reported low social
constraints, suggesting that EW buffers the deleterious effects of inhibition. Similarly,
Zakowski et al. (2004) reported that patients with high social constraints who wrote
349
expressively had lower distress and avoidance of cancer-related thoughts compared to

high-social-constraint controls who wrote about trivial topics. Again, the levels of distress
and avoidance reported by the EW group paralleled patients with few social constraints.
Other studies found evidence that individual differences including emotional support,
time since diagnosis, intentional avoidance of cancer-related thoughts/feelings and
alexithymia (i.e., inability to identify and describe ones emotions) moderated the
effectiveness of EW in patients with breast cancer. Low et al. (2010) found that women
with poor emotional support at baseline were benefited by EW such that they had fewer
intrusive thoughts and somatic symptoms (trend only; p = .08) at follow-up, whereas
women with greater support were not impacted by EW. In this same study, women who
were diagnosed recently and wrote expressively reported fewer somatic symptoms at
follow-up, whereas women who had been diagnosed for more than 20 months were not
benefited by EW. This study also revealed a potentially harmful result of EW for women
who had been living with cancer for a longer time: women in the EW condition who were
diagnosed more than 4.7 years prior to the study reported increased sleep disturbances. In
Stanton et al.s (2002) sample, women who reported low levels of avoidance and used the
emotional prompt (but not the benefit-finding prompt) reported lower distress than
women with high avoidance. There was also a trend in the opposite direction for the
benefit-finding prompt: women with high avoidance who wrote about benefit-finding
reported less distress than women who were less avoidant (Stanton et al., 2002). In the
study by Jensen-Johansen et al. (2012), two aspects of alexithymia (i.e., externally
oriented thinking and difficulty describing feelings) moderated the effectiveness of EW.
Women who reported a higher cognitive focus on inner experiences who wrote
expressively reported lower distress; women with difficulty describing their feelings
who wrote expressively reported a more positive mood (Jensen-Johansen et al., 2012).
Writing topic and level of emotional disclosure within writing passages were also
significant moderators. In Jensen-Johansens (2012) study, wherein patients could choose
their EW topic (cancer vs. another trauma), women who wrote about another trauma
reported lower avoidance, whereas women who wrote about cancer reported decreased
depression and improved mood. Notably, patients who wrote about cancer were more
likely to have a history of critical medical interventions (e.g., mastectomy and
chemotherapy), pain medication usage and psychotherapy (Jensen-Johansen et al.,
2012). Only Cepeda et al. (2008) examined the influence of linguistic characteristics in
EW texts; these findings are discussed in the Measurement of Written Content section.
Several studies reported qualitative feedback regarding patient experiences with EW.
Gellaitry et al. (2010) noted that EW gave participants the opportunity to disclose
feelings they found difficult to share with those closest with them because they did not
want to cause them more worry (p. 82). Craft and colleagues (2013) also published
positive written reflections suggesting that the patients listening to themselves led to
perceptions of increased quality of life (p. 312). Similarly, Walker and colleagues (1999)
found that many women described the exercise as helpful with regard to their end-oftreatment concerns, expression of their emotions and consolidation of their thoughts
without burdening their loved ones. deMoor et al. (2002) found comparably favourable
results; over 80% of the sample reported that they would like to participate again, and
90% reported that they would recommend EW to a friend. In another sample, the majority
of women (59%) stated that they felt positively about writing, while 34% felt neutrally
(Henry et al., 2010). A number of these participants also indicated that the intervention
aided in their coping efforts, although some expressed that they did have difficulty
350
E.L. Merz et al.
writing (Henry et al., 2010). Stanton et al. (2002) also reported that respondents found
EW to be a valuable/meaningful experience that provided them with positive, long-lasting
effects, and did not elicit negative effects.
Measurement of written content

Five studies (Craft et al. 2013; Gellaitry et al., 2010; Henry et al., 2010; Jensen-Johansen
et al., 2012; Rosenberg et al., 2002) did not evaluate written content. Of the eight studies
that used a method of text analysis, seven used written characteristics solely as a
manipulation check.3 Three studies used the LIWC program as a manipulation check,
finding that respondents in the experimental conditions used greater numbers of emotional and cognitive words compared to those in the control conditions [deMoor et al.,
2002 (32 categories evaluated, significant differences in 24 categories); deMoor
et al., 2008 (10 categories evaluated, significant differences in nine categories); Mosher
et al., 2012 (two categories evaluated, significant differences in both)], indicating
compliance with the writing instructions. One study (Walker et al., 1999) used the LIWC
program for a substantive question, finding that patients who were assigned to have three
doses of writing (vs. one dose) used more positive emotion words, and wrote more
about the present and future and less about the past in their third essay than their first
essay.
Trained judges were used to code written content as a manipulation check in four
studies [Low et al., 2010 (one judge, 94% accuracy); Mosher et al., 2012 (one judge, 99%
accuracy); Stanton et al., 2002 (one judge, 95% accuracy); Walker et al., 1999 (four
judges, accuracy not reported)] and as a process variable in one study (Cepeda et al.,
2008; 2 judges, k = .8, p = .0001). Across these studies, judge ratings suggested that
narratives from the experimental group were more personal and revealing compared to
narratives from the control conditions. The study by Cepeda et al. (2008), which was the
only study in the current review to evaluate whether the written content was related to
health outcomes, reported that individuals who wrote at least one essay with very much
disclosure reported less pain and greater well-being over eight weeks compared to those
who never had an essay rated as very much disclosure (disclosure was rated on a sixpoint scale ranging from none to very much). This suggests that exposure to ones
cancer-related emotions engendered the effectiveness of EW in this study. However, the
categorisation of emotional disclosure by dichotomising the blind judges ratings in this
study was unconventional; given that the authors reported that no other categorisations
had produced statistically significant results, these findings may represent a reporting
bias. Three studies (Mosher et al., 2012, Stanton et al., 2002, Zakowski et al., 2004) used
self-rating as a manipulation check; EW participants rated their essays as more personal
and revealing.
Risk of bias within studies
Table 2 provides a summary of risk of bias in the current review. Eight studies were
considered low risk of selection bias due to random sequence generation. Four studies
used a computerised random number generator, conferring the absolute lowest risk
(Cepeda et al., 2008; Gellaitry et al., 2010; Low et al., 2010; Stanton et al., 2002) and
four studies used restricted randomisation procedures that balance characteristics between
groups but can increase the predictability of group assignment by the investigator (e.g.,
minimisation, random permuted block; deMoor et al., 2002; deMoor et al., 2008; Jensen
351
Table 2. Risk of bias in studies reporting experimental trials of expressive writing interventions in
cancer patients.
Random
sequence
Allocation
generation concealment
Cepeda
et al. (2008)
Craft
et al. (2013)
deMoor
et al. (2002)
deMoor
et al. (2008)
Gellaitry
et al. (2010)
Henry
et al. (2010)
Jensen-Johansen
et al. (2012)
Low
et al. (2010)
Mosher
et al. (2012)
Rosenberg
et al. (2002)
Stanton
et al. (2002)
Walker
et al. (1999)
Zakowski
et al. (2004)
Blinding of
participants
and
personnel
Blinding
of
outcome
assessment
Incomplete
outcome
data
addressed
Selective
reporting
Note: low risk of bias: +, high risk of bias: , unclear risk of bias: ?
Johansen et al., 2012; Mosher et al., 2012). Two studies did not indicate the method of
random assignment (Rosenberg et al., 2002; Zakowski et al., 2004). The three studies
with the most significant risk used sequential assignment (Craft et al., 2013; Walker et al.,
1999) or matched controls (Henry et al., 2010).
Three studies were considered low risk of selection bias due to allocation
concealment (Cepeda et al., 2008; Low et al., 2010; Stanton et al., 2002) and three
studies were considered high risk (Craft et al., 2013; Henry et al., 2010; Walker et al.,
1999). The remaining studies did not provide sufficient details to draw a conclusion,
although Jensen-Johansen et al. (2012) did indicate that allocation was concealed prior to
group assignment without specifying the method.
As in any EW study, the risk of performance bias in the reviewed papers was high
overall. It is not possible to fully blind participants to condition assignment, as awareness
of the prompt is required to complete the task. Only one study described efforts to
352
E.L. Merz et al.
attenuate performance bias by blinding personnel to study purpose/hypotheses during the

intervention (Stanton et al., 2002).
Evidence of detection bias was largely unclear given that most studies used written
self-report measures for the primary study outcomes but did not provide details regarding
participant knowledge of the study purpose and hypotheses (n = 9). Three studies that
used written self-report were considered low risk because some efforts to obscure the
study purpose were described (Craft et al., 2013; Jensen-Johansen et al., 2012; Stanton
et al., 2002). One study was considered high risk because the nurse interviewers who
conducted the baseline assessment were not blind to allocation (Henry et al., 2010).
There was a relatively low attrition bias, with most studies reporting participant
retention rates in excess of 80%. Both samples with significant attrition comprised breast
cancer patients at different points in their treatment [i.e., definitive treatment had been
completed within the past two years (Craft et al., 2013); immediately prior to surgery
after completing neoadjuvant chemotherapy (deMoor et al., 2008)]. Notably, Craft et al.
(2013) used intent-to-treat analyses to attenuate the potential for misleading results from
their data. One study (Jensen-Johansen et al., 2012) did not describe the reasons for
attrition and was therefore considered an unclear risk.
Overall it appeared there was a low risk for reporting bias. However, Cepeda et al.
(2008) noted that while multiple categorisations of the emotional disclosure variable had
been tested, only the one that emerged as significant was presented; therefore, it is likely
that some amount of reporting bias was conferred. It is also worth mentioning that three
studies reported non-statistically significant trends (Craft et al., 2013, Low et al., 2010,
Rosenberg et al., 2002). While this does not formally constitute reporting bias, this can
create the appearance of greater significance than was evident in the data.
Discussion
This review summarised 13 experimental trials of EW in cancer. The samples were not
representative of the cancer population at large in terms of ethnicity or age, with an
overwhelming majority of participants being White and also younger than the overall
cancer population (median age of onset is 66; Horner et al., 2009). The younger age may
be due in part to the overrepresentation of breast cancer, which has a somewhat earlier
age of onset (median age of onset is 61years; Horner et al., 2009). Thus, the conclusions
drawn are limited in generalisability.
Findings from the reviewed studies were mixed, with some main effects reported for
sleep, pain, physical and mental health, and healthcare utilisation. The sizes of the
significant main effects varied widely, although the majority suggested a small to
moderate impact. This corroborates Frattarolis (2006) analysis of written and spoken
emotional disclosure, as well as meta-analyses by Frisina, Borod, and Lepore (2004) who
found an overall effect size of .084 (d = .19) in nine clinical samples, and Smyth (1998)
who found an overall effect size of .230 (d = .47) in 13 non-clinical samples. However, in
the present review there were few similarities in the designs of the six studies with
significant benefits. All but one (Henry et al., 2010) had participants complete four EW
sessions, although the time between sessions ranged widely. There was also great
variability in prompts [emotional (deMoor et al., 2002; Mosher et al., 2012; Stanton et al.,
2002; Rosenberg et al., 2002) vs. benefit-finding (Henry et al., 2010; Stanton et al., 2002)
vs. different prompts (Gellaitry et al., 2010)] and controls [usual care (Gellaitry et al.,
2010; Henry et al., 2010; Rosenberg et al., 2002) vs. neutral writing about trivial topics
353
(Mosher et al., 2012), health behaviours (deMoor et al., 2002) or cancer facts (Stanton
et al., 2002)], making it difficult to ascertain which aspects of EW are therapeutic, if any.
The implications of the demonstrated benefits of EW should also be considered
within the context of potential threats to the validity of these findings. Risk of biases
varied across the studies in the current review, although several trends were observed. In
general, there was little risk of biased findings due to selection, attrition or reporting bias.
Among the studies with significant main effects, Henry et al. 2010) conferred the greatest
amount of risk mainly because the matched groups design increased selection bias.
All EW studies carry a high risk of performance and detection biases. As part of the
informed consent process, experimenters warn participants about the risks and benefits of
the study and that they may be asked to write about their emotional experience with
cancer, which could be upsetting. Upon receiving instructions, participants become aware
of which condition they were assigned to, potentially creating different emotional and
behavioural reactions between the groups. One consequence of performance and
detection biases is that individuals in the experimental condition may experience (i.e.,
performance bias) or report (i.e., detection bias) improvements that are due to their
expectations about EW, rather than the actual EW task. The highest risk for this is in
studies where non-writing controls are used (e.g., usual care or attentional control; six
studies in the current review), because allocation is more obvious than in studies that
employ some type of neutral writing. Three of these non-writing control studies yielded
main effects (Gellaitry et al., 2010; Henry et al., 2010; Rosenberg et al., 2002), which is
noteworthy given that these represent half of the studies in the review with significant
results of experimental condition.
Although performance and detection biases cannot be totally eliminated given the
nature of the EW task, they can be attenuated by blinding study personnel to condition or
to purpose/hypotheses during all parts of the study that require participant contact. Some
studies reported efforts within the early stages of a participants protocol [e.g.,
interviewers in Rosenberg et al.s (2002) study were blind to condition before the
baseline assessment], during the intervention (Stanton et al., 2002) and during outcome
assessment (Cepeda et al., 2008; Mosher et al., 2012; Walker et al., 1999). However, it
should be noted that even in studies that did not report these efforts, contact between
researchers and participants was typically minimal after the instructions were
administered.
A number of methodological issues should also be considered when appraising the
quality of the evidence in the reviewed studies. Variability in the experimental and control
conditions may have impacted the findings. Most studies used a uniform EW prompt
across administrations, although Gellaitry et al., (2010) used different prompts across
sessions, and Jensen-Johansen et al. (2012) varied the instructions, threatening the
consistency (and thus the internal validity) of the study. Emotional, benefit-finding and
multiple-prompt writing instructions all elicited significant and null main effects across a
myriad of outcomes. Control conditions also ranged from usual care, to receiving
attention, to neutral writing about non-cancer topics, to cancer facts, which more
stringently controls for mere exposure to cancer-related content. Thus, the mixed results
produced by the included studies may not be due entirely to EW being ineffective in this
population, but rather an artefact of comparing apples with oranges. The specific features
of EW that are effective in cancer populations are clearly not well understood.
It is interesting that most EW protocols instructed participants to write expressively
about their cancer. Although it is reasonable to assume that a cancer diagnosis is the most
354
E.L. Merz et al.
central traumatic issue in ones life, this is not necessarily the case. For example, the
study by Jensen-Johansen et al. (2012), which allowed participants to self-modify the
expressive writing topic, found no main effects for EW. However, chosen topic (cancer
vs. other) moderated the efficacy of EW, with each group reporting different benefits
(Jensen-Johansen et al., 2012). This raises an interesting point: in healthy populations, for
which EW effects have been greater, participants can write about any trauma; in cancer,
researchers have generally tailored the prompt to be cancer specific. In doing so, the
possibility that some patients may be even more distressed by other issues in their lives
may have been overlooked. In allowing for a choice, Jensen-Johansen et al. (2012)
provided the opportunity for participants to tailor the intervention, leading to improved
outcomes for both topics. Such flexibility in clinical settings has also been generally
recommended by Pennebaker (2010) who wrote, Encourage people to write about what
is bothering them, rather than what you think is bothering them (p. 24). Interestingly,
Craft et al. (2013) also included a condition wherein patients were instructed to write
about any trauma (the nature of the chosen topics was not reported; therefore, it is
unknown whether any of these patients wrote about cancer) in addition to the cancer
trauma EW group. All of the writing groups in Crafts study (i.e., emotional writing about
cancer, emotional writing about any trauma, neutral writing about cancer) had an increase
in quality of life from baseline to the first follow-up, whereas the usual care group did
not. However, when all four groups were compared against one another, the only
significant group difference that emerged was between those who wrote neutrally about
cancer and those who received usual care and therefore did not write anything.
Threats to internal validity due to lower control in clinical studies are also relevant.
Given the challenges of recruiting and retaining medical patients, location of settings
varied (e.g., home and waiting rooms) and participants often completed writing sessions
over periods that were longer than the original recommendation of four consecutive days.
The extent to which these differences influenced outcomes is unclear; however, it is
important to recognise that the many differences between the clinic and highly controlled
laboratory settings may be important.
Differences in the timing of follow-up assessments should also be considered. In the
study by deMoor et al. (2002), the first follow-up assessment was conducted immediately
following the intervention, which may have affected the pooled estimates of the
outcomes. People typically feel somewhat worse after EW (Pennebaker & Beall, 1986;
Smyth, 1998); thus, the observed effects for EW on distress, stress and mood may have
been diminished in this study. In addition, Cepeda et al. (2008) conducted the first three
assessments (out of eight) during the same weeks that patients engaged in EW. Although
no main effects for EW on pain or quality of life emerged even by the eighth week of the
study, it is possible that the immediacy of the first several follow-ups contaminated the
outcome measurements.
Another possibility is that the EW protocols (whether one session of writing or
multiple sessions) were considered single interventions. Health-promoting therapies are
not generally administered once and then expected to provide benefits into the future
without repetition. Perhaps EW should be considered a repeatable exercise, or part of a
larger therapeutic programme that includes other components such as psychoeducation
and group therapy, as was recently demonstrated in patients with colorectal cancer
(Carmack et al., 2011).
A wide array of dependent variables was evaluated in the included studies, increasing
the possibility of chance findings. Also, p-value corrections were not typically applied,
355
including in studies that evaluated a variety of outcomes from the same domain [e.g.,
deMoor et al. (2002) evaluated 19 outcomes of stress, mood and sleep; and Craft et al.
(2013) analysed the same quality of life variable in several ways with alphas set at .05].
The reviewed studies included outcomes ranging from affect and cognitions, to sleep and
pain, to disease markers and immune functioning. It seems that a kitchen sink of
outcomes has been evaluated, rather than identifying key areas of change expected from
EW in the context of cancer.
Alternatively, the effectiveness of EW could be dependent upon individual differences
(e.g., social constraints, alexithymia, avoidance, emotional support and time since
diagnosis) and thus the effects of the intervention may not be detectable until these
features are partialed out. Several studies evaluated potential moderators to this end;
however, more research on moderators is warranted as these findings were derived from
single studies, except those regarding social constraints.
Another potential explanation is that the effect size for EW is simply smaller for
clinical samples as compared to healthy samples (Frisina et al., 2004). Perhaps clear
reductions in pathology are more than should be expected from EW in the context of a
life-altering and potentially life-ending disease. Even when EW did not elicit changes in
the presence or absence of symptoms, participant feedback was overwhelmingly positive
and patients generally reported appreciating the exercise. Researchers and clinicians may
want to consider the possibility that EW provides benefits that are more existential in
nature. Notably, one of the reviewed studies (Mosher et al., 2012) investigated the
possibility of existential gains using standardised self-report tools, without significant
findings. Nevertheless, there is merit in continuing to explore this possibility.
Use of written content
This review found that text analysis is used primarily as a manipulation check and is
greatly underused as a process variable within experimental trials of cancer patients. This
is unfortunate given that linguistic information could enhance theoretical understanding
of why EW works and for whom, within this population. Such information could also
serve as an aid in decision-making with regard to intervention design (e.g., how many
sessions to prescribe). For example, Walker et al. (1999) evaluated written content in their
substantive analyses, finding that emotional disclosure and present/future orientation
words increased from the first essay to the third. The authors concluded that the practical
implication of this finding is that greater doses of EW in clinical settings encourage
more disclosure, although notably this did not elicit differences on the studied outcomes
(Walker et al., 1999). The only study from the current review to investigate linguistic
characteristics as a process variable found that greater disclosure (i.e., exposure to ones
emotions) predicted reduced pain, suggesting that disclosure may be the key mechanism
through which EW is effective (Cepeda et al., 2008). However, caution regarding this
finding is warranted given the evidence for selective reporting bias in this study.
Other studies have analysed text in cancer studies with promising results. In a reanalysis of Stanton et al.s (2002) data, a greater number of negative emotion words was
associated with fewer somatic symptoms, suggesting that the positive effects of EW are
contingent on habituation to an emotional reaction to cancer through the writing process
(Low et al., 2006). That is, it may be necessary to expose the writer to their emotions
related to cancer, rather than the facts of the event itself, for EW to be effective (Low
et al., 2006). In a correlational study of women with metastatic breast cancer, Laccetti
356
E.L. Merz et al.
(2007) found that more positive words were associated with better emotional well-being.
Smith et al. (2005) also found that greater negative emotion words predicted higher
anxiety and depression after adjusting for baseline mood in a prepost study of newly
diagnosed breast cancer patients. In contrast, Morgan, Graves, Pogi, and Cheson (2008)
found that while greater use of affect, positive emotion and anxious words correlated with
better physical health-related quality of life in leukaemia and lymphoma patients, and
using more sad and anxious words correlated with poorer mental health-related quality of
life, the relationship was explained by pre-intervention quality of life.
Limitations of the review

There were several limitations to the current review. The search was limited to
publications in English, thus excluding at least one relevant study [e.g., Park and Yi
(2012) is published in Korean]. Further, because grey literature was excluded, the current
review may not be fully inclusive of all research conducted on EW in cancer. The review
was unable to account for publication bias and the potential of incomplete reporting of all
null findings. In addition, a pooled quantitative analysis of the findings was not
performed given the diversity of study conditions and outcomes.
Future research
The extant literature limits the ability to draw definitive conclusions about EW as an
effective intervention for cancer patients. Given the heterogeneity of studies and results,
further investigation is warranted, with specific recommendations. Recruitment of
patients with other common cancers, men and ethnically diverse samples is encouraged
to increase generalisability. It is also recommended that researchers adhere to Pennebaker
and Bealls (1986) original design of using four writing administrations to enhance
standardisation within cancer samples and comparability to other EW literature, unless
the question being asked warrants a variation (e.g., dosing). As much as possible,
researchers are encouraged to blind personnel to study condition and study hypotheses, to
mask the purpose of the study to participants, and to articulate these efforts in their final
publications.
Researchers should build upon the strongest studies in the current review and design
protocols with respect to both basic and clinical research questions specific to cancer.
Tasks should be designed to systematically explicate the components of writing that are
potentially efficacious (e.g., neutral writing about anything vs. neutral writing about
cancer vs. emotional writing about cancer vs. benefit-finding about cancer vs. emotional
writing about anything), and also explore the advantages and disadvantages of these
different prompts. Text analysis should also be utilised to test and refine the explanatory
models of the mechanisms underlying EW-based changes in cancer patients.
Moderators examined in the current review also need to be replicated, and possible
interactions with gender (e.g., Range & Jenkins, 2010), personality (e.g., Zakowski et al.,
2011) and other individual differences (e.g., type of cancer diagnosis, stage, treatment,
age and education) should also be considered. It is also worth exploring whether
individual differences may be used to maximise the effectiveness of EW interventions by
matching individuals with prompts. For example, Stanton et al. (2002) found that low
avoidance individuals benefit from the emotional prompt, whereas high avoidance
individuals benefit from the benefit-finding prompt. Multiplicative interactions may also
be important to consider given Zakowski et al.s (2003) finding that male cancer patients
357
who are inhibited from expressing their emotions to their spouses suffer more problems in
mood and intrusive thoughts than female cancer patients with similar levels of inhibition.
This recommendation has both theoretical and practical advantages because further
analysis of moderators may enhance understanding of the fundamental bases of how EW
works within the context of cancer, and also may inform the identification of individuals
in cancer clinics who may benefit from or be injured by EW.
More practical questions should focus on feasibility and effectiveness in clinical
settings. Great efforts were taken in the current studies to encourage task compliance, but
clinical researchers may want to design studies wherein these contingencies are not in
place to determine how viable EW is in real-world settings. It is also notable that there
was significant attrition among the pre-surgical patients in the study by deMoor et al.
(2008), whereas there was less attrition in studies that enrolled patients at other points in
their cancer treatment. At least one other published study has also reported completion
issues, with only two of the 24 participants in palliative care for advanced cancer
completing all four sessions (Bruera, Willey, Cohen, & Palmer, 2008). It is possible that
writing during a trauma or immediately afterwards could cause more harm than good
(Pennebaker, 2010), potentially explaining this high level of dropout. Thus, it may be
worthwhile to determine whether there is an optimal time or condition for EW
interventions.
In addition, while symptomatology relevant to cancer patients (e.g., physical health,
affect, mood and fatigue) should continue to be assessed in EW studies, qualitative and
descriptive outcome assessments should also be considered. Pennebaker and Beall (1986)
suggested that EW improves physical health (e.g., number of illnesses and number of
days one is restricted due to illness), which may be more relevant in healthy populations.
For cancer patients, EW may have a less obvious effect on health; rather, it may provide
comfort in other ways, such as with existential concerns, personal growth or meaningmaking.
Finally, it is also recommended that researchers expand their use of text analysis. In
particular, LIWC and trained judges (whenever possible, given time and cost constraints)
are recommended. The LIWC program appears to have many potential benefits, including
demonstrated sensitivity, convergent validity and discriminant validity for identifying
emotional expression in cancer patients (Bantum & Owen, 2009). It has also been used to
demonstrate how cognitive and affective word choice in EW passages predicts quality of
life outcomes in cancer studies that did not meet the current reviews inclusion criteria.
Conclusions
Taken together, EW appears to be generally feasible for cancer patients, and there were
some positive outcomes. Given the current findings, a reliable recommendation cannot be
made, and healthcare providers should recognise that the evidence is not clear about the
therapeutic value of EW in cancer. There were generally no adverse consequences, with
the exception of increased sleep disturbance among patients who had been living with
breast cancer for more than 4.7 years (Low et al., 2010) and increased sleep medication
usage in breast cancer patients three days prior to surgery (although it is unknown what
the mechanism for the uptake of these hypnotics was; deMoor et al., 2008). This is
consistent with other EW studies, suggesting that EW is likely safe to use with this
population, even if the measured gains are not remarkable (Baikie & Wilhelm, 2005).
Many participants reported enjoying the intervention, and thus, for cancer patients who
358
E.L. Merz et al.
are interested in EW as a complementary health promotion/coping strategy, EW could

represent a simple, accessible, inexpensive exercise that may offer some relief.
Acknowledgements
We are grateful to Linda C. Gallo, PhD for helpful comments on an earlier version of this paper. We
thank Scott C. Roesch, PhD and Sharon Danoff-Burg, PhD for providing additional guidance and
clarification. We thank three anonymous reviewers for their constructive comments which
contributed to improving the final version of this paper.
Supplemental material
Supplemental material for this article can be accessed here: http://dx.doi.org/10.1080/17437199.201

4.882007
Notes
1. Although including grey literature may attenuate publication bias, given that published studies
generally have larger treatment effects than unpublished studies (Hopewell, McDonald, Clarke,
& Egger, 2007), grey literature can also introduce bias due to poorer methodological quality
(Egger, Juni, Barlett, Holenstein, & Sterne, 2003) and comprehensiveness of locatable grey
literature (Sterne, Egger, & Moher, 2011). Given that reviews that include grey literature produce
similar results to those that do not (Egger et al., 2003), it was decided to restrict the search to
published findings.
2. The authors described the neutral writing condition as an experimental group; however, for
consistency across studies it was categorized as a control condition for the purpose of this
review.
3. Jensen-Johansen and colleagues (2012) performed their manipulation check by evaluating
participants mood immediately following the EW intervention, rather than assessing the
linguistic characteristics of the essays.
References
Andersson, M. A., & Conley, C. S. (2008). Expecting to heal through self-expression: A perceived
control theory of writing and health. Health Psychology Review, 2, 138162. doi:10.1080/
17437190802660890
Baikie, K. A., & Wilhelm, K. (2005). Emotional and physical health benefits of expressive writing.
Advances in Psychiatric Treatment, 11, 338346. doi:10.1192/apt.11.5.338
Bantum, E. O., & Owen, J. E. (2009). Evaluating the validity of computerized content analysis
programs for identification of emotional expression in cancer narratives. Psychological
Assessment, 21, 7988. doi:10.1037/a0014643
Broderick, J. E., Stone, A. A., Smyth, J. M., & Kaell, A. T. (2004). The feasibility and effectiveness
of an expressive writing intervention for rheumatoid arthritis via home-based videotaped
instructions. Annals of Behavioral Medicine, 27(1), 5059. doi:10.1207/s15324796abm2701_7
Bruera, E., Willey, J., Cohen, M., & Palmer, J. L. (2008). Expressive writing in patients receiving
palliative care: A feasibility study. Journal of Palliative Medicine, 11(1), 1519. doi:10.1089/
jpm.2007.0112
Carmack, C. L., Basen-Engquist, K., Yuan, Y., Greisinger, A., Rodriguez-Bigas, M., Wolff, R. A.,
Pennebaker, J. W. (2011). Feasibility of an expressive-disclosure group intervention for posttreatment colorectal cancer patients: Results of the healthy expressions study. Cancer, 117, 4993
5002. doi:10.1002/cncr.26110
Cepeda, M. S., Chapman, C. R., Miranda, N., Sanchez, R., Rodriguez, C. H., Restrepo, A. E.,
Carr, D. B. (2008). Emotional disclosure through patient narrative may improve pain and wellbeing: Results of a randomized controlled trial in patients with cancer pain. Journal of Pain and
Symptom Management, 35, 623631. doi:10.1016/j.jpainsymman.2007.08.011
Craft, M. A., Davis, G. C., & Paulson, R. M. (2013). Expressive writing in early breast cancer
survivors. Journal of Advanced Nursing, 69, 305315. doi:10.1111/j.1365-2648.2012.06008.x
359
Creswell, J. D., Lam, S., Stanton, A. L., Taylor, S. E., Bower, J. E., & Sherman, D. K. (2007). Does
self-affirmation, cognitive processing, or discovery of meaning explain cancer-related health
benefits of expressive writing? Personality and Social Psychology Bulletin, 33, 238250.
doi:10.1177/0146167206294412
Danoff-Burg, S., Mosher, C. E., Seawell, A. H., & Agee, J. D. (2010). Does narrative writing
instruction enhance the benefits of expressive writing? Anxiety, Stress, and Coping, 23, 341352.
doi:10.1080/10615800903191137
deMoor, C., Moy, L., Low, D., Rivera, E., Singletary, S. E., Fouladi, R. T., & Cohen, L. (2008).
Expressive writing as a presurgical stress management intervention for breast cancer patients.
Journal of the Society for Integrative Oncology, 6, 5966.
deMoor C., Sterner J., Hall M.Warneke, C., Gilani, Z., Amato, R., & Cohen, L. (2002). A pilot
study of the effects of expressive writing on psychological and behavioral adjustment in patients
enrolled in a Phase II trial of vaccine therapy for metastatic renal cell carcinoma. Health
Psychology, 21, 615619. doi:10.1037//0278-6133.21.6.615
Egger, M., Juni, P., Bartlett, C., Holenstein, F., & Sterne, J. (2003). How important are
comprehensive literature searches and the assessment of trial quality in systematic reviews?
Empirical study. Health Technology Assessment, 7, 176.
Frank, A. W. (1991). At the will of the body: Reflections on illness. Boston, MA: Houghton Mifflin.
Frattaroli, J. (2006). Experimental disclosure and its moderators: A meta-analysis. Psychological
Bulletin, 132, 823865. doi:10.1037/0033-2909.132.6.823
Frisina, P. G., Borod, I. C., & Lepore, S. J. (2004). A meta-analysis of the effects of written
emotional disclosure on the health outcomes of clinical populations. Journal of Nervous and
Mental Disease, 192, 629634. doi:10.1097/01.nmd.0000138317.30764.63
Gellaitry, G., Peters, K., Bloomfield, D., & Home, R. (2010). Narrowing the gap: The effects of an
expressive writing intervention on perceptions of actual and ideal emotional support in women
who have completed treatment for early stage breast cancer. Psycho-Oncology, 19, 7784.
doi:10.1002/pon.1532
Harris, A. H. S., Thoresen, C. E., Humphreys, K., & Faul, J. (2005). Does writing affect asthma? A
randomized trial. Psychosomatic Medicine, 67, 130136. doi:10.1097/01.psy.0000146345.73
510.d5
Hart, S. L., Hoyt, M. A., Diefenbach, M., Anderson, D. R., Kilbourn, K. M., Craft, L. L.,
Stanton, A. L. (2012). Meta-analysis of efficacy of interventions for elevated depressive
symptoms in adults diagnosed with cancer. Journal of the National Cancer Institute, 104, 990
1004. doi:10.1093/jnci/djs256
Henry, E. A., Schlegel, R. J., Talley, A. E., Molix, L. A., & Bettencourt, B. A. (2010). The
feasibility and effectiveness of expressive writing for rural and urban breast cancer survivors.
Oncology Nursing Forum, 37, 749757. doi:10.1188/10.ONF.749-757
Hoey, L. M., Ieropoli, S. C., White, V. M., & Jefford, M. (2008). Systematic review of peer-support
programs for people with cancer. Patient Education and Counseling, 70, 315337. doi:10.1016/j.
pec.2007.11.016
Hopewell, S., McDonald, S., Clarke, M. J., & Egger, M. (2007). Grey literature in meta-analyses of
randomized trials of health care interventions (Review). Cochrane Database of Systematic
Reviews, 2, MR000010. doi:10.1002/14651858.MR000010.pub3
Horner, M. J., Ries, L. A. G., Krapcho, M., Neyman, N., Aminou, R., Howlander, N., Edwards,
B. K. (Eds.). (2009). SEER cancer statistics review, 19752006. Bethesda, MD: National Cancer
Institute. Retrieved from http://seer.cancer.gov/csr/1975_2006
Hybye, M. T., Dalton, S. O., Deltour, I., Bidstrup, P. E., Frederiksen, K., & Johansen, C. (2010).
Effect of Internet peer-support groups on psychosocial adjustment to cancer: A randomised study.
British Journal of Cancer, 102, 13481354. doi:10.1038/sj.bjc.6605646
Jensen-Johansen, M. B., Christensen, S., Valdimarsdottir, H., Zakowski, S., Jensen, A. B., &
Bovbjerg, D. H. (2012). Effects of an expressive writing intervention on cancer-related distress in
Danish breast cancer survivors results from a nationwide randomized clinical trial. PsychoOncology, 22, 14921500. doi:10.1002/pon.3193
Kallay, E., & Baban, A. (2008). Emotional benefits of expressive writing in a sample of Romanian
female cancer patients. Cognition, Brain, Behavior, 12, 115129.
Laccetti, M. (2007). Expressive writing in women with advanced breast cancer. Oncology Nursing
Forum, 34, 10191024. doi:10.1188/07.ONF.1019-1024
360
E.L. Merz et al.
Lepore, S. J., Greenberg, M. A., Bruno, M., & Smyth, J. M. (2002). Expressive writing and health:
Self-regulation of emotion-related experience, physiology, and behavior. In S. J. Lepore & J. M.
Smyth (Eds.), The writing cure: How expressive writing promotes health and emotional well
being (pp. 99117). Washington, DC: American Psychological Association.
Low, C. A., Stanton, A. L., Bower, J. E., & Gyllenhammer, L. (2010). A randomized controlled trial
of emotionally expressive writing for women with metastatic breast cancer. Health Psychology,
29, 460466. doi:10.1037/a0020153
Low, C. A., Stanton, A. L., & Danoff-Burg, S. (2006). Expressive disclosure and benefit finding
among breast cancer patients: Mechanisms for positive health effects. Health Psychology, 25,
181189. doi:10.1037/0278-6133.25.2.181
Lu, Q., Zheng, D., Young, L., Kagawa-Singer, M., & Loh, A. (2012). A pilot study of expressive
writing interventions among Chinese-speaking breast cancer survivors. Health Psychology, 31,
548551. doi:10.1037/a0026834
Manzoni, G. M., Castelnuovo, G., & Molinari, E. (2011). The written-heart study (expressive
writing for heart healing): Rationale and design of a randomized controlled clinical trial of
expressive writing in coronary patients referred to residential cardiac rehabilitation. Health and
Quality of Life Outcomes, 9, 51. doi:10.1186/1477-7525-9-51
Morgan, N. P., Graves, K. D., Poggi, E. A., & Cheson, B. D. (2008). Implementing an expressive
writing study in a cancer clinic. The Oncologist, 13, 196204. doi:10.1634/theoncologist.
2007-0147
Mosher, C. E., DuHamel, K. N., Lam, J., Dickler, M., Li, Y., Massie, M. J., & Norton, L. (2012).
Randomised trial of expressive writing for distressed metastatic breast cancer patients,
Psychology & Health, 27, 88100. doi:10.1080/08870446.2010.551212
Owen, J. E., Giese-Davis, J., Cordova, M., Kronenwetter, C., Golant, M., & Spiegel, D. (2006).
Self-report and linguistic indicators of emotional expression in narratives as predictors of
adjustment to cancer. Journal of Behavioral Medicine, 29, 335345. doi:10.1007/s10865-0069061-8
Park, E. Y., & Yi, M. (2012). Development and effectiveness of expressive writing program for
women with breast cancer in Korea [Abstract]. Journal of the Korean Academy of Nursing, 42,
269279. doi:10.4040/jkan.2012.42.2.269
Pennebaker, J. W. (1993). Putting stress into words: Health, linguistic and therapeutic implications.
Behavior Research and Therapy, 31, 539548. doi:10.1016/0005-7967(93)90105-4
Pennebaker, J. W. (2010). Expressive writing in a clinical setting. Independent Practitioner, 30,
2325.
Pennebaker, J. W., & Beall, S. K. (1986). Confronting a traumatic event: Toward an understanding
of inhibition and disease. Journal of Abnormal Psychology, 95, 274281. doi:10.1037/0021843X.95.3.274
Pennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). Linguistic inquiry and word count: LIWC
[Computer software]. Austin, TX: LIWC.net.
Pennebaker, J. W., & Chung, C. K. (2007). Expressive writing, emotional upheavals, and health. In
H. S. Friedman & R. C. Silver (Eds.), Foundations of health psychology (pp. 263284). New
York, NY: Oxford University.
Pennebaker, J. W., & Chung, C. K. (2011). Expressive writing and its links to mental and physical
health. In H. S. Friedman (Ed.), Oxford handbook of health psychology (pp. 417437). New
York, NY: Oxford University Press.
Pennebaker, J. W., Mayne, T. J., & Francis, M. E. (1997). Linguistic predictors of adaptive
bereavement. Journal of Personality and Social Psychology, 72, 166183. doi:10.1037/00223514.72.4.863
Range, L. M., & Jenkins, S. R. (2010). Who benefits from Pennebakers expressive writing
paradigm? Research recommendations from three gender theories. Sex Roles, 63, 149164.
doi:10.1007/s11199-010-9749-7
Rivkin, I. D., Gustafson, J., Weingarten, I., & Chin, D. (2006). The effects of expressive writing on
adjustment to HIV. AIDS and Behavior, 10(1), 1326. doi:10.1007/s10461-005-9051-9
Rosenberg, H. J., Rosenberg, S. D., Ernstoff, M. S., Wolford, G. L., Amdur, R. J., Elshamy, M. R.,
Pennebaker, J. W. (2002). Expressive disclosure and health outcomes in a prostate cancer
population. International Journal of Psychiatry in Medicine, 32, 3753. doi:10.2190/AGPFVB1G-U82E-AE8C
361
Servaes, P., Vingerhoets, A. J. J. M., Vreugdenhil, G., Keuning, J. J., & Broekhuijsen, A. M.
(1999). Inhibition of emotional expression in breast cancer patients. Behavioral Medicine, 25,
2327. doi:10.1080/08964289909596735
Smith, S., Anderson-Hanley, C., Langrock, A., & Compas, B. (2005). The effects of journaling for
women with newly diagnosed breast cancer. Psycho-Oncology, 14, 10751082. doi:10.1002/
pon.912
Smyth, J. M. (1998). Written emotional expression: Effect sizes, outcome types, and moderating
variables. Journal of Consulting and Clinical Psychology, 66, 174184. doi:10.1037/0022006X.66.1.174
Smyth, J., True, N., & Souto, J. (2001). Effects of writing about traumatic experiences: The
necessity for narrative structuring. Journal of Social and Clinical Psychology, 20, 161172.
doi:10.1521/jscp.20.2.161.22266
Stanton, A. L., Danoff-Burg, S., Sworowski, L. A., Collins, C. A., Branstetter, A. D., RodriguezHanley, A., Austenfeld, J. L. (2002). Randomized, controlled trial of written emotional
expression and benefit finding in breast cancer. Journal of Clinical Oncology, 20, 41604168.
doi:10.1200/JCO.2002.08.521
Sterne, J. A. C., Egger, M., & Moher, D. (Eds.). (2011). Chapter 10.3.2: Including unpublished
studies in systematic reviews. In J. P. T. Higgins and S. Green (Eds.), Cochrane handbook for
systematic reviews of interventions, Version 5.1.0. Retrieved from http://handbook.cochrane.org/
Tamagawa, R., Giese-Davis, J., Speca, M., Doll, R., Stephen, J., & Carlson, L. E. (2013). Trait
mindfulness, repression, suppression, and self-reported mood and stress symptoms among
women with breast cancer. Journal of Clinical Psychology, 69, 264277. doi:10.1002/jclp.21939
Tausczik, Y. R., & Pennebaker, J. W. (2010). The psychological meaning of words: LIWC and
computerized text analysis methods. Journal of Language and Social Psychology, 29(1), 2454.
doi:10.1177/0261927X09351676
Walker, B. L., Nail, L. M., & Croyle, R. T. (1999). Does emotional expression make a difference in
reaction to breast cancer? Oncology Nursing Forum, 26, 10251032.
Zakowski, S. G., Harris, C., Krueger, N., Laubmeier, K. K., Garrett, S., Flanigan, R., & Johnson, P.
(2003). Social barriers to emotional expression and their relations to distress in male and female
cancer patients. British Journal of Health Psychology, 8, 271286. doi:10.1348/135910703322
370851
Zakowski, S. G., Herzer, M., Barrett, S. D., Milligan, J. G., & Beckman, N. (2011). Who benefits
from emotional expression? An examination of personality differences among gynaecological
cancer patients participating in a randomized controlled emotional disclosure intervention trial.
British Journal of Psychology, 12, 355372. doi:10.1348/000712610X524949
Zakowski, S. G., Ramati, A., Morton, C., Johnson, P., & Flanigan, R. (2004). Written emotional
disclosure buffers the effects of social constraints on distress among cancer patients. Health
Psychology, 23, 555563. doi:10.1037/0278-6133.23.6.555

Health Psychology Review: Click For Updates

Uploaded by

Copyright:

Available Formats

Health Psychology Review: Click For Updates

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Health Psychology Review: Click For Updates

Uploaded by

Copyright:

Available Formats

This article was downloaded by: [New York University]

On: 24 April 2015, At: 08:24

Health Psychology Review

Expressive writing interventions in

Erin L. Merz , Rina S. Fox & Vanessa L. Malcarne

SDSU/UCSD Joint Doctoral Program in Clinical Psychology, 6363

Click for updates

PLEASE SCROLL DOWN FOR ARTICLE

Downloaded by [New York University] at 08:24 24 April 2015

Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

Health Psychology Review, 2014

Expressive writing interventions in cancer patients:

Downloaded by [New York University] at 08:24 24 April 2015

(Received 16 May 2013; accepted 8 January 2014)

*Corresponding author. Email: [email protected]

E.L. Merz et al.

Downloaded by [New York University] at 08:24 24 April 2015

Health Psychology Review

Downloaded by [New York University] at 08:24 24 April 2015

Expressive writing and cancer

E.L. Merz et al.

Downloaded by [New York University] at 08:24 24 April 2015

OR illness narrative OR reflective writing OR writing cure) and cancer (cancer OR

Downloaded by [New York University] at 08:24 24 April 2015

Health Psychology Review

(i.e., performance bias) is problematic in all EW studies because a true double-blind

Task and follow-up

Results; (Cohens d) for main effects

Cepeda et al. (2008)

EW: STORY (n = 65)

Craft et al. (2013)

EW: EMO (n =26)

deMoor et al. (2002)

EW: EMO (n = 21)

deMoor et al. (2008)

EW: EMO (n = 24)

E.L. Merz et al.

Downloaded by [New York University] at 08:24 24 April 2015

Study and sample

Results; (Cohens d) for main effects

EW: MULTIPLE (n = 38)

Henry et al. (2010)

EW: BEN (n = 40)

At 3 and 9 months, EW did not directly improve somatisation or retarded activity

EW: EMO (n = 253)

Health Psychology Review

Downloaded by [New York University] at 08:24 24 April 2015

Study and sample

Task and follow-up

Results; (Cohens d) for main effects

Low et al. (2010)

EW: EMO (n = 31)

At 3 months, EW did not directly improve depression (d = .05), intrusive thoughts (d =

Mosher et al. (2012)

EW: EMO (n = 44)

At 8 weeks, EW did not directly improve existential well-being (meaning/peace d = .25,

EW: EMO (n = 16)

At 6 months, EW did not directly improve immune-directed cancer activity markers

Stanton et al. (2002)