follow us on twitter: @GoAfricaNetwork
Importance Mindfulness-based interventions may be acceptable to veterans who have poor adherence to existing evidence-based treatments for posttraumatic stress disorder (PTSD).
Objective To compare mindfulness-based stress reduction with present-centered group therapy for treatment of PTSD.
Design, Setting, and Participants Randomized clinical trial of 116 veterans with PTSD recruited at the Minneapolis Veterans Affairs Medical Center from March 2012 to December 2013. Outcomes were assessed before, during, and after treatment and at 2-month follow-up. Data collection was completed on April 22, 2014.
Interventions Participants were randomly assigned to receive mindfulness-based stress reduction therapy (n = 58), consisting of 9 sessions (8 weekly 2.5-hour group sessions and a daylong retreat) focused on teaching patients to attend to the present moment in a nonjudgmental, accepting manner; or present-centered group therapy (n = 58), an active-control condition consisting of 9 weekly 1.5-hour group sessions focused on current life problems.
Melissa A. Polusny, PhD1,2,3; Christopher R. Erbes, PhD1,2,3; Paul Thuras, PhD1,3; Amy Moran, MA1; Greg J. Lamberty, PhD1,3; Rose C. Collins, PhD1; John L. Rodman, PhD1; Kelvin O. Lim, MD1,3
Main Outcomes and Measures The primary outcome, change in PTSD symptom severity over time, was assessed using the PTSD Checklist (range, 17-85; higher scores indicate greater severity; reduction of 10 or more considered a minimal clinically important difference) at baseline and weeks 3, 6, 9, and 17. Secondary outcomes included PTSD diagnosis and symptom severity assessed by independent evaluators using the Clinician-Administered PTSD Scale along with improvements in depressive symptoms, quality of life, and mindfulness.
Results Participants in the mindfulness-based stress reduction group demonstrated greater improvement in self-reported PTSD symptom severity during treatment (change in mean PTSD Checklist scores from 63.6 to 55.7 vs 58.8 to 55.8 with present-centered group therapy; between-group difference, 4.95; 95% CI, 1.92-7.99; P=.002) and at 2-month follow-up (change in mean scores from 63.6 to 54.4 vs 58.8 to 56.0, respectively; difference, 6.44; 95% CI, 3.34-9.53, P < .001). Although participants in the mindfulness-based stress reduction group were more likely to show clinically significant improvement in self-reported PTSD symptom severity (48.9% vs 28.1% with present-centered group therapy; difference, 20.9%; 95% CI, 2.2%-39.5%; P = .03) at 2-month follow-up, they were no more likely to have loss of PTSD diagnosis (53.3% vs 47.3%, respectively; difference, 6.0%; 95% CI, −14.1% to 26.2%; P = .55).
Conclusions and Relevance Among veterans with PTSD, mindfulness-based stress reduction therapy, compared with present-centered group therapy, resulted in a greater decrease in PTSD symptom severity. However, the magnitude of the average improvement suggests a modest effect.
Trial Registration clinicaltrials.gov Identifier: NCT01548742
Posttraumatic stress disorder (PTSD) affects 23% of veterans returning from deployments in Afghanistan (Operation Enduring Freedom) and Iraq (Operation Iraqi Freedom).1 Left untreated, it is associated with high rates of comorbidity, disability, and poor quality of life.2 The US Department of Veterans Affairs (VA) has invested heavily in the dissemination of prolonged exposure therapy and cognitive processing therapy.3Robust evidence supports the efficacy of these 2 first-line treatments.4 Yet 30% to 50% of veterans participating in prolonged exposure or cognitive processing therapy fail to show clinically significant improvements,5,6 and dropout is high, ranging from 30% to 38%5– 7 in randomized trials and 32% to 44% in clinic-based studies.8,9 A recent chart review found that 60% of eligible Operation Enduring Freedom/Operation Iraqi Freedom veterans failed to begin or dropped out of these treatments.10Avoidance and difficulties tolerating trauma-focused material likely contribute to dropout.7,11 Thus, research aimed at testing novel treatments for PTSD in this population is important.
Evidence suggests that mindfulness-based stress reduction, an intervention that teaches individuals to attend to the present moment in a nonjudgmental, accepting manner,12 can result in reduced symptoms of depression and anxiety.13 By encouraging acceptance of thoughts, feelings, and experiences without avoidance, mindfulness-based interventions target experiential avoidance, a key factor in the development and maintenance of PTSD.14 This randomized clinical trial compared mindfulness-based stress reduction with an active, credible intervention, present-centered group therapy. We hypothesized that veterans randomly assigned to mindfulness-based stress reduction would show greater reductions in self-reported and interview-rated PTSD severity and loss of diagnosis after treatment and at 2-month follow-up compared with those randomized to present-centered group therapy.
Participants were veterans who met the following inclusion criteria: (1) current full PTSD according to theDiagnostic and Statistical Manual of Mental Disorders (Fourth Edition) (DSM-IV)15 or subthreshold PTSD, defined as endorsement of DSM-IV criterion A1 and at least 1 symptom each from criteria B, C, and D with significant impairment; (2) agreement to not receive other psychotherapy for PTSD during study; and (3) if being treated with psychoactive medications, a stable regimen for at least 2 months prior to study entry. Exclusion criteria were (1) current substance dependence (except nicotine or caffeine); (2) current psychotic disorder (eg, schizophrenia, bipolar disorder); (3) prominent current suicidal or homicidal ideation; and (4) cognitive impairment or medical illness that could interfere with treatment.
Patients were recruited through advertisements and clinical referrals at a large VA medical center. All patients provided written informed consent for participation in this study, which was approved by the Minneapolis VA Medical Center institutional review board. Participants completed a 5-hour eligibility and baseline assessment that included a structured clinical interview and self-report measures. Master’s-level assessors supervised by study authors (C.R.E. and G.J.L.) served as independent evaluators blinded to treatment condition. Posttraumatic stress disorder and Axis I psychiatric disorders were assessed using structured clinical interviews.16,17 Outcomes were assessed before treatment, at 3-week intervals during treatment (weeks 3 and 6), after treatment (week 9), and at 2-month follow-up (week 17). Data collection was completed April 22, 2014.
Randomization was conducted using SAS PROC PLAN in blocks of 4 to ensure even randomization across the length of the study. A restricted electronic randomization chart was provided to the study coordinator by the statistician. Veterans were randomized approximately every 2 months over a 19-month period, for a total of 9 cohorts composed of 1 group each of the 2 conditions.
The trial protocol is available in the Supplement.
Treatment was delivered in a group format according to manualized protocols by 2 instructors/clinicians. For mindfulness-based stress reduction, lead instructors completed a 9-day intensive practicum training at the University of Massachusetts Center for Mindfulness. Each lead instructor was assisted by a doctoral-level clinician. All instructors/clinicians completed a 2-day training and received weekly or biweekly supervision (by senior staff at the University of Minnesota Center for Spirituality and Healing for mindfulness-based stress reduction; by developer Melissa Wattenberg, PhD, for present-centered group therapy).
Mindfulness-Based Stress Reduction
Standard protocol consists of 8 weekly 2.5-hour group sessions and a daylong retreat.12 The intervention was modified to include an orientation to the program that incorporated PTSD psychoeducation and treatment rationale (session 1), followed by 7 weekly 2.5-hour group sessions and a 6.5-hour retreat, for a total of 9 sessions. The program teaches participants to attend to the present moment (immediate emotional and physical states, including discomfort) in a nonjudgmental and accepting way. Sessions include didactic training and formal practice in 3 meditation techniques. The body scan is a guided exercise that systematically directs attention through various areas of the body. Sitting meditation involves developing capacity for sustained self-observation through directing attention to specific experiences (eg, the breath, physical sensations, thoughts, emotions, sensory stimuli). Mindful yoga involves gentle stretches and movements practiced with present-moment attention, which encourages greater body awareness. Meditation techniques were taught in the same manner as is typical in programs offered in the community, and no modifications were made to specifically accommodate PTSD. The daylong silent retreat provides an opportunity for sustained practice of these techniques. Additionally, the program encourages individuals to practice meditation techniques at home and to cultivate present-moment awareness in ordinary daily activities (eg, mindful eating and driving).
Present-Centered Group Therapy
Present-centered group therapy is an active-control condition shown to benefit individuals with PTSD.18,19It controls for nonspecific therapeutic factors by providing professional contact, a credible therapeutic rationale, and corresponding specific ingredients (eg, problem solving) for reducing distress, with positive therapeutic expectancy similar to mindfulness-based stress reduction.13 The intervention consists of 9 weekly 1.5-hour group sessions focused on current life problems as manifestations of PTSD.20 Session 1 focuses on providing psychoeducation about PTSD and treatment rationale, building group cohesion, and goal setting. Sessions 2 through 8 focus on discussing daily difficulties. Session 9 focuses on reviewing accomplishments and planning for the future. Therapists are nondirective and encourage patients to provide each other with support, problem solving, and validation. There was no discussion of mindfulness meditation techniques or traumatic experiences.
The primary outcome, change in PTSD symptom severity over time, was assessed using the PTSD Checklist21 (PCL; range, 17-85; higher scores indicate more severe symptoms) at all assessment points (baseline and weeks 3, 6, 9, and 17). It has excellent internal consistency (Cronbach α = 0.94-0.97), test-retest reliability (0.96), and concurrent validity.22 The minimal clinically important difference (MCID) for self-reported PTSD symptom severity is a reduction of 10 or more points on the PCL.23
Diagnosis and symptom severity of PTSD were also assessed using the Clinician-Administered PTSD Scale (CAPS)16 before and after treatment and at 2-month follow-up (baseline, week 9, and week 17). Potentially traumatic events were identified using the Life Events Checklist and further assessed during interview.24We used the recommended 1/2 scoring rule, whereby a frequency score of 1 (0 = none of the time; 4 = most or all of the time) and intensity score of 2 (0 = none; 4 = extreme) is required to consider each symptom as present.22 Diagnoses were based on DSM-IV criteria for PTSD; a severity score was also calculated by summing frequency and intensity scores for all 17 symptoms (range, 0-136; higher scores indicate more severe PTSD). A reduction of 10 or more points is considered the MCID for the CAPS.5 Comorbid depression symptoms were assessed using the Patient Health Questionnaire 925 (PHQ-9; range, 0-27; higher scores indicate more symptoms). The MCID for the PHQ-9 is a reduction of 5 or more points.26Quality of life was assessed using the World Health Organization Quality of Life–Brief (WHOQOL-BREF).27 This study reports the summed total score (range, 0-130; higher scores indicate greater quality of life).
Mindfulness skills (observing, describing, acting with awareness, nonjudging of inner experience, and nonreactivity to inner experience) were assessed using the Five Facet Mindfulness Questionnaire (FFMQ)28at all assessment points (range, 39-195; higher scores indicate greater mindfulness).
Participants’ beliefs about the rationale and logic of the treatment (credibility scale range, 1-9; higher scores indicate more rationality/logic) and likelihood of the treatment’s success (expectancy scale range, 1-9; higher scores indicate greater expectations of success) in reducing PTSD symptoms were assessed using the Credibility/Expectancy Questionnaire29 at week 3. Participants reported treatment satisfaction at week 9 using a scale ranging from 1 to 4 with higher scales indicating greater satisfaction. Demographic information, including self-reported race/ethnicity, was collected at baseline to characterize the sample. Mental health treatment history and psychotherapy health care visits (individual and group therapy delivered both in specialty PTSD and mental health clinics) from October 1, 1999, to participants’ baseline dates were extracted from VA electronic medical records. We determined the mean duration (in months) of previous mental health care, total number of psychotherapy mental health visits, and whether a participant had previously received 8 or more therapy sessions at baseline.
All treatment sessions were videotaped. Two senior clinicians independent of treatment delivery rated 10% of sessions from each condition using a rating tool adapted from other trials of PTSD group treatment.30
Intention-to-treat analyses were conducted for all outcomes. Baseline differences between groups were examined using analysis of variance for continuously measured variables and χ2 statistics for noncontinuous variables. Mixed-effects models were used to analyze the efficacy of mindfulness-based stress reduction compared with present-centered group therapy in reducing PTSD symptoms over 9 weeks of treatment and at 2-month follow-up.31 Mixed-effects models are flexible regression methods for incomplete repeated-measures data and allow continuous and categorical covariates, fixed and time-dependent covariates, and a specification of unstructured as well as structured covariance matrix. The analysis for each outcome consisted of a maximum likelihood growth curve model that included treatment, time, and treatment × time interaction as fixed effects and the intercept and slope as random effects with an unstructured covariance matrix. Since treatment groups are expected to be similar at baseline, the effect of treatment is captured through the treatment × time interaction (ie, differential temporal patterns of PTSD symptoms for 2 treatment groups). Between-group effect sizes were computed as Cohen d, the standardized mean difference,32 and defined as small (d = 0.25), medium (d = 0.50), and large (d = 0.80). We calculated the percentage of participants who showed clinically significant improvement on primary and secondary outcomes based on MCIDs established in the literature. Finally, because of the sex imbalance between the groups, we also conducted a series of exploratory growth curve models adding sex as a control variable to determine the effect of this imbalance on study findings.
Power analyses were based on expected effect sizes (0.5-0.9) drawn from a prior pilot study33 and estimates for means, standard deviations, and covariances from an unpublished open trial of mindfulness-based stress reduction in a group of 24 veterans with PTSD. Power analyses were conducted for the primary outcome (PCL score) using Nquery Advisor 4 (Statistics Solutions) under the following assumptions: (1) repeated-measures analysis of variance with main effects of treatment (mindfulness-based stress reduction vs present-centered group therapy), time (0, 3, 6, 9, and 17 weeks), and treatment × time interaction; (2) compound symmetric covariance matrix; and (3) α<.05 significance level. A sample size of 60 to 65 patients after attrition (we expected attrition of approximately 30%; hence, our initial recruiting goal was 90 per group) was estimated to provide 80% power to detect an effect size of d = 0.52. This effect size corresponds to a difference in total scores between groups of 6.4 for the PCL, 8.6 for the CAPS, 2.8 for the PHQ-9, and 7.6 for the WHOQOL-BREF. All analyses were conducted using SPSS software version 19.0 (IBM Corp). All tests were 2-tailed and α<.05 was considered statistically significant. No adjustment for multiple comparisons were made, so secondary outcomes, including tests of MCIDs, should be considered exploratory.
Figure 1 depicts participant flow through the study. Table 1 provides information on demographics, mental health treatment history, and trauma exposure for the sample. There were no differences between groups in demographic characteristics with the exception of ethnicity and sex. There were more African American (10.3% [n = 6] vs 5.2% [n = 3]) and American Indian (3.4% [n = 2] vs 0%) patients in the present-centered therapy group, but there were more patients of mixed ethnicity (10.3% [n = 6] vs 0%) and more women (20.7% [n = 12] vs 10.3% [n = 6]) in the mindfulness-based stress reduction group. Groups were similar in terms of comorbid mood disorder, psychoactive medication use, mental health treatment history, number of traumatic events, combat exposure, and ratings of treatment credibility and expectancies. However, participants randomized to mindfulness-based stress reduction therapy more frequently reported a history of sexual trauma (37% [n = 21] vs 19% [n = 11]) and had greater severity of PTSD symptoms at baseline as measured by the PCL (mean score, 63.6 vs 58.8) and CAPS (mean score, 69.9 vs 62.5) compared with participants randomized to present-centered group therapy. There was no difference in PTSD symptom severity as measured by the PCL between those who met full PTSD and those who met subthreshold PTSD (mean score, 61.2 vs 60.7). There were no differences between groups in depression symptoms (PHQ-9 mean score, 15.5 vs 14.6), quality of life (WHOQOL-BREF mean score, 75.6 vs 76.4), or mindfulness skills (FFMQ mean score, 105.7 vs 108.1) reported at baseline.
All participants provided written informed consent prior to enrollment in the study, which was approved by the University of Pennsylvania institutional review board. All protocols and blinded data were reviewed at intervals by a data and safety monitoring board.
The intervention group received a 20- to 30-minute manual-guided motivational intervention, delivered by master’s-level therapists during the ED visit and a telephone booster at 10 days. The treatment training manual was modeled on brief ED interventions targeting drinking and risky driving17– 19 and used motivational interviewing20 and empowerment theoretical frameworks21 (eMethods in Supplement 2). Extensively pilot tested and revised after input from female patients and experts in IPV,22 drinking among women,23 and motivational interviewing,24 the intervention formalizes measurable components of reflective feedback, avoidance of confrontation, respect, empathy, and empowerment, which are all consistent with the principles of IPV advocacy.21
The goal of the intervention was to elicit the patient’s self-identified reasons for change and personal goals.20,24 During the intervention sessions, which were recorded and analyzed for fidelity, the therapists encouraged participants to identify any linkages between their drinking and IPV, and helped them to resolve ambivalence regarding behavioral change, while also supporting the patient’s autonomy and personal choice.20,24 An unrecorded telephone follow-up booster with the same therapist occurred approximately 10 days later to consolidate and reinforce the motivational interviewing session.17
All therapists had training and experience working with abused women before the initiation of this study. They received extensive additional training and supervision by motivational interviewing experts through biweekly review of recorded sessions. Study participants had the option not to have their brief interventions recorded.
An independent motivational interviewing consultant reviewed and rated 10% of randomly sampled recordings throughout the study with ongoing feedback to the therapists. At study completion, 203 recordings (85%) were available for quality ratings by the developer of the motivational interviewing adherence scale.25 Three raters were used and a high intraclass correlation coefficient (ICC) was obtained for adherence scores (ICC = 0.86); there was 98.4% adherence to motivational interviewing techniques.25
All demographic, baseline, and outcome measures were collected from participants via self-report (Table 1). To describe our patient population, race/ethnicity was collected using fixed categories with the option for participants to specify other or multiple races. Primary outcomes were assessed weekly for 12 weeks using an the interactive voice response system (IVRS) among the brief intervention group and the assessed control group. The primary drinking outcome was days of heavy alcohol consumption (>4 drinks/day)29during the past week from the Alcohol Use Disorders Identification Test alcohol consumption30 (AUDIT-C) questions. The primary IPV outcome was any past week verbal, physical, or sexual abuse (8 items from the Revised Conflict Tactics Scales [CTS2S]).31
Secondary outcomes were assessed at 3, 6, and 12 months for the brief intervention group and the assessed control group but only once at 3 months for the no-contact control group. Secondary drinking outcomes included changes in the full AUDIT28 score and both quantity and frequency of drinking using the Timeline Follow-Back technique.32 Secondary IPV outcomes included frequency and severity of experiencing or perpetrating IPV (16 items) from the full CTS2S31 and changes in the Composite Abuse Scale.33 Other health-related outcomes included measures of self-rated health, depression, sleep, social support, quality of life, satisfaction with relationship, and engagement with treatment.16 All non-IVRS postbaseline assessments were conducted by trained interviewers blinded to treatment assignment.
We used a modified intent-to-treat analysis that included all participants with complete enrollment and at least 1 postbaseline assessment, regardless of whether the participant received the intervention. The presence or absence of past week heavy drinking and experiencing IPV collected by the IVRS were the primary binary end points for comparing the intervention group and the assessed control group. The analyses of the primary longitudinal end points from the 12 weeks of IVRS assessments used a hierarchical generalized linear model to accommodate the covariance structure and the correlation among repeated measurements over time observed for each patient. Similarly, secondary outcome measures collected at baseline and at 3, 6, 9, and 12 months were compared (brief intervention group vs assessed control group) using the hierarchical generalized linear model. For outcomes demonstrating a large proportion of zeros, zero-inflated Poisson and zero-inflated negative binomial models were generated.
The mixed-effects models used a maximum likelihood estimation approach that assumed any missing outcome data to be missing at random (ie, missing data including those due to dropout can be dependent on any previously observed outcomes or treatment assignment). With this approach, all data that have been collected without regard to whether data are missing for a patient at another visit, including dropouts, and without explicit imputation of missing data, can be used. Analyses were generated to test whether missing data were related to key treatment or baseline demographic characteristics. Analyses of 3-month CTS2S and AUDIT data were performed to determine the assessment reactivity effect on the primary outcomes, with specific contrasts between the assessed control group and the no-contact control group.
A priori power analysis was based on a 2-sided .03 significance level to accommodate 2 primary outcomes: heavy drinking days and IPV incidents, with significance on either outcome providing evidence of a positive study. Based on prior studies,19 199 patients per group were required to achieve at least 80% power to detect 2 days of heavy drinking per month. Without prior effect sizes for IPV interventions, we designed the study to have 94% power to detect a 20% difference between groups in the incidence rate of IPV (assuming 30% attrition) during the first 12 weeks of treatment.
From January 2011 to November 2013, there were a total of 112 167 ED visits by 59 326 unique female patients between the ages of 18 and 64 years (Figure 1). Approximately 53.4% of 59 924 total visits by 38 255 unique patients occurred during data collection time frames (Monday-Saturday, 9 am-7 pm). Of these, 28 813 unique patients were approached and 6768 were assessed for study eligibility.
Based on positive screenings for past 3-month IPV (CTS2S score >1)31 and heavy drinking (National Institute on Alcohol Abuse and Alcoholism sex-specific criteria ≥4 drinks/day or AUDIT score >4),29 1245 women were eligible for the study. The main reason for ineligibility was failing to meet criteria for both risk factors. Of those eligible, 24% declined to participate and 27% were discharged before enrollment could occur; 48% (N = 600) signed written informed consent and were randomized (2:2:1) as planned into the 3 groups. Eight randomized participants were excluded due to incomplete enrollment without follow-up contact, leaving 592 (98.7%) for analysis with 239 in the brief intervention group, 232 in the assessed control group, and 121 in the no-contact control group.
Compared with women who did not enroll, eligible women who enrolled were older, more likely of black race, and had higher IPV and drinking severity scores (eTable 1 in Supplement 2). Overall attrition rates and missing data did not vary by group at any time point; 78% of participants completed the 3-month interview, 79% completed the 6-month interview, and 71% completed the 12-month interview without differential attrition by severity of IPV or drinking. Participant safety was carefully tracked; no harms related to the intervention were identified.
Baseline demographics and risk factors were fairly well balanced across the groups (Table 1). Exceptions were slightly more white patients in the intervention group, the no-contact control group had higher rates of IPV at baseline, and more women in the assessed control group had previously used community-based IPV services compared with the intervention group (10% vs 4%, respectively).
Black women comprised 80% of the study, the mean age was 32 years, most were in relationships (88%) but unmarried (86%), 71% presented to the ED for medical concerns, 14.5% were pregnant, approximately half had children younger than 18 years, were employed, and living in households with incomes of less than $20 000. Comorbid risk factors were prominent: 60% smoked cigarettes, 17% disclosed using illicit drugs (46% including marijuana), 12% were positive for nonmedical use of prescription drugs, 43% disclosed a history of child sexual abuse, 40% screened positive for posttraumatic stress disorder, and 86% screened positive for depression. Baseline levels of alcohol dependence were 16% (AUDIT score >13)28 and mean IPV levels were severe on the Women’s Experience with Battering Scale (score range: 10-40; ≥20 indicates higher severity)26 and Danger Assessment (score range: 0-39; ≥18 indicates extreme danger).27
The primary outcomes of experiencing any IPV and any days of heavy drinking during the past week appear in Figure 2. This was tracked for 12 weeks after the ED visit only for women in the intervention group and the assessed control group; 89% of participants completed at least 2 of the weekly IVRS calls and 71% completed 50% or more. During the 12-week period following the brief motivational intervention, there were no significant differences between the intervention group and the assessed control group on weekly assessments for experiencing IPV (odds ratio [OR], 1.02; 95% CI, 0.98-1.06) or heavy drinking (OR, 0.99; 95% CI, 0.96-1.03) (eTable 2 in Supplement 2).
The incidence of past week heavy drinking at baseline (week 1) was 51% (95% CI, 44%-57%) in the intervention group (120 of 236 women) and 46% (95% CI, 40%-53%) in the assessed control group (107 of 231 women), which decreased by week 12 to 43% (95% CI, 36%-50%) in the intervention group (83 of 194 women) and 41% (95% CI, 34%-48%) in the assessed control group (77 of 187 women); however, the change was not statistically significant (P = .74 for treatment group × time interaction). The baseline incidence of past week IPV was 57% (95% CI, 50%-63%) in the intervention group (134 of 237 women) and 63% (95% CI, 57%-69%) in the assessed control group (145 of 231 women), which decreased by week 12 to 43% (95% CI, 36%-50%) in the intervention group (83 of 194 women) and 41% (95% CI, 34%-48%) in the assessed control group (77 of 187 women) (P = .33 for treatment group × time interaction).
The test for assessment reactivity (assessed control group vs no-contact control group) and the comparison of IPV and drinking outcomes (intervention group vs each of the control groups at 3 months) appear inTable 2. Controlling for baseline differences, experiencing abuse during the last 3 months was significantly reduced for the no-contact control group compared with the assessed control group, indicating that the assessment did not serve as an intervention. However, there were no significant differences in IPV and drinking outcomes between the intervention group and either control group at 3 months.
aReasons for loss to follow-up are unknown.
From March 2012 to December 2013, 603 veterans were screened and 167 completed clinical interviews to assess eligibility; 118 were eligible and 116 were randomly assigned to mindfulness-based stress reduction (n = 58) or present-centered group therapy (n = 58). For each cohort, there was an average of 6.4 (range, 4-11) veterans in each group. Recruitment was terminated prior to reaching the target recruitment goal of 90 per condition prior to attrition because of lower-than-anticipated attrition rates and lack of sufficient funds. Of the 116 participants randomized, 99 (85.3%) completed treatment, defined as receiving at least 7 of the possible 9 treatment sessions. The mean number of sessions attended was 6.96 (SD, 2.56) in mindfulness-based stress reduction and 8.08 (SD, 1.84) in present-centered group therapy (P=.008). Treatment dropout was higher in mindfulness-based stress reduction (22.4% [n = 13]) than in present-centered group therapy (6.9% [n = 4]; χ2 = 5.58; P = .02). There were no differences in demographic or baseline clinical characteristics between patients who completed treatment and those who dropped out. There was 1 serious adverse event in present-centered group therapy, in which a patient made a suicide attempt. Adherence to prescribed elements of the treatment was 96.25% for mindfulness-based stress reduction and 100% for present-centered group therapy; use of proscribed elements was extremely low, with no use of proscribed elements in mindfulness-based stress reduction. For our primary outcome, the PCL, we obtained 93.6% of all possible assessments. We obtained 93.4% of all possible interviewer-rated PTSD assessments and 93.5% of all possible secondary self-report outcomes. We found no evidence that missing data were due to any substantial demographic or clinical characteristics.
Figure 2 presents mean PCL scores from baseline to 2-month follow-up. There was an initial increase in PTSD symptom severity in both groups; however, this worsening of PTSD symptoms from baseline (mean score, 58.8) to week 3 (mean score, 61.7; t55.8 = 8.4; P = .005) was significant for present-centered group therapy participants only. There was a significant group × time interaction in relation to PTSD symptom severity (F1,106.9 = 8.78; P = .004). Table 2 shows that mean PCL scores improved from baseline to 2-month follow-up for both groups (mindfulness-based stress reduction, from 63.6 to 54.4; present-centered group therapy, from 58.8 to 56.0). However, growth curve mixed-effects models showed that improvement in mindfulness-based stress reduction was significantly greater than improvement in present-centered group therapy (mean difference in improvement, 6.44; 95% CI, 3.34-9.53; t = 4.08; P < .001; d = 0.40).
All participants also showed significant improvements in interview-rated PTSD severity on the CAPS from baseline to 2-month follow-up. However, a significant group × time interaction showed that improvement was greater in mindfulness-based stress reduction than in present-centered group therapy (F1,106.8 = 4.75; P = .03; mean CAPS score improvement: mindfulness-based stress reduction, from 69.9 to 49.8 vs present-centered group therapy, from 62.5 to 50.6; mean difference in improvement, 7.89; 95% CI, 2.58-10.6; t = 2.91; P=.004; d = 0.41). There was no difference between mindfulness-based stress reduction and present-centered group therapy in rates of loss of diagnosis at posttreatment (42.3% [n = 22] vs 43.9% [n = 25]; mean difference, 1.6%; 95% CI, −20.6% to 17.4%; χ2 = 0.03; P = .87) or at 2-month follow-up (53.3% [n = 24] vs 47.3% [n = 26]; mean difference, 6.0%; 95% CI, −14.1% to 26.2%; χ2 = 0.36; P = .55).
Mindfulness-based stress reduction participants reported greater improvement in quality of life on the WHOQOL-BREF from baseline to 2-month follow-up than did those in present-centered group therapy (mean score improvement: mindfulness-based stress reduction, from 75.6 to 80.2 vs present-centered group therapy, from 76.4 to 75.8; mean difference in improvement, 5.22; 95% CI, 1.73-8.71; t = 2.94;P=.004; d = 0.41). While mindfulness-based stress reduction participants reported greater improvement in depressive symptoms on the PHQ-9 from baseline to 2-month follow-up than did those in present-centered group therapy, this differential did not reach the level of significance (mean score improvement: mindfulness-based stress reduction, from 15.5 to 13.3 vs present-centered group therapy, from 14.6 to 13.8; mean difference in improvement, 1.34; 95% CI, −0.07 to 2.75; t = 1.87; P = .06; d = 0.26).
A significant group × time interaction was observed in relation to mindfulness skills from baseline to 2-month follow-up (F1,118.5 = 12.7; P = .001), with participants receiving mindfulness-based stress reduction therapy reporting greater improvement in mindfulness as measured by the FFMQ (mean score improvement, mindfulness-based stress reduction, from 105.7 to 116.3 vs present-centered group therapy, from 108.1 to 108.6; mean difference in improvement, 9.73; 95% CI, 4.42-15.04; t = 3.59; P < .001; d = 0.36). Improvement in FFMQ scores from baseline to posttreatment for participants receiving mindfulness-based stress reduction therapy was associated with greater reductions in PTSD symptom severity (PCL, r = −0.46; CAPS, r = −0.33) and depressive symptoms (r = −0.44) as well as improved quality of life (r = 0.42) at 2-month follow-up.
Using a 10-point or greater reduction on the PCL as an indicator of MCID,23 we found that a greater percentage of participants receiving mindfulness-based stress reduction therapy showed clinically significant improvement in PTSD symptom severity at 2-month follow-up compared with participants receiving present-centered group therapy (48.9% [n = 23] vs 28.1% [n = 16]; mean difference, 20.9%; 95% CI, 2.2%-39.5%; χ2 = 4.79; P = .03). Using a 10-point or greater reduction on the CAPS, groups were similar in the percentage of participants showing clinically significant improvement in interview-rated PTSD symptom severity at 2-month follow-up (mindfulness-based stress reduction, 66.7% [n = 30] vs present-centered group therapy, 54.5% [n = 30]; mean difference, 12.1%; 95% CI, −7.5% to 31.7%; χ2 = 1.52; P = .22). Treatments showed similar percentages of participants reporting clinically significant improvement in depressive symptoms on the PHQ-9 (mindfulness-based stress reduction, 27.7% [n = 13] vs present-centered group therapy, 22.8% [n = 13]; mean difference, 4.9%; 95% CI, −12.2% to 21.9%; χ2 = 0.32; P = .57).
Due to the imbalance in sex distribution at baseline, we conducted a series of growth curve analyses controlling for sex. For our primary outcome, we continued to find a significant group × time interaction (F1,106.7 = 8.80; P = .004). For secondary outcomes, the group × time interaction for interview-rated PTSD severity was also significant (F1,106.9 = 4.81; P = .03), as were the interactions for the WHOQOL-BREF (F1,109.4 = 4.99; P = .03) and FFMQ (F1,108.5 = 12.79; P = .001). The group × time interaction for the PHQ-9 did not meet significance (F1,107.8 = 2.22; P = .14), as was the case before controlling for sex.
Findings from the present study provide support for the efficacy of mindfulness-based stress reduction for the treatment of PTSD among veterans. Participants randomized to receive mindfulness-based stress reduction therapy showed greater improvement in self-reported PTSD symptom severity during treatment than those randomized to receive present-centered group therapy. However, the magnitude of the average improvement suggests a modest effect. Results of this study also support the modestly increased efficacy of mindfulness-based stress reduction therapy through the 2-month follow-up, with participants randomized to mindfulness-based stress reduction therapy showing greater improvement in both self-reported and interview-rated PTSD severity than those randomized to present-centered group therapy. Participants receiving mindfulness-based stress reduction therapy appeared to demonstrate improvements in depressive symptoms and quality of life after treatment, but these improvements were not observed among those who received present-centered group therapy.
These findings are consistent with previous studies demonstrating robust changes in anxiety and depressive symptoms between pretreatment and posttreatment.34 However, pilot studies evaluating mindfulness-based stress reduction as a treatment for PTSD have shown mixed results. While promising, previous studies had methodological shortcomings, including small sample sizes,33,35,36 pre-post designs,33,37 and lack of blinding of outcome assessments33,35– 37 and evaluation of treatment fidelity,33,35– 37 which preclude clear conclusions regarding efficacy.13 Few previous studies have evaluated the efficacy of mindfulness-based stress reduction relative to active treatment controls.36 Our findings add to the literature by demonstrating the comparative efficacy of mindfulness-based stress reduction for improving PTSD as well as possibly improving depressive symptoms and quality of life. Sustained improvements in PTSD symptoms observed at 2-month follow-up in the current study are consistent with pilot findings reported by Kearney et al37 but challenge those of Kearney et al35 showing no effect on PTSD and those of Niles et al36 showing clinically significant albeit temporary improvements in PTSD symptoms following mindfulness-based stress reduction therapy.
The findings of the present study are consistent with previous studies showing that mindfulness-based stress reduction is well tolerated by trauma survivors and veterans with PTSD.33,35,36 Although treatment dropout was higher among those randomized to mindfulness-based stress reduction therapy than to present-centered group therapy (22.4% vs 6.9%), dropout rates were lower than those reported for veterans receiving prolonged exposure (28.1%-44%)5,8,9,11 and cognitive processing therapy (26.8%-35%)6,7,19 in clinical trials and real-world treatment settings.
Half to two-thirds of patients who received mindfulness-based stress reduction therapy showed clinically meaningful treatment benefits at 2-month follow-up (48.9% based on the PCL; 66.7% based on the CAPS). These clinical response rates are similar to those reported for first-line PTSD treatments with veterans (49%-68.8%).5,6,11,19 Yet half of participants in both groups continued to meet criteria for PTSD diagnosis at 2-month follow-up.
Findings from the present study suggest that veterans who received mindfulness-based stress reduction therapy reported significant improvement in mindfulness skills after treatment, while there appeared to be little change in mindfulness skills reported by veterans who received present-centered group therapy. Moreover, findings suggest that greater reductions in PTSD symptom severity were associated with changes in mindfulness over the course of treatment. Improvements in quality of life made during treatment appeared to be maintained through the 2-month follow-up for participants receiving mindfulness-based stress reduction therapy, but reports of quality of life appeared to return to baseline levels for present-centered group therapy participants during this same follow-up period. Taken together, these findings suggest that mindfulness-based stress reduction may provide veterans with internal tools for promoting self-management of PTSD symptoms and quality of life.
The quality of scientific evidence supporting the efficacy of mindfulness-based interventions has recently been criticized.13 This study improves on shortcomings of previous trials by comparing mindfulness-based stress reduction with an active, credible control condition, taking steps to ensure treatment fidelity, and using both patient-reported and blinded clinician ratings of PTSD outcomes.13 Although groups were structurally equivalent in number of weekly sessions, therapist training and qualifications, and group format, present-centered group therapy may not have fully accounted for all nonspecific factors present in mindfulness-based stress reduction (eg, therapist expectations) and was unequal in duration of sessions. Because our intent was to study mindfulness-based stress reduction in the format it is typically taught (2.5-hour group sessions) compared with present-centered group therapy as generally provided within the VA system (1.5-hour group sessions), participants received less contact with clinicians in the control condition (13.5 hours) than in mindfulness-based stress reduction (26.5 hours). Another important limitation was the short follow-up period. Given the chronicity of the study group in terms of treatment history and modest average treatment effects observed in this study, it is possible that some participants may have relapsed after the 2-month follow-up. Future trials with longer-term follow-up (≥6 months) are needed to evaluate the durability of treatment benefits over time. Despite randomization and inclusion criteria requiring diagnosis of PTSD or subthreshold PTSD, the 2 groups differed in baseline PTSD symptom severity, with present-centered group therapy participants reporting lower symptoms. This limitation may have influenced results. Current findings are also limited by the sample, predominantly white men from 1 geographical region who served during the Vietnam era, and results may not generalize to nonveterans or veterans from other eras or areas. Replication with more diverse samples at additional centers is needed.
Among veterans with PTSD, mindfulness-based stress reduction therapy, compared with present-centered group therapy, resulted in a greater decrease in PTSD symptom severity. However, the magnitude of the average improvement suggests a modest effect.