Expressive Writing as a Therapeutic Process for Drug Dependent Women

Abstract

Background

Although women with Substance Use Disorders (SUD) have high rates of trauma and post-traumatic stress, many addiction programs do not offer trauma-specific treatments. One promising intervention is Pennebaker’s expressive writing, which involves daily, 20-minute writing sessions to facilitate disclosure of stressful experiences.

Methods

Women (N = 149) in residential treatment completed a randomized clinical trial comparing expressive writing to control writing. Repeated measures analysis of variance was used to document change in psychological and physical distress from baseline to 2-week and 1-month follow-ups. Analyses also examined immediate levels of negative affect following expressive writing.

Results

Expressive writing participants showed greater reductions in post-traumatic symptom severity, depression, and anxiety scores, when compared to control writing participants at the 2-week follow-up. No group differences were found at the 1-month follow-up. Safety data were encouraging; while expressive writing participants showed increased negative affect immediately after each writing session, there were no differences in pre-writing negative affect scores between conditions the following day. By the final writing session, participants were able to write about traumatic/stressful events without having a spike in negative affect.

Conclusions

Results suggest expressive writing may be a brief, safe, low cost, adjunct to SUD treatment that warrants further study as a strategy for addressing post-traumatic distress in substance-abusing women.

INTRODUCTION

Studies point to overwhelming rates of trauma in the lives of women with Substance Use Disorders (SUDs). In SUD treatment settings, 55% to 99% of women report at least one lifetime trauma (1). As many as 80% of women seeking drug abuse treatment report lifetime histories of physical or sexual assault (2), and many display Posttraumatic Stress Disorder (PTSD) symptomatology (3). Trauma and addiction comorbidity studies reveal these women are likely to have poorer health, increased disability, more severe clinical profiles, and poorer treatment adherence, than those without trauma or PTSD (4).

While trauma interventions have traditionally not been incorporated into SUD treatment programs, studies indicate that addressing trauma during SUD treatment may improve treatment outcomes (5). Historically, there has been an ongoing debate regarding which problem to treat first; the trauma or the SUD. Among SUD treatment providers, the general consensus was that addressing trauma during the early phase of SUD recovery may “open Pandora’s box” and derail SUD treatment improvement (5). Recent research, however, suggests the opposite; that trauma should be treated concurrently, even in the earliest stages of SUD treatment (6).

Accumulating evidence over the past two decades indicates that disclosure of traumatic or stressful experiences through Pennebaker’s expressive writing has widespread benefits. The typical laboratory research for Pennebaker’s expressive writing involves randomly assigning individuals to one of two conditions (e.g., expressive writing [emotional disclosure] versus control writing [neutral topic]). Expressive writing involves writing about the most traumatic or stressful event of one’s life for 15 to 20 minutes a day over three to five consecutive days, and is typically done with no feedback given (7). Studies have demonstrated the efficacy of expressive writing as a brief therapeutic intervention, with significant reductions in distress (8), and improved psychological (9) and physical health (10).

For instance, Pennebaker’s expressive writing has revealed that writing about a stressful or traumatic event is associated with a reduction in the degree to which the experience is distressing or painful (11), reduced depression among depressed women (12), a decrease in health center and doctor’s visits (10; 13), fewer missed days of work or school (14), improved immune functioning (15), and decreases in sympathetic nervous system activity (16).

In a meta-analysis of expressive writing studies, Smyth (17) found a 23% symptom improvement in the traumatic disclosure writing condition over a control writing condition. The average effect size was comparable or larger than those found in other psychological, educational, or behavioral interventions (8). Cognitive Processing Therapy (CPT), which also employs writing about stressful events, has recently revealed decreases in PTSD and depression symptoms (18). Furthermore, research indicates no harm to participants has been encountered as a result of Pennebaker’s expressive writing. While there is a rise in negative affect immediately after trauma writing, it is usually transient; and at follow-up, participants generally report that they feel better than they did before writing (19).

Although much attention has focused on benefits of Pennebaker’s expressive writing, to date, it has not been examined in populations affected by SUDs. Expressive writings’ efficiency (15 to 20 minutes over three to five days), efficacy, and low cost suggest that writing as a means for disclosing traumatic experiences may be a powerful adjunct to traditional SUD treatment.

The purpose of this study was to determine whether Pennebaker’s expressive writing is beneficial as a brief adjunct to traditional treatment for women currently undergoing residential treatment for SUDs. This randomized clinical trial (RCT) tests two hypotheses. First, it was hypothesized that participants randomized to the expressive writing condition would show more improvement in psychological and physical health at the 2-week and 1-month follow-up, compared to control writing participants. While it was expected that there would be improvements for both writing conditions as a function of being in residential treatment, expressive writing participants were expected to show greater improvement. Second, it was hypothesized that negative affect measured immediately after each writing session would be higher among expressive writing participants compared with control writing participants, but that these increases in negative affect would be short-lived.

METHODS

Participants

Participants were 149 women admitted to a gender-specific residential SUD treatment facility from June 17, 2007 to November 6, 2008. Specific services within this facility include individual counseling, motivational enhancement therapy groups, and case management for such needs as housing, transportation, and childcare. To be eligible for the study, women had to: (a) be at least 18 years old; (b) meet DSM-IV criteria for a SUD; and (c) have approval for 60 days of residential treatment from a third-party payer to help facilitate presence for the 1-month follow-up. Women were ineligible if they: (a) had an acute mental disorder (e.g., current suicidality) that would make it difficult to provide informed consent and/or follow the study protocol; or (b) had literacy problems that would prevent them from being able to complete the writing sessions or the research assessments.

Measures

With the exception of demographic questions and the Structured Clinical Interview for DSM-IV-TR - Alcohol and Substance Use Disorders Module (SCID) (20), measured at baseline, the same battery of measures was administered at baseline, and at the 2-week and 1-month post-writing follow-up. All assessments were completed on-site and instructions were read aloud by the researcher. The primary outcome measure included the Posttraumatic Stress Diagnostic Scale (PDS) (21). Secondary outcomes included the: Center for Epidemiological Studies-Depression Scale (CES-D) (22), Pennebaker Inventory of Limbic Languidness (PILL) (23), and Brief Symptom Inventory (BSI) (24). Process measures included the: Positive and Negative Affect Scale (PANAS) (29), measured pre- and post-writing session, as well as the Essay Evaluation Measure (EEM) (30), measured post-writing session, and the follow-up questionnaire, measured at 2-week and 1-month post-writing intervention.

The SCID (20) is a diagnostic interview assessing SUD diagnosis, including alcohol and other drugs. The SCID has demonstrated good validity and high interrater reliability for SUDs (25).

The PDS (21) is a 49-item self-report measure focused on PTSD symptom severity and diagnosis, with items parallel to DSM-IV criteria (26). The PDS has high test-retest reliability (r = .83), high internal consistency (α = .92), and high convergent validity (27). Trauma symptom severity was defined as the sum scores for items focused on re-experiencing, arousal, and avoidance symptoms. At baseline, internal consistency for trauma symptom severity for this sample was .99.

The CES-D (22) is a 20-item self-report measure of depression, and has high internal consistency in psychiatric settings (α. = .90). Internal consistency at present study baseline for this sample was .87.

The PILL (23) is a 54-item scale that assesses the frequency of common physical symptoms and sensations (e.g. “headaches” “congested nose” “coughing”). Chronbach’s alphas for the PILL range from .88 to .91, with 2-month test-retest reliabilities of .79 to .83. PILL has a mean score of 112.7 (SD = 24.7) (23). At baseline, internal consistency for this sample was .94.

The BSI (24), a 53-item self-report measure, is a shortened version of the Symptom Check List-90 (SCL-90). It assesses nine symptoms of distress (Somatization; Obsessive-Compulsive; Interpersonal Sensitivity; Depression; Anxiety; Hostility; Phobic Anxiety; Paranoid Ideation; Psychoticism). The BSI has high scale-by-scale correlations with the SCL-90, as well as high internal consistency (Cronbach’s α.: .71-.85), convergent, discriminant, and test-retest reliability (r = .68–.91), and construct validity (28). At baseline, internal consistencies for the 9 BSI indices for this sample were .74-.87.

Process Measures

The PANAS (29) assesses negative affect (NA) and positive affect (PA), using 10 items for NA and 10 items for PA. It was administered pre- and post-writing, and only the negative affect scale was used in this study. The NA scale has high internal consistency reliabilities, ranging from .84 to .87, and they are unchanged by the time point used. Internal consistency for the post-writing NA scale on Day 1 of writing for this sample was .92.

The EEM (30) served as manipulation check to evaluate participants’ self-reports of their response to the experimental and control manipulation after each writing session. It assesses participants’ report, along separate 7-point scales, ranging from 1 (not at all) to 7 (a great deal), the extent to which their essays were meaningful, personal, and revealing, and the extent they had actually talked to others, had wanted to talk to others, and had actively held back from talking to others about the subject(s) of the essays (31). After the first writing session, internal consistency of the measure for this sample was .79.

The follow-up questionnaire was an investigator-developed survey that examined issues of cross-contamination inherent in studies done in residential settings. The follow-up questionnaire served as a fidelity check, and measured possible diffusion of treatment. The questionnaire included closed and open-ended questions and surveyed whether participants heard about, shared or received writing instructions from others in the facility. It also addressed whether the participant found the writing helpful and talked about the subject matter from her essays during treatment.

Procedures

The University’s Institutional Review Board approved research procedures. A CONSORT diagram (Figure 1) summarized study screening, enrollment, treatment, assessment, and retention of participants. Admission records identified potential study participants and those women were approached within the first few days of treatment. Those who met inclusion criteria were invited for participation in a study that involved writing stories related to their life (32).

Figure 1.

Women providing informed consent were randomly assigned (using a random numbers table) to one of two conditions: (a) expressive writing (emotional topic) or (b) control writing (neutral topic). Writing instructions were based on protocols used by Pennebaker et al. (14) and Sloan and Marx (33). In addition to being read aloud by the investigator before each writing session, a copy of the writing instructions was attached to the front flap of each journal so that participants could re-read instructions at any time.

Writing Procedure

Once enrolled, participants completed baseline assessments, four 20-minute writing sessions on consecutive days, and both a 2-week and 1-month follow-up assessment. In order to reduce unintentional expectancy effects and investigator bias, one experimenter administered baseline assessments and all writing instructions, while a different experimenter, blind to study condition, administered PANAS, EEMs, and follow-up assessments.

Writing Session One

The day after completing baseline assessments, each participant was administered the writing instructions. Participants were asked not to discuss the study or writing instructions. After receiving writing instructions from one experimenter, participants were introduced to a second experimenter.

The second experimenter had participants complete a PANAS (29) prior to handing them a blank notepad. Participants located a comfortable place where they would not be disturbed (e.g., their room), and wrote about their assigned topic for 20 minutes. Twenty minutes later, the second experimenter administered the PANAS and EEM (33). Notepads, marked by participant ID, were then dropped into a cardboard box along with participants’ responses to PANAS and EEM responses.

Writing Sessions Two through Four

Participants were greeted for each session by the first experimenter, who read the writing instructions aloud. The second experimenter instructed participants to complete PANAS before writing for 20 minutes. The PANAS and EEM were completed after each writing session.

2-Week and 1-Month Follow-Up

Participants completed follow-up assessments 2-weeks and 1-month post-writing intervention. A 2-week follow-up was added after 22 participants had already been enrolled; this was done to insure some data collection from participants who might leave against medical advice (AMA) prior completing their 60-day treatment, as well as to assess the shorter-term effects of the intervention.

Participants received a $5 gift card after completing the baseline assessment, a $5 gift card after each writing session ($20 total), a $10 gift card after completing the 2-week follow-up, and a $15 gift card upon completion of the 1-month follow-up.

Data Analysis

Kolmogorov-Smirnov and Shapiro-Wilk statistics determined that primary and secondary outcomes were normally distributed. T- and chi-square tests examined whether significant baseline differences existed between conditions. A series of analyses of variance (ANOVA) were then used to test for between-group differences in (a) 2-week clinical outcomes, and (b) clinical outcomes at the end of the 1-month follow-up. For the first research hypothesis, 2-week effects were examined with a 2 Condition (expressive writing; control writing) by 2 Time (baseline; 2-week follow-up) repeated-measures analysis of variance (ANOVA) separately for each outcome measure (PDS, CES-D, PILL, BSI). One-month effects were also tested by examining between-group differences over time using a 2 Condition (expressive writing; control writing) by 2 Time (baseline; 1-month follow-up) repeated-measures analysis of variance (ANOVA) for each outcome measure (PDS, BSI, CES-D, PILL). Condition × Time interaction effects were examined for differential treatment effects with post-hoc analysis of the main effect of time within each treatment group. Since an investigator was present during assessments, missing data were minimal. In the intention-to-treat analyses, with the entire sample (149 participants), however, missing outcome data for ‘non-completers’ (e.g., participants who did not complete all writing sessions, the 2-week, and/or the 1-month follow-up assessments) were addressed by carrying forward each participant’s previous score.

For the second research hypothesis, a repeated measures analysis of negative affect was derived from PANAS. A 2 Condition (expressive writing; control writing) by 4 Session (writing 1; writing 2; writing 3; writing 4) by 2 Time (pre- and post-writing session) repeated measures multivariate analysis of variance (MANOVA) examined the interaction between pre-post writing NA and condition at each of the four sessions.

T- and chi-square tests compared writing conditions on items of the EEM and follow-up questionnaire to evaluate participants’ self reports of their response to the experimental manipulation. Because participants completed the EEM four times (after each writing session), the mean of the four days for each participant was used in analyses.

RESULTS

Participant Characteristics

Overall, women enrolled in the study (Table 1) were an average of 36.3 (SD = 8.6) years old; and had completed an average of 11.2 (SD = 1.8) years of education; most (92.7%) were single/never married; and a majority (70%) identified their race as African-American. Most participants reported at least one lifetime trauma at baseline (assessed by PDS). Over half met diagnostic criteria for current PTSD. SUD women with co-morbid PTSD reported significantly more types of trauma than women without PTSD, t (147) = 5.1, p < .005. A majority of participants indicated cocaine as one of their primary drugs of dependence, with over half meeting DSM-IV criteria for substance dependence for more than one drug.

Table 1.

Participant Characteristics: Expressive Writing (n = 82) and Control Writing (n = 67)

	Expressive Writing % or M (SD)	Control Writing % or M (SD)	Total Sample % or M (SD)
Age (years)	37.7 (8.7)	34.6 (8.2)	36.6 (8.3) *
Education (last grade completed)	11.5 (1.6)	10.9 (2.0)	11.2 (1.8)
Marital Status
Married	4.4%	9.8%	7.3%
Single/Never Married	95.6%	90.2%	92.7%
Race
% Caucasian	25.6%	23.5%	24.7%
% African-American	69.5%	70.6%	70.0%
% Hispanic	1.2%	0.0%	0.7%
% Other	3.7%	5.9%	4.7%
At least One Trauma	96.3% (n = 79)	95.5% (n = 64)	96.0% (n = 143)
Mean Number of Different Types of Trauma Events	3.8 (SD = 2.2)	3.6 (SD = 2.4)	3.7 (SD = 2.3)
PTSD diagnosis (n = 81)	4.6 (SD = 2.2)	4.5 (SD = 2.6)	4.6 (SD = 2.3)
No PTSD diagnosis (n = 68)	2.8 (SD = 1.8)	2.8 (SD = 1.8)	2.8 (SD = 1.8)
PTSD Diagnosis	58.5% (n = 48)	49.3% (n = 33)	54.4% (n = 81)
Trauma Symptom Severity	21.3 (SD = 13.4)	17.5 (SD = 12.4)	19.5 (SD = 13.1)
CES-D depression score	25.2 (SD = 12.0)	21.1 (SD = 10.9)	23.3 (SD = 11.7)*
CES-D Cutoff ≥ 16	76.8%	67.2%	72.5%
PILL score	118.8 (SD = 37.6)	113.5 (SD = 34.2)	116.4 (SD = 36.0)
DSM-IV Substance Dependence Diagnosis (current)
Alcohol	32.9%	23.9%	28.9%
Amphetamine/Stimulant	0.0%	1.5%	0.7%
Cannabis	11.0%	9.0%	10.1%
Cocaine	80.5%	83.6%	81.9%
Hallucinogen	1.2%	0.0%	0.7%
Inhalant	0.0%	0.0%	0.0%
Opioid	47.6%	41.8%	45.0%
Sedative	6.1%	4.5%	5.4%
More than One Drug	63.4%	49.3%	57.0%

Randomization

Participants randomized to expressive writing had higher CES-D depression scores than controls (M = 25.2 [SD = 12.0] versus M = 21.1 [SD = 10.9], t (148) = 2.2, p < .05), and higher BSI Anxiety scores (M = 1.3 [SD = 1.1] versus M = 0.8 [SD = 0.8]), t (148) = 3.0, p < .005. There were no other baseline differences between conditions.

Attrition

Writing Session Attrition

A chi-square analysis showed no significant relationship between condition assignment and number of writing sessions completed, p > .05. As depicted in the CONSORT diagram, rates of compliance for both conditions were high. Specifically, 97.6% of expressive writing and 97.0% of control writing participants completed at least two writing sessions, and most (94.6%) completed all four writing sessions.

Follow-Up Assessment Attrition

Among expressive writing participants, there were no differences between ‘completers’ versus ‘non-completers’ on dependent variables at 2-week follow-up. Among control participants, women who did not complete the 2-week follow-up were significantly more depressed at baseline compared to those who completed the follow-up (M = 28.0 [SD = 6.8] versus M = 20.0 [SD = 11.1]), t (65) = 2.1, p < .05), and were more likely to have had clinically elevated depression (≥ 16 CES-D cutoff) at baseline (100 % versus 62.1%), χ² (2, N = 67) = 5.1, p < .05. No differences between ‘completers’ versus ‘non-completers’ were found at 1-month follow-ups on any dependent variables, p > .05.

Hypothesis 1: 2-Week and 1-Month Effects of Expressive Writing

Trauma Symptom Severity

There was a Condition × Time interaction for levels of traumatic stress at 2-week follow-up, F (1,149) = 5.8, p < .05, partial eta² = .04. Planned contrasts revealed a significant decrease in trauma symptom severity from baseline to 2-weeks among expressive writing, F (1,82) = 8.6, p < .005, partial eta² = .10, but not among control writing participants, p > .05. There was no interaction for trauma symptom severity at 1-month follow-up, p > .05; however, both conditions revealed significant improvements (Table 2). Expressive writing participants revealed a significant decrease from baseline to 1-month follow-up, F (1,82) = 11.1, p < .005, partial eta² = .12, as did controls, F (1,67) = 7.3, p < .01, partial eta² = .10.

Table 2.

Effects of Expressive Writing versus Control Writing on Two-Week Follow-Up (Intent-to-Treat)

Groups	Expressive writing (N = 82)			Control Writing (N = 67)
	Baseline Mean (SD)	2-Week-Posttest Mean (SD)	1 Month Posttest Mean (SD)	Baseline Mean (SD)	2 Week-posttest Mean (SD)	1 Month-posttest Mean (SD)
Trauma Symptom Severity	21.3 (13.4)	18.0* (12.3)	17.0 (12.6)	17.5 (12.6)	18.0 (11.7)	14.5 (12.1)
Depression	25.2 (12.0)	21.9* (10.7)	19.4 (11.8)	21.1 (10.9)	21.3 (11.8)	17.6 (11.1)
PILL	118.8 (37.6)	113.4 (36.4)	106.8 (39.1)	113.5 (34.2)	108.8 (28.3)	105.6 (39.0)
BSI Anxiety Index	1.2 (1.0)	0.9* (0.9)	0.9 (1.0)	0.8 (0.8)	0.9 (0.8)	0.6 (0.7)
BSI Somatic Index	1.0 (0.8)	1.0 (0.8)	0.8 (0.8)	0.8 (0.8)	1.0 (0.9)	0.7 (0.8)

^*p < .05

Depression

There was a significant Condition × Time interaction for depression, F (1,149) = 4.3, p < .05, partial eta² = .03 at 2-weeks. Planned contrasts demonstrated significant decreases in depression among expressive writing, F (1,82) = 7.4, p < .01, partial eta² = .08, but not among control writing participants, p > .05. There was no interaction at the 1-month follow-up (p > .05), however, both conditions revealed significant improvements in depression. Expressive writing participants showed significant improvements in depression at 1-month follow-up, F (1,82) = 16.1, p < .005, partial eta² = .17, as did control participants, F (1,67) = 9.3, p < .005, partial eta² = .12.

Physical Health

There was no Condition × Time interaction in the reduction of physical health symptoms and sensations at the 2-week, p > .05, or 1-month follow-up, p > .05. However, significant improvements were found at 1-month follow-up for both the expressive writing, F (1,82) = 18.3, p < .005, partial eta² = .18, and control writing condition, F (1,67) = 4.5, p < .05, partial eta² = .06.

BSI

There was a significant Condition × Time interaction for anxiety, F (1,149) 13.9, p < .005, partial eta² = .09. Expressive writing participants showed significant improvements in anxiety scores at the 2-week follow-up, F (1,82) = 14.1, p < .005, partial eta² = .15, while control participants did not, p >.05. There was no interaction at 1-month follow-up, p > .05.

There was a trend approaching a significant interaction for the Somatic Index, F (1,149) = 3.8, p = .05. While the control condition revealed an increase in somatic symptoms from baseline to 2-week follow-up, F (1,67) = 4.4, p < .05, partial eta² = .06, the expressive writing condition revealed no differences suggesting a possible protective factor of the writing. This trend was not maintained at the 1-month follow-up, yet overall improvements were seen for both conditions. There were no interactions between conditions over time for the remaining seven BSI indices (all > .05).

Hypothesis 2: Expressive Writing on Negative Affect

There was a significant Condition × Session × Time interaction for PANAS negative affect scores, F (4,149) = 7.3, p < .005, partial eta² = .30. Additionally, univariate F tests revealed that groups did not differ in negative affect prior to writing, suggesting that the immediate increase in negative affect after writing was brief (Table 3).

Table 3.

Negative Affect: Pre-Post Writing Session

Session	Expressive Writing Mean (SD)	Control Writing Mean (SD)
Writing Session 1 Time 1	18.7 (7.9)	16.2 (7.2)
Writing Session 1 Time 2	24.0 (10.9)**	15.2 (8.2)
Writing Session 2 Time 1	17.6 (8.2)	13.3 (4.9)
Writing Session 2 Time 2	22.4 (11.2)**	13.5 (5.2)
Writing Session 3 Time 1	16.8 (7.7)	14.1 (6.2)
Writing Session 3 Time 2	19.5 (9.5)**	13.4 (5.1)
Writing Session 4 Time 1	17.6 (8.4)	13.9 (6.2)
Writing Session 4 Time 2	18.7 (9.0)	13.6 (5.4)

Note: Time 1 PANAS was assessed after writing instructions were given, but prior to writing.

^**p < .005

Univariate tests revealed a significant interaction between time by condition at writing session one, F (1,149) = 29.9, p < .005, with a significant group main effect, partial eta² = .17. There was a significant time by condition interaction at writing session two, F (1,149) = 15.6, p < .005, with a significant group main effect, partial eta² = .10. The interaction continued to be significant at writing session three, F (1,149) = 9.3, p < .005, with a significant group main effect, partial eta² = .06. At writing session four, however, there was no interaction, F (1,147) = 2.0, p > .05. As illustrated, analyses revealed that the main effects decreased over time; by writing session four, expressive writing participants were able to write about traumatic event without having the spike in negative affect afterward.

Manipulation and Fidelity Checks

EEM

When compared to control writing participants, expressive writing participants described their essays as more meaningful, t (139) = 9.1, p < .0005, more personal, t (139) = 13.3, p < .0005, and more revealing of one’s emotions, t (139) = 18.0, p < .0005. They were more likely to have wanted to discuss it with others, t (139) = 8.7, p < .0005, than control writing participants. They also reported being more likely to have talked with others about the content of their essays, t (139) = 10.0, p < .0005, and also to have actively held back from talking with others about the content, t (139) = 4.5, p < .0005, than controls.

Follow-up Questionnaire

Only 2 (1.7%) participants in the control condition reported hearing about the expressive writing instructions from another participant. Furthermore, expressive writing participants were more likely than controls to: report finding the writing to be helpful (98.6% versus 72.0%), χ² (2, N = 120) = 18.6, p < .005, continue journaling using the writing instructions given after the sessions ended (42.0% versus 14.0%), χ² (2, N = 120) = 10.8, p < .005; and to have talked about the subject matter form their essays during treatment (15.9% versus 0.0%), χ² (2, N = 120) = 8.8, p < .005 . All (100%) of the expressive writing participants who talked about the subject matter from their essays during their treatment found it to be helpful. Expressive writing participants reported that expressive writing helped them work through issues and deal more effectively with emotions related to trauma/stress (Table 4).

Table 4.

Follow-Up Questionnaire: 5 Randomly Selected Quotes from Each Condition

Expressive Writing Participants	Question 4: “Did you find the writing helpful? Why? Why not? (please put your answer below)”
“Very helpful. I’ve never in my life experienced something like this. It hit the core of everything”
“It made me realize that what happened to me wasn’t my fault”
“Very helpful. It got me in touch with my feelings and helped me work through some of my issues and problems. It was good for me”
“Made me bring some of the issues that wrote about, especially the attempted rape and how I was able to not have sex because of fear”
“Because it made me think back to some issues that were hindering me from moving forward. It helps me let go and move forward. I feel much better”
Control Writing Participants	Question 4: “Did you find the writing helpful? Why? Why not? (please put your answer below)”
“It was just being plain, you can’t express no emotions”
“Kind of neutral! Not aware of what the goal was. Helps me in my purpose in gaining knowledge of it”
“Because it made me realize that I can write just about anything”
“I found out that it was kind of hard trying to remember everything I did in a day’s time”
“Shallow subject matter”

DISCUSSION

The present study collected benchmark data on whether Pennebaker’s expressive writing could be applied to women in residential SUD treatment. Trauma rates in this study were comparable to those previously reported in the literature (5), supporting the rationale for the clinical trial and study hypotheses.

2-week and 1-Month Outcomes

Supporting hypothesis one, expressive writing participants showed greater reductions in trauma symptom severity, depression, and anxiety than control writing participants at the 2- week follow-up. Expressive writing may be a protective factor in somatic symptoms. While no group differences were found at the 1-month follow-ups, both conditions showed significant positive changes, likely due to being in residential treatment for 30 days (34). Due to baseline differences between conditions, results should be interpreted with caution.

Trauma

Short-term (2-week) findings are consistent with research suggesting that writing about emotional experiences using Pennebaker’s expressive writing results in a decreased degree to which the experience is emotionally distressing or painful (35). One potential reason is based on exposure theory, which suggests that attempting not to think about traumatic events in order to avoid overwhelming emotions may generate further feelings of distress and fear (11). Thus, expressive writing may act as a medium for exposure to previously avoided stimuli. Repeat exposure over several writing sessions may allow for habituation and extinction of negative emotional associations and/or offer corrective information to the person about the stimuli, responses, and meaning of the event (32).

While it is generally thought that exposure to the same traumatic event is essential for extinction, expressive writing does not require writing about the same event at each session, suggesting that stimulus-related habituation may take place regardless (7). Studies to-date comparing instructions for participants to write about the same or different topics each day have yielded mixed results. For instance, Sloan and colleagues (35) found college students with a trauma history who wrote about the same trauma experience at each session showed improvements at follow-up, while those instructed to write about a different trauma each day did not. Other studies have obtained more robust results when participants were allowed to choose their writing topic at each session (8; 36).

The importance of short-term improvements at 2-weeks versus 1-month should not be discounted, particularly since we are dealing with a clinical population and a chronic disorder. It is likely that extending the intervention in ways that “boost” the intervention (e.g., more writing sessions during different points in treatment) would be helpful. While writing conditions did not differ at 1-month, both conditions demonstrated positive changes on core outcomes. The present study is unique, however, in that it was conducted in the context of residential treatment, which, in and of itself, is likely to produce improvements in physical and psychosocial functioning over time.

Depression

Expressive writing participants showed significant reductions in depressive symptoms at the 2-week follow-up, while controls did not. Results are consistent with research among women who had endured intimate partner violence (12), showing greater drops in depression among those in the traumatic disclosure as compared to the control writing condition. However, given that baseline depression differences existed, this finding should also be interpreted with caution.

Health

There were no differences between the two conditions in the reduction of physical health symptoms and sensations over time, which conflicts with an extensive body of research that suggests physical health benefits from expressive writing (37). However, there were improvements in physical symptoms regardless of condition at 1-month follow-up. While we did not find the health improvements we expected from Pennebaker’s expressive writing, the present study uses a sample of women who, generally speaking, were receiving little to no healthcare treatment prior to admittance. Once admitted into treatment, they began receiving psychotherapy and/or medication, which likely influenced symptoms and self-reports above and beyond improvements we may otherwise witness as a result of the writing intervention.

BSI

Expressive writing participants reported greater reductions in anxiety than controls at 2-weeks; however, there were no group difference at the1-month follow-up. Again, caution should be used when interpreting these findings due to baseline differences. Nonetheless, these outcomes suggest expressive writing might assist with emotion regulation by promoting attention, habituation, and cognitive restructuring. Expressive writing can potentially help influence attention toward or away from different dimensions of a stressor (11). Another theory suggests that cognitive restructuring may initiate changes in stress-related thoughts and appraisals. To the degree that people can cognitively negotiate stressful experiences, they should experience a decrease in and possible extinction of anxiety and intrusive thoughts.

Analyses revealed a trend approaching significance for the Somatic Index, which might suggest expressive writing could have a protective effect during the early stages of residential treatment. If women with SUDs are likely to show initial increases in somatic problems (start of treatment), possibly as a result of their bodies readjusting to no longer being physically dependent on drugs, then expressive writing may benefit women by preventing somatic deterioration.

Expressive Writing on Negative Affect

Consistent with hypothesis two, negative affect produced by expressive writing was short-lived. Increased negative affect immediately following writing is consistent with previous literature (38), and provides evidence for emotional engagement and (fear) activation (39). Trauma experts claim emotional engagement is essential for habituation (40). In order for successful habituation, individuals should endure strong negative emotions initially (11), followed by gradual reductions in negative emotions within and across writing sessions. Such habituation may indicate desensitization to trauma-related thoughts and memories (i.e., stimulus-specific habituation), and possibly an enhanced tolerance of ones’ negative emotional responses to stress-provoking stimuli in general (i.e., response-specific habituation) (11). Thus, initial spikes in distress and consequent habituation may be reflective of emotional processing and enhanced self-regulation, and should be associated with improvements in one’s reaction (psychological and physical) to traumatic memories (41).

Limitations

The study had a number of limitations. First, although participants were randomized to conditions, expressive writing participants started with higher pre-writing scores; while, depression and anxiety scores were the only significant pre-writing differences, these differences raise concerns that the effects could be the result of a regression to the mean. Another limitation is that one month is a relatively short follow-up period. Future studies should examine group differences over longer periods of time post-intervention (including post-residential treatment outcomes and substance use), even though some studies show the most dramatic effects more immediately following the intervention.

Furthermore, it was realized 22 participants into recruitment that short-term follow-up would be beneficial in catching participants who did not stay the intended 60 days; therefore, we did not have 2-week assessments on those 22 participants. It would also be useful to examine potential moderators, such as high versus low avoidance personalities (42). Finally, individuals with more severe mental health problems (e.g., SUD, PTSD) may require more writing sessions or longer writing sessions to realize positive effects.

Study Implications and Applications

This study has a number of important implications. First, it provides evidence that trauma-focused interventions can be incorporated into traditional residential SUD treatment without triggering more harm. Women safely expressed their deepest emotions and thoughts, many for the first time, about a traumatic and/or stressful event without any indications of a lasting increase in psychiatric distress. During follow-up assessments, the majority of expressive writing participants reported that it helped them work through issues and deal more effectively with emotions related to trauma/stress. These reports were offered spontaneously by participants, as well as through the follow-up questionnaire.

To the best of our knowledge, this study was the first to examine the utility of Pennebaker’s expressive writing with drug-dependent women. The ability to reduce trauma symptom severity, depression, and anxiety early in drug treatment can be beneficial in many ways (e.g., increasing patients’ ability to concentrate during counseling sessions, reducing risks for leaving treatment prematurely, and reducing risks for relapse). Thus, future studies should explore whether extending the intervention in ways that maintain or further ‘boost’ short-term improvements at a later time, possibly through more writing sessions distributed over the course of SUD treatment, may enhance longer-term effects. Future analyses should examine essay content and congruence between topic chosen and measures of traumatic stress, as well as whether the same versus difference topic across all writing sessions.

The population also represents women of diverse race, substance use diagnoses, and psychiatric symptom severity. The majority of women had less than a high school level of education, were under-employed, uninsured, single/never married, and were thus a vulnerable population with a variety of medical and psychosocial problems and fewer resources than others with SUDs. It is noteworthy that in a population that generally has difficulty adhering to treatment, expressive writing was well received and regarded as helpful by the participants, themselves (43). Thus, expressive writing was found to be a brief and low cost adjunct to current residential SUD treatment. Expressive writings’ efficacy, efficiency, and cost-effectiveness, suggest that writing as a means of disclosing traumatic experiences may be a useful adjunct to traditional SUD treatment.