The role of reduced sense of isolation in group-format PTSD treatment

Abstract

With increasing application of group-based interventions for PTSD, more research is needed on what might contribute to the therapeutic effects of group-based interventions. This study conducted secondary analyses with data from a previous randomized controlled trial, which compared two manualized 16-week videoconferencing PTSD treatments in a hybrid format with 16 group sessions and 8 individual sessions. One treatment arm focused on cultivating inner compassion and understanding internal conflicts through a contemplative Internal Family Systems (IFS) approach, while the other treatment arm focused on watching nature videos and discussing nature’s potential for stress reduction. The original RCT showed comparable effects on PTSD symptom reduction. Therefore, we conducted secondary analysis to elucidate whether both two group-based interventions, despite very different curriculum content, might have induced “reduced sense of isolation”, which might be a shared therapeutic element in both treatment arms. Findings from Linear Mixed Effects model analysis and Variance Decomposition Analyses (VDA) showed significant reduction of sense of isolation at the end of both interventions, which was one of the major contributors to reduction of PTSD symptoms. Path analyses showed similarities between the two group-based treatment arms in which reduced sense of isolation was associated with improvements in self-judgment and emotion regulation, all of which contributed to PTSD symptom reduction. These findings may inspire and inform future PTSD treatment development to place more emphasis on reducing patients’ sense of isolation.

Post-traumatic stress disorder (PTSD) is a psychological disorder in response to traumatic experiences¹. Current evidence-based psychotherapy techniques for PTSD treatment, such as prolonged exposure therapy² or cognitive processing therapy³, primarily take place in individual therapy settings. Yet, community mental health centers that serve a large number of underserved populations with high PTSD prevalence often struggle with staff shortage for providing individual therapies⁴. For the purposes of cost effectiveness, accessibility, and increasing patient engagement, especially in community mental health centers, group-based PTSD treatment options are increasingly being developed and utilized. However, a recent meta-analysis of k = 16 randomized controlled trials (RCTs) of total n = 1686 participants demonstrated that group treatments for PTSD were only superior compared to waitlist control (d = 0.56) but not significantly different from other active group-based treatments such as psychoeducation or supportive counseling (d = 0.09)^5,6. Such an intriguing phenomenon points to the possibility of intrinsic therapeutic effects from being part of a treatment group. Treatment groups can provide a safe environment to cope with social functioning challenges such as social isolation, difficulty with trusting others, or feeling misunderstood, which are very common among patients with PTSD⁷.

Despite ongoing innovations with group-based PTSD treatments, such as group-based Skills Training in Affective and Interpersonal Regulation (STAIR)^8,9, group-based mindfulness-based interventions¹⁰, group therapy in primary care settings^11,12, as well as the “seeking safety”^13,14 program for PTSD with comorbid substance use disorder, most research on these programs are focused on the content of the curricula, while there is still very limited understanding on the potential contribution of the “group format”. Yalom et al. have conducted decades of research on the theories and practices of group psychotherapy^15,16 that have identified multiple therapeutic factors including e.g., catharsis, universality, altruism, instillation of hope, practicing social skills, group cohesion and imitative behaviors, which has been summarized in a recent book by Yalom and Leszcz¹⁷. Nevertheless, quantitative research is still lacking, with prior studies mostly focused on factors such as group discussion topics and group leaders’ styles and efforts¹⁸, but little is known regarding the effects of being part of a therapy group on participants’ internal social cognitions, such as reducing the sense of isolation and tendencies for self-judgment.

In Buddhist philosophy, “suffering” is considered a common human condition that can happen to anyone at any time, because the fundamental cause of suffering is the desire of the human mind to have control over the constantly changing external world and to avoid dissatisfaction¹⁹. From this perspective, when people do not understand this principle, they operate under the illusory assumption that most other people are “happy” or “free of suffering” most of the time. Therefore, suffering is considered as an “abnormal” experience that seemingly only occasionally happens to the unfortunate few. This can lead to judgmental inferences that there is something “wrong” with the person that is suffering. This kind of illusion is in fundamental denial of the universality of suffering as a common human experience and an intrinsic element of human life²⁰. Judith Herman’s work in the 1990s has pointed out that trauma denial is a combination of individual survivors’ defense mechanisms and societal pressure of silencing and invalidating trauma narratives, creating a vicious cycle of denial and re-traumatization²¹ that hinders trauma healing²². Therefore, we theorize that group-based PTSD treatments provide opportunities and social contexts to break this vicious cycle by enabling patients to feel they are not suffering alone, to reduce the emotional impact of self-judgment, and to normalize emotional reactions to trauma experiences, through recognizing the commonality of PTSD symptoms via witnessing and interacting with others in the treatment group.

To quantitatively investigate how group-based treatments contribute to PTSD symptom reduction, this study conducted secondary data analyses of psychological measures from a recent Randomized Controlled Trial (RCT) that compared two group-based therapeutic programs for PTSD with distinct curriculum content and theoretical background, with one focused on contemplative techniques for emotion regulation, inner compassion, and reconceptualizing internal conflicts^23,24, and the other focused on watching nature videos and understanding the role of nature in stress reduction²⁵. The RCT took place during the Covid-19 pandemic, and all intervention sessions were delivered through online synchronous videoconferencing. Recent publication from the original RCT showed both two treatment arms had significant reduction of PTSD symptoms without significant group difference. Therefore, the primary objective of the present secondary data analysis study was to elucidate how reduced sense of isolation might be a potential universal contributor for therapeutic effects of the two treatment arms despite very different curricula content. Linear Mixed Effects (LME) model analysis, Variance Decomposition Analysis (VDA), and hypothesis-driven path analysis were utilized to investigate data from two distinct group-based PTSD treatments.

In the present study, subjective sense of isolation was measured by the “isolation” subscale of the Self-Compassion Scale (SCS)²⁶-Short Form²⁷, which reflects the tendency for seeing one’s suffering as alone and separate instead of seeing suffering as common human experiences shared by many other people. We hypothesized that participation in group-based interventions reduced the sense of isolation, which might have contributed to improvement in self-judgment and emotion regulation, all of which could contribute to PTSD symptom reduction. In the present study, the tendency for self-judgment was measured by the “self-judgment” subscale of Self-Compassion Scale (SCS), which reflects internalized social cognition of how one sees themselves in terms of their relationships with the external world, e.g., judging oneself for not meeting societal expectations, shame, or self-criticism, which are all common symptoms among trauma survivors²⁸. Emotion regulation capacity was measured by the Difficulty in Emotion Regulation Scale (DERS)²⁹, and PTSD symptom severity was measured by PTSD Checklist for DSM-5 (PCL-5)³⁰. The overarching hypothesis was that group-based treatments could reduce perceived sense of isolation, enabling patients to reduce their tendency for judging themselves as broken or flawed for having PTSD, which eventually enhanced their emotion regulation capacity to effectively cope with PTSD symptoms.

The original RCT was conducted in a community mental health center setting that was facing staff shortage for providing individual therapies for the large volume of patients with PTSD⁴, therefore, there was a strong need for cost-effective group-based interventions. The RCT was conducted via videoconferencing during the period of Covid-19 pandemic. Social isolation during the pandemic³¹ has been shown to have detrimental effects on a wide range of mental health conditions^32,33. The purpose of the original RCT was to test the feasibility and obtain preliminary data on the therapeutic effects of adapting a group format Internal Family Systems (IFS) psychotherapy³⁴, which was traditionally provided in individual therapy settings. An active control condition was developed to match with the target intervention in terms of format, time duration, attention, and the wide range of affect and arousal levels evoked by experiential practices. The active control intervention consisted of watching nature videos together as a group and discussing how nature could help with stress reduction and trauma healing, and purposefully did not contain any elements from IFS nor any teaching of Buddhism or mindfulness.

Although the primary hypothesis of the original RCT was that the IFS-based intervention program would have superior effects in reducing PTSD symptoms compared to the active control intervention^24,25,35, both treatment arms showed significant PTSD symptom reduction without significant group differences³⁶. Therefore, the current secondary data analyses research focused on investigating potential shared factors that might be driving the therapeutic effects of both treatment programs, e.g., how participating in group-based interventions might reduce the sense of isolation, and how reduced sense of isolation could contribute to reduced PTSD symptoms.

Materials and methods

Study design

This study conducted secondary analyses of psychological variables from a prior RCT that compared two online group-based psychotherapy programs for PTSD: Program for Alleviating and Resolving Trauma and Stress (PARTS)^24,35,36 vs. Nature-Based Stress Reduction for Trauma (NBSR-T) which was developed as an active control in this RCT based on a previous NBSR program²⁵. The original RCT was registered on ClinicalTrials.gov (NCT05095428) and was approved by the Institutional Review Board (IRB) of Cambridge Health Alliance (CHA) on 2022-04-15 with IRB approval #CHA-IRB-21-22-18. All research procedures were performed during 2022-04-28 to 2023-09-12 in accordance with the guidelines and regulations of the Declaration of Helsinki. All research procedures were conducted according to IRB approved study protocol; any changes after the initial approval were IRB-approved again. All research participants provided written informed consent with IRB-approved consent form according to IRB-approved consenting procedures. The IRB approval remains active for secondary data analyses and is renewed annually, all analysis in this manuscript was permitted in the original informed consent.

Participants

Research participants were recruited from the CHA community outpatient psychiatry clinics through digital advertising (e.g., webpages and email lists) or provider referrals, for “a program for stress reduction and alleviating trauma in patients with PTSD”. Eligibility criteria included: (1) age 18–75 years; (2) have a current diagnosis of PTSD or a score ≥ 58 from the PTSD module of Computerized Adaptive Testing for Mental Health (CAT-MH)³⁷ and having active PTSD symptoms prior to randomization with PCL-5 score ≥ 31³⁸ and CAPS-5³⁹ score ≥ 26; (3) having reliable access and the ability to use electronic devices to attend all therapy sessions and complete online questionnaires. Exclusion criteria included active psychosis (CAT-MH psychosis score ≥ 60⁴⁰), history of bipolar I disorder or active mania (CAT-MH mania score ≥ 70⁴⁰), severe depression (CAT-MH depression score ≥ 75⁴⁰), acute suicidality or self-injurious behavior, acute homicidality with plan and/or intent, severe borderline personality disorder, or active substance use disorder with use of illicit drugs within the past 3 months. CAT-MH and PCL-5 were administered online, and CAPS-5 were administered via telephone-based clinician interview to verify PTSD diagnoses. Furthermore, every prospective participant met with a clinician for a psychiatric intake evaluation and the clinician also conducted an electronic health record review to make a final determination regarding the eligibility of each prospective participant.

The original RCT randomized N = 60 eligible patients into two treatment arms using 1:1 ratio with block size = 4³⁶. Data from all 60 patients were analyzed in this secondary data analysis study. Fisher’s exact tests showed that the patient samples of the two treatment arms had no significant differences in distribution of gender, race, ethnicity, and education levels (Table 1).

Table 1.

Demographic information of research participants in both treatment arms. P values were from Fisher’s exact tests for comparing categorical ratios between PARTS and NBSR-T, demonstrating that the two treatment arms had no significant differences in patient sample distribution of gender, race, ethnicity, and education levels.

	PARTS (n = 30)	NBSR-T (n = 30)	Total (n = 60)	p-value
Female, N (%)	23 (76.7%)	22 (73.3%)	45 (75.0%)	0.77
Race, N (%)				0.32
American Indian or Alaska Native	1 (3.3%)	0 (0%)	1 (1.7%)
Asian	4 (13.3%)	4 (13.3%)	8 (13.3%)
Black, Haitian, or African American	4 (13.3%)	5 (16.7%)	9 (15.0%)
Prefer not to say	2 (6.7%)	4 (13.3%)	6 (10.0%)
Unreported	2 (6.7%)	2 (6.7%)	4 (6.7%)
White	17 (56.7%)	11 (36.7%)	28 (46.7%)
More than 1 category	0 (0%)	4 (13.3%)	4 (6.7%)
Ethnicity, N (%)				0.55
Hispanic, Spanish, or Latinx origin	7 (23.3%)	6 (20.0%)	13 (21.7%)
Not Hispanic, Spanish or Latinx	22 (73.3%)	21 (70.0%)	43 (71.7%)
Prefer not to say	1 (3.3%)	1 (3.3%)	2 (3.3%)
Unreported	0 (0%)	2 (6.7%)	2 (3.3%)
Highest Level of Education, N (%)				0.71
Some high school	1 (3.3%)	1 (3.3%)	2 (3.3%)
Completed high school or GED	2 (6.7%)	2 (6.7%)	4 (6.7%)
Technical and/or vocational training	1 (3.3%)	0 (0%)	1 (1.7%)
Some college or 2-year degree	11 (36.7%)	7 (23.3%)	18 (30.0%)
Completed 4-year degree	7 (23.3%)	13 (43.3%)	20 (33.3%)
Some graduate school	1 (3.3%)	1 (3.3%)	2 (3.3%)
Completed graduate school	7 (23.3%)	6 (20.0%)	13 (21.7%)

Interventions

All intervention sessions were held via secure, HIPAA-compliant Zoom videoconference. Both intervention programs included 16 weekly 90-minute group sessions, as well as 8 biweekly 55-minute individual counseling sessions. The purpose of the individual sessions was to ensure psychological safety, to resolve ambivalence about group sessions, or to address any barriers to participating in group activities.

The PARTS program was derived from the IFS psychotherapy framework³⁴ and has been shown to significantly reduce PTSD symptoms²⁴. The PARTS program focused on cultivating meta-awareness of internal conflicts and applying compassion towards the self, using experiential exercises such as contemplative guided meditation, body awareness, dramatic play, insight inquiries, and group discussions. Topics of the 16-week curriculum included: (weeks 1–2) introducing the conceptual framework of IFS and its treatment approach for psychological trauma; (week 3) acknowledging the core “Self” with “curious, clear and compassionate” qualities; (weeks 4–8) recognizing internal conflicts as “parts”, understanding how “parts” are separate from the core “Self”, and learning to “un-blend” from the “parts”; (weeks 9–11) recognizing defense mechanisms as “protectors” of the internal system and recognizing the role of “protectors” in the “un-blending” process; (weeks 12–14) recognizing the “parts” that carry the burden of past trauma as “exiles” of the internal system, and learning how to access and help them; (weeks 15–16) making plans for continued healing beyond the program. “Center practices” were frequently utilized in group sessions in which one volunteer participant verbalized their internal experiences as they went through a particular therapeutic process while the rest of participants watched and later debriefed and discussed with the whole group.

The NBSR curriculum was originally designed as a control program for Mindfulness-Based Stress Reduction (MBSR)²⁵, and NBSR-T was further developed in this RCT from the original NBSR curriculum to match the duration, format, attention, valence and arousal levels of the 16-week PARTS curriculum. The NBSR-T curriculum focused on watching nature videos of various valence (pleasant, neutral, unpleasant), with didactic teaching on the health benefits of connecting to nature, as well as guided reflective group discussions. The NBSR-T program purposefully avoided instructions related to IFS, Buddhism, breathing techniques, and mindfulness. Nature videos covered a wide range of nature themes such as grassland, mountain, ocean, forest, and savannah, with scenes of natural scenery as well as various animal life clips. Weekly themes of curriculum included: (weeks 1–2) program orientation and introduction of basic concepts and components; (weeks 3–4) appreciating the sounds and sights of nature and its qualities; (weeks 5–6) understanding the connection between the natural world and the human mind; (weeks 7–8) confronting fears of nature; (weeks 9–10) understanding how exposure to nature can improve health; (weeks 11–12) reflecting on human relationships with the natural world; (weeks 13–14) caring about our natural world; (weeks 15–16) setting goals and plans for activities to connect with nature. NBSR-T was developed based on the previous NBSR program²⁵ by integrating trauma-specific curriculum content into the first 4 sessions: The first session introduced NBSR-T as an intervention for helping people cope with trauma, and discussed creating psychological safety in group settings. The second session introduced the concept of “window of tolerance” in trauma treatment⁴¹. The third session focused on how nature can help heal trauma. The fourth session introduced the 8 Fs (8Fs: Free, Facing Fear, Flow, Feeling, Focus, Forward thinking, Fun, seeing the Forest) to demonstrate how nature can be a tool for coping with the impact of trauma. These 4 sessions set the stage for the trauma-specific focus in the NBSR-T groups. To match the “center practices” of the PARTS program, similar practices were adapted in NBSR-T in which one volunteer talked about what they were noticing as they watched nature videos together with the group and later debriefed and discussed with the whole group.

Self-report questionnaires

The following self-report questionnaires, collected at week 0, week 8, and week 16 of the 16-week interventions, and follow-up at week 24, were utilized in the current secondary data analysis study:

PTSD Checklist for DSM-5 (PCL-5): PCL-5 is a widely utilized self-report questionnaire for measuring PTSD symptoms according to DSM-5, with high internal consistency with Cronbach’s α = 0.94³⁰.

Self-Compassion Scale-Short form (SCS-SF): SCS-SF is a 12-item self-report questionnaire²⁷ for measuring self-compassion with 3 pairs of subscales: Self-Kindness vs. Self-Judgment, Mindfulness vs. Over-identification, Common-Humanity vs. Isolation. Factor analyses were conducted in the previous study that validated this instrument²⁷, which has been shown with good internal consistency (α = 0.87) and near-perfect correlations with the long form SCS (r ≥ 0.97)²⁷. Below are the items for each of the subscales:

Self-Kindness: “I try to be understanding and patient towards those aspects of my personality I don’t like”, “When I’m going through a very hard time, I give myself the caring and tenderness I need.” This subscale had α = 0.55²⁷.

Self-Judgment: “I’m disapproving and judgmental about my own flaws and inadequacies”, “I’m intolerant and impatient towards those aspects of my personality I don’t like.” This subscale had α = 0.81²⁷.

Common-Humanity: “I try to see my failings as part of the human condition”, “When I feel inadequate in some way, I try to remind myself that feelings of inadequacy are shared by most people.” This subscale had α = 0.60²⁷.

Isolation: “When I’m feeling down, I tend to feel like most other people are probably happier than I am.” “When I fail at something that’s important to me, I tend to feel alone in my failure.” This subscale had α = 0.77.

Mindfulness: “When something painful happens I try to take a balanced view of the situation”, “When something upsets me, I try to keep my emotions in balance.” This subscale had α = 0.64²⁷.

Over-identification: “When I fail at something important to me, I become consumed by feelings of inadequacy,” “When I’m feeling down, I tend to obsess and fixate on everything that’s wrong.” This subscale had α = 0.69²⁷.

Difficulty in Emotion Regulation Scale (DERS): DERS is a 36 item self-report questionnaire for assessing emotional awareness, ability to regulate emotion, and the impacts of emotion dysregulation²⁹. It was shown to have good internal consistency (α = 0.93).

Connectedness to Nature Scale (CNS): CNS is a 14-item self-report questionnaire for assessing emotional connection with nature, example items include “I often feel a sense of oneness with the natural world around me”, “When I think of my life, I imagine myself to be part of a larger cyclical process of.

living”. This scale has been shown to have adequate internal consistency (α = 0.75)⁴².

Statistical analysis

All data was analyzed with R statistical software. To correct for multiple comparisons, the “p.adjust” R function was utilized with the Benjamini-Hochberg (‘BH’) procedure for controlling False Discovery Rate⁴³, with statistical significance threshold set at alpha level of 0.05.

First, to study treatment effects on PCL-5, SCS, DERS and CNS, LME model analyses were conducted with the “nlme” R package, in which scores of PCL-5, SCS subscales, DERS and CNS were the dependent variables, and “intervention type” (PARTS vs. NBSR-T) and “time” (week # 0, 8, 16, 24), as independent variables with group by time interaction included in the model, with demographic variables as covariates including age, gender, race, educational level, as well as intervention cohorts. Variance for each group was separately accounted for in the model, the alphanumerical ID assigned to each subject was used to model random effect, and the “REML” method was chosen to maximize the restricted log-likelihood.

Then linear regression analysis was conducted for each treatment arm separately, in which PCL-5 was the dependent variable, with independent variables including scores of each SCS subscale, DERS and CNS, “time” (week # 0, 8, 16, 24), as well as covariates: age, gender, race, educational level and intervention cohorts. VDA was then conducted using the “relaimpo” R package, which estimates percentages of variances explained by each regressor while accounting for the correlations among regressors⁴⁴.

Path analysis was conducted for each treatment arm separately using maximum likelihood estimation with the “sem” R function in the “lavaan” package to test the significance of hypothesized path: Δ SCS-Isolation ◊ Δ SCS-Self-Judgment ◊ Δ DERS ◊ Δ PCL-5, with the following pre-determined model fit parameter thresholds: Comparative Fit Index (CFI) ≥ 0.95 and Standardized Root Mean Square Residual (SRMR) ≤ 0.08⁴⁵.

Results

LME analyses of PCL-5 showed significant effect of time (F = 20.24, p < 0.001). The isolation (F = 6.07, p < 0.01, Fig. 1A), self-judgment (F = 4.95, p < 0.05, Fig. 1B) and over-identification (F = 9.92, p < 0.01) subscales of SCS showed significant effects of time, but the rest of SCS subscales did not show significant effects of time. DERS (F = 12.74, p < 0.001, Fig. 1C) and CNS also only showed significant effect of time (F = 8.78, p < 0.01, Fig. 1D) There were no significant effects of intervention type, or time by intervention type interaction with any of the outcome measures in the present study.

Fig. 1.

Estimated marginal means of the scores of (A) SCS-SF subscale “isolation”, (B) SCS-SF subscale “self-judgment”, (C) DERS, and (D) CNS, across assessment time points throughout the 16-week interventions (week 0, 8 and 16) and at post-intervention follow-up at week 24. All scales had significant effect of time but no significant effect of treatment arm (PARTS vs. NBSR-T).

VDA analyses separately conducted for each treatment arm demonstrated that DERS and SCS-isolation explained the highest amounts of variance for PCL-5 (Fig. 2), with DERS explaining 13.6% in the PARTS treatment arm and 17.3% in NBSR-T, and with SCS-isolation explaining 10.0% in PARTS and 8.8% in NBSR-T (Fig. 2). The “self-judgment” subscale of SCS explained 7.6% in PARTS vs. 3.6% in NBSR-T, the “self-kindness” subscale explained 4.4% for PARTS vs. 0.9% in NBSR-T, and the “over-identification” subscale explained 4.1% variance for PARTS vs. 6.8% for NBSR-T. CNS and the rest of SCS subscales (common humanity, mindfulness) explained very small amounts of variance in PARTS (1.2%, 2.8%, 2.3%) and NBSR-T (0.6%, 0.6%, 1.6%).

Fig. 2.

Results from Variance Decomposition Analysis (VDA). Percent variance explained by each variable for the PARTS (above the horizontal line) and NBSR-T (below the line) treatment arms for predicting the temporal variations of PTSD symptoms across all time points are demonstrated in separate rectangles, with the size of the rectangles representing the relative proportions of variance explained by the temporal variations (Δ) of each variable in the PARTS treatment arm.

Path analyses separately conducted for each treatment arm showed that both treatment arms had a significant path (p < 0.001) of ΔSCS-Isolation◊ Δ SCS-Self-Judgment◊ Δ DERS◊ Δ PCL-5 (Fig. 3). There was also a significant path of Δ SCS-Isolation◊ Δ DERS◊ Δ PCL-5 (p < 0.001) for each treatment arm, as well as a direct path of Δ SCS-Isolation◊ Δ PCL-5 (p < 0.05). The path models for both treatment arms showed good model fit (CFI = 1.00, SRMR < 0.001). The path model of the PARTS treatment arm had Akaike Information Criterion (AIC) = 1891.07, Bayesian Information Criterion (BIC) = 1914.43, whereas the path model of the NBSR-T treatment arm had AIC = 1656.90, BIC = 1678.88, with Root Mean Square Error of Approximation (RMSEA) < 0.001 for both treatment arms.

Fig. 3.

Results from path analyses for therapeutic elements of the (A) PARTS and (B) NBSR-T treatment arms. Findings from the path analyses showed similar paths of reduced sense of isolation being associated with improvements in self-judgment and emotion regulation, with all of which contributing to reduced PTSD symptoms. The similar paths between the two treatment arms with very different curriculum content suggested reduced sense of isolation might play a fundamental therapeutic role in group PTSD treatments.

Discussion

Findings from this study demonstrated the importance of reduced sense of isolation in group-format PTSD psychotherapy treatments. In both treatment arms, reduction of sense of isolation was associated with improvements in self-judgment and emotion regulation, all of which were associated with reduced PTSD symptoms. Similar findings from the two treatment arms with very different curriculum content suggest reduced sense of isolation might play a fundamental role in group PTSD treatments.

Despite very different curriculum content, both PARTS and NBSR-T treatment arms showed significantly reduced sense of isolation by the end of the 16-week interventions (Fig. 1A). Both two treatment arms included 16 weekly group sessions, during which participants had the opportunity to share experiences, interact with other participants, and observe social interactions in the group setting. Both interventions included a “Center Practice” in which a volunteer participant verbalized their internal experiences while the rest of the group watched and later debriefed. For the PARTS group sessions, this kind of practice might be particularly helpful for group participants to realize they were not suffering alone, and that other people could also have moments of distressing experiences such as fear, worry, or self-judgment; this “center practice” could also cultivate feelings of hope when they witnessed healing and growth in other participants. The NBSR-T control program had “center practices” where one volunteer talked about what they were noticing as they watched nature videos with the group. Although this was not a typical psychotherapeutic process, watching nature videos that depicted suffering among animals could encourage participants adapting broadened perspectives, such as realizing “suffering” as a universal experience among all living beings; watching interactions among animals could also encourage participants putting their personal experiences with human social dynamics into a bigger perspective. Watching nature videos also provided an opportunity to feel connected to something larger than oneself, which is one of the documented therapeutic effects of nature-based therapy⁴⁶. Such broadened perspectives, especially after discussions with the group, could reduce feelings of isolation for patients in the NBSR-T treatment arm. Furthermore, both treatment arms had a hybrid format that included 16 weekly group sessions and 8 bi-weekly individual sessions, and the individual sessions might have also played an important role in addressing barriers for group participation. Therefore, participants from both treatment arms showed reduced sense of isolation by the end of the 16-week intervention, which could be the result of multiple factors that can be further explored in future research.

This secondary data analysis study utilized the SCS-SF subscale to quantify “isolation”. SCS²⁶ has been widely utilized for studying psychological health and emotional resilience, for which social wellbeing is an essential element^20,26. Although self-compassion is towards the “self”, the “isolation” subscale^26,27 can reflect internalized social cognitions, e.g., comparing oneself to others (“When I’m feeling down, I tend to feel like most other people are probably happier than I am”), or feeling “alone” in one’s failures (“When I fail at something that’s important to me, I tend to feel alone in my failure.”). In theory, group therapy creates opportunities for patients to interact with other patients with similar conditions (in this case PTSD), which can facilitate awareness that each patient is not alone in struggling with these conditions, and that symptoms experienced by one patient bear similarity with symptoms of many other patients in the group. Therefore, this secondary data analysis study hypothesized that reduced sense of “isolation”, as measured by the SCS-SF subscale, was one of the critical contributing factors for the therapeutic effects of the two interventions that had very different curriculum content but both relied on a group format. This hypothesis was supported by observation of significantly reduced SCS-isolation scores in both treatment arms at the end of the intervention (Fig. 1.A), and findings from both treatment arms that changes in sense of isolation was associated with improvement in self-judgment and emotion regulation that eventually contributed to reduction of PTSD symptoms (Fig. 3).

Prior research demonstrated group-based therapies had strong effects for enhancing treatment adherence and therapeutic alliance^47,48. In recent years, “peer support” has been increasingly utilized in mental health treatment^49,50. A meta-analysis of k = 18 RCTs with a total of N = 5,597 participants with severe mental illness (schizophrenia spectrum or bipolar disorder) showed peer support enhanced patients’ senses of hope and empowerment⁵¹. Although group-based treatments have been utilized for decades⁵² with a wide variety of formats and purposes, there has been very limited mechanistic understanding. Existing mechanistic research of group-based therapy has focused on external factors from providers’ perspectives, such as didactic vs. member-led format, group discussion topics, as well as the styles and effort levels of group leaders¹⁸. The current study is innovative for investigating the effects of group therapies on participants’ internal experiences such as changes in sense of isolation, which was significantly reduced in both treatment arms in the current study (Fig. 1A). Because NBSR-T was designed as an active control condition for PARTS with matched format but purposefully avoided overlaps in curriculum content, it is plausible the similar effects of reduced senses of isolation could be due to the group-based format in both treatment arms.

Self-judgment also has a strong social component. The psychological trait for self-judgment is usually considered to develop from internalized criticisms from others during childhood⁵³. Our recent study demonstrated a strong neurobiological foundation for reductions in self-judgment as a result of Mindful Self-Compassion training⁵⁴, from which patients with above median levels of childhood trauma, compared to the rest of the patients, demonstrated more post-intervention reduction in self-judgment and increase of self-compassion, indicating the importance of self-judgment reduction in trauma healing. The similar paths found in both group-based treatment arms of the present study suggests that reduced sense of isolation appears to have facilitated the reduction of self-judgment, which could have arisen from the strong efforts in both treatment arms to create a trauma sensitive and psychologically safe group environment, which not only reduced participants’ sense of isolation but also reduced the tendency to be overly self-critical. It is also possible that practicing compassion for others in group sessions could reduce self-judgment. A large longitudinal study followed n = 1,090 adults for up to 7.5 years and found strong intercorrelations between compassion for others and self-compassion, with both predicting lower levels of loneliness⁵⁵. The hypothesized path of reduced isolation contributing to reduced self-judgment, instead of reduced self-judgment contributing to reduced isolation, was based on the common elements in the two treatment arms being group-based psychotherapy, which provided opportunities to observe other participants and exchange personal experiences with other participants. The PARTS curriculum and pedagogy relied on “Center Practice” exercise, in which other participants observed a volunteer participant going through a particular therapeutic process; through such process, group participants could learn to be less self-judgmental through learning about similar struggles of other participants. The NBSR-T curriculum did not have explicit teaching for cultivating self-compassion or reducing self-judgment; nevertheless, NBSR-T can direct patients’ attention towards appreciating nature and understanding human connection with nature and away from the self and self-related cognitive and emotional processes, which bears similarity with distress tolerance skills in Dialectical Behavioral Therapy (DBT)⁵⁶. Therefore, reducing the sense of isolation may be a potentially universal fundamental therapeutic element in both treatment arms, either through observing the healing process of other participants, or through reducing self-focus and potentially feeling connected with something beyond oneself such as nature^46,57.

Emotion dysregulation is a fundamental symptom of PTSD that affects not only symptom severity but also treatment responses⁵⁸. Neuroimaging research has demonstrated that emotion dysregulation in PTSD is a result of neural dysfunction involving multiple brain regions such as exaggerated response in the amygdala and insula as well as deficiency with emotion appraisal and management by the anterior cingulate cortex and prefrontal cortex⁵⁹. Therefore, the present study included DERS scores as a variable in the path model. Findings from our path analyses (Fig. 3) showed that reduced sense of isolation both directly and indirectly benefited emotion regulation and PTSD symptom severity reduction, but reduced self-judgment did not directly contribute to PTSD symptom reduction, highlighting the relative importance of reduced sense of isolation. Consistent with the above findings from path analyses, results from VDA (Fig. 2) also demonstrated that score changes of DERS and SCS-isolation were the most prominent predictors of PCL-5 score changes in both treatment arms. Consistent with findings from path analysis, self-judgment explained more variance in the PARTS arm than the NBSR-T arm (Fig. 2). Other subscales of SCS-SF explained much less variance for PTSD symptom changes, supporting our hypotheses being focused on the isolation and self-judgment subscales.

Although NBSR-T was originally designed as an active control condition for PARTS, it demonstrated comparable effects in reducing PTSD symptoms, with reduced sense of isolation as a shared therapeutic element as PARTS. Prior research suggested potential beneficial effects of nature-based therapies for PTSD⁶⁰. One prior single arm study observed improved psychological wellbeing after white water rafting trips organized for military veterans and underserved youth⁶¹, and proposed the emotional experience of “awe” as the mechanism for potential healing effects⁶¹. The NBSR-T curriculum in our study facilitated nature-based experience through watching and discussing nature videos, which is much more feasible and cost-effective, and elicits a wide range of feelings from “awe” to “aww” and many feelings in between. CNS score changes after NBSR-T demonstrated increased senses of connectedness to nature (Fig. 1D), but temporal variations of CNS scores only contributed to a very small portion of variance (Fig. 2) for changes in PTSD symptoms; therefore CNS score change was not the primary mechanism for the therapeutic effects of NBSR-T. Findings from the present study suggests reduced sense of isolation might have been a major therapeutic element of NBSR-T, which could have been achieved through a wide range of factors such as group peer support⁴⁷, obtaining broader perspectives instead of focusing on oneself⁴⁶, or feeling connected with something larger than oneself⁴⁶.

Our study revealed that reduced sense of isolation could be a potential shared therapeutic element of group-based PTSD treatment programs. This secondary data analyses study had several limitations. (1) The original RCT was not designed for specifically studying group effects or social wellbeing. Furthermore, both interventions were in hybrid format with both group and individual sessions. To specifically study the social effects of group format, future studies should design to directly contrast group-format interventions with individual-based interventions and should include more measurements for quantifying internalized social cognition or external social functioning. (2) The PARTS and NBSR-T are relatively novel interventions in the field of PTSD treatment and the field of contemplative practices; future studies should investigate whether reduced sense of isolation and self-judgment are also key therapeutic elements for other well-established programs such as MBSR, which was also found to have comparable but not superior therapeutic effects compared to active control groups for trauma survivors^62–65. (3) The present study only demonstrated associations among psychological variables but could not make any causal inferences. The analyses in the present study did not model temporal sequence among psychological variables, which limited the ability to make any causal inference. According to Kazdin’s criteria for establishing mechanisms of therapeutic changes⁶⁶, temporal sequence and elimination of alternative “common” causes beyond modeled variables are necessary to establish causality, which were not demonstrated in this paper. Furthermore, the present study only had two active intervention arms without another non-interventional control group, and the present study also did not include another treatment arm with only individual-based therapy without any group elements, which limited the ability to infer whether reduced isolation is unique to group-based interventions. Due to all of the above limitations, this study is not intended for drawing causal conclusions, and extreme caution shall be applied when considering the inference potentials and generalizability of the research findings. (4) As a secondary data analyses study, we could only utilize variables included in the original RCT, in which the “isolation” subscale of SCS-SF was the only variable directly explicitly related to the effects of both group interventions; future research can be specifically designed to systematically quantitatively investigate the group therapeutic factors outlined by Yalom et al.^15,16, such as catharsis, universalism, altruism, instillation of hope, practicing social skills, group cohesion and imitative behaviors^15,16. Furthermore, only SCS-SF was used in this RCT, which was a brief version of the original SCS. The “isolation” and “self-judgment” subscales of SCS-SF had the highest internal consistency among all sub-scales²⁷, which were the two utilized in the present study based on strong hypotheses related to these two concepts. It is possible that other components of self-compassion, especially when measured by the long version of SCS²⁶, might also have important potential contributions for PTSD treatments. (5) As a secondary data analysis study of an RCT originally designed for testing feasibility and obtaining preliminary data on therapeutic effects, the present study had a small sample size; therefore, all findings from this study shall be interpreted with caution and require replication and validation in future studies with larger sample sizes. Despite these limitations, this study contributes to the psychotherapy literature by highlighting the potentially fundamental role of reduced sense of isolation in therapeutic effects of group-based interventions.

Author contributions

DJ conceptualized and conducted the secondary data analysis and drafted the manuscript. AC and ZSO managed the original clinical trial. HSR was a lead instructor for the PARTS program and AR was the lead instructor for the NBSR-T program. MS provided clinical supervision and co-developed the PARTS program. ZSO acquired funding for the original clinical trial and supervised the study. All authors reviewed the manuscript.

Data availability

This manuscript reports results from secondary analysis of a previously collected dataset, data availability is subject to restrictions with the original dataset.

Declarations

Competing interests

Zev Schuman-Olivier and Diane Joss report financial support for the research was provided by funding from the Foundation for Self Leadership awarded to Cambridge Health Alliance. Martha Sweezy reports financial support was provided by Routledge, Guilford and PESI, as well as private practice with IFS-based psychotherapy. Hanna Soumerai Rea reports financial support was provided by private practice with IFS-based psychotherapy.