A total of 175 women diagnosed with PTSD according to DSM-5 criteria [1] participated in this study. All PTSD diagnoses were specifically linked to experiences of sexual or physical abuse before the age of 18. Participants, aged 18 to 65 years, were recruited as part of a multicenter RCT (German Clinical Trials registration number DRKS00005578), titled “Treating Psychosocial and Neural Consequences of Childhood Interpersonal Violence in Adults.” The trial compared the efficacy of DBT-PTSD and CPT, with detailed treatment protocols published previously [12]. Participants received 1 year of treatment, comprising up to 45 weekly sessions, followed by 3 monthly booster sessions.

Inclusion criteria were a PTSD diagnosis related to childhood abuse and the presence of at least 3 DSM-5 BPD criteria, including affective instability (criterion 6). Participants were excluded if they had a lifetime diagnosis of schizophrenia or bipolar 1 disorder, intellectual disability, current substance dependence, a BMI below 16.5 kg/m², pregnancy, unstable life circumstances (eg, homelessness), life-threatening suicide attempts within 2 months before enrollment, planned residential treatment, or recent (within the past year) CPT or DBT-PTSD therapy. Participants engaging in high-risk behaviors, continued self-harm, or suicidality (eg, suicide attempts considered to be at low risk of death) were not excluded.

Of the 193 participants initially enrolled in the RCT, 18 (9.3%), including 8 (4.1%) in DBT-PTSD and 10 (5.2%) in CPT, were excluded from the e-diary analysis due to technical problems or incomplete participation. This resulted in a final sample of 175 participants who completed at least 1 e-diary assessment (DBT-PTSD: n=90, CPT: n=85).

Ethical approval was obtained from the relevant institutional review boards, including the Medical Faculty Mannheim at Heidelberg University (reference 2013‐635 N-MA), as well as the ethics committees of Goethe University and Humboldt University. Written informed consent was obtained from all participants before their inclusion. Safety and data quality were independently monitored by the Coordination Centre for Clinical Trials, Heidelberg. Participants did not receive any compensation.

A detailed description of the study protocol and interventions has been published elsewhere [12].

During an initial diagnostic session, trained clinical psychologists assessed inclusion and exclusion criteria, as well as mental health diagnoses. PTSD diagnoses were determined using the Clinician Administered PTSD Scale for DSM-5 [3]. Comorbid Axis 1 disorders were assessed via the Structured Clinical Interview for DSM-4 Axis 1 disorders [24], and BPD was evaluated with the International Personality Disorder Examination [25]. Participants also completed self-report questionnaires and provided demographic data, as detailed more extensively for the full sample in the original trial publication [14].

At each of the 3 assessment points (baseline, after 6 months, and after 12 months of therapy), participants completed a 7-day e-diary protocol using a study smartphone (Samsung Galaxy S3 mini; Samsung Electronics) equipped with the movisensXS application (movisens GmbH). Before data collection began, participants received standardized instructions and a brief training session to ensure familiarity with the procedure.

The protocol consisted of 2 consecutive phases. During the first phase, which is not part of this manuscript, participants responded for 2 days to time-based prompts that appeared at semirandom intervals, approximately every 30 minutes, between 10 AM and 10 PM. On the evening of the second day (8 PM), participants received an automated reminder message informing them about the transition to the event-based phase starting the following day. This reminder explained that, from day 3 onward, prompts would be reduced to a single morning notification at 9 AM, and participants were instructed to initiate an entry themselves whenever intrusive thoughts or memories of their most distressing traumatic event occurred during the day. The message also reminded participants how to access and activate the event-based entry function (Thoughts/Memory button) on the app’s home screen.

During the subsequent 5 days, participants followed the event-based sampling approach. They were instructed to initiate an entry each time they experienced an intrusive memory of their most distressing traumatic event. For the present analyses, only data from the 5-day event-based phase at each assessment point were included.

At each self-initiated entry, participants rated their current level of inner tension on a visual analog scale ranging from 0 (no tension) to 100 (maximum tension) in response to the item: At this moment, I am experiencing unpleasant inner tension. They then indicated whether the memory or thought was related to their most distressing traumatic event (yes or no). Only entries identified as trauma-related were included in the analysis; non-trauma-related entries were automatically discontinued.

All entries were automatically time-stamped, encrypted, and securely transmitted to the study server. The app interface, including the visual analog scale (panel A) and the trauma-relatedness question (panel B), is illustrated in Figure S1 in Multimedia Appendix 1.

Statistical analyses were conducted in R (R Foundation for Statistical Computing). Baseline levels of intrusion frequency and related inner tension were derived by averaging all available entries from the 5-day pretreatment e-diary period within each participant. Group differences (DBT-PTSD vs CPT) were examined using 2-tailed independent-samples t tests.

Subsequent analyses used multilevel modeling to account for the nested data structure (ie, repeated measures within participants). To examine whether treatment group, time, and their interaction predicted the daily frequency of intrusive memories, a multilevel zero-inflated negative binomial model was used. This approach was chosen to appropriately model the characteristics of the outcome variable (intrusion frequency), which exhibited substantial overdispersion and a high proportion of zero values. For momentary tension ratings, a linear mixed-effects model was conducted, with treatment group, time, and their interaction as predictors. A binary dropout variable was included in both models to address potential biases related to missing data.

e-Diary data from 175 participants, including 90 (51.4%) in DBT-PTSD and 85 (48.6%) in CPT, were included in the analysis. Across measurement points, participants completed, on average, 11.4 (SD 10.1; range 1‐53) e-diary assessments in the CPT group and 12.0 (SD 9.94; range 1‐50) assessments in the DBT-PTSD group. As presented in Table 1, the treatment groups showed no significant differences in demographic or clinical baseline characteristics.

At baseline, participants reported an average of 0.99 (SD 1.01) intrusive memories per day and a mean inner tension score of 62.48 (SD 23.10; scale 0‐100). Baseline levels of intrusive memories and inner tension did not differ significantly between the 2 treatment groups (Table 1).

On average, participants demonstrated reductions in intrusive memory frequency and related inner tension over time, with a more pronounced improvement in inner tension observed in the DBT-PTSD group. At the 12-month assessment, participants in the DBT-PTSD group reported a mean of 0.31 (SD 0.55) intrusions per day and a mean inner tension score of 38.15 (SD 27.56). In comparison, participants in the CPT group reported an average of 0.26 (SD 0.42) intrusions per day and a mean inner tension score of 56.98 (SD 30.51).

For intrusion frequency, a significant main effect of time was found, indicating a general reduction in intrusions across both treatment groups (Table 2). However, neither the group effect nor the time × group interaction reached significance. The inclusion of dropout status did not significantly alter the results.

For intrusion-related inner tension, a significant main effect of time also indicated reductions across the study period. Importantly, a significant time × group interaction was found: participants in the DBT-PTSD group showed significantly greater reductions in inner tension over time compared with those in the CPT group (Table 2 and Figure 1). As with intrusion frequency, neither the main effect of group nor dropout status were significant. Full parameter estimates from the multilevel models are presented in Table 2.

This study used real-time e-diary assessments to evaluate intrusive memories and momentary inner tension experienced during these intrusions in daily life, comparing 2 PTSD treatments—DBT-PTSD and CPT—in a large, clinically complex sample of women with PTSD related to childhood abuse and co-occurring BPD features. By leveraging longitudinal e-diary data, this approach provided a nuanced perspective on individual symptom trajectories while minimizing recall bias and enhancing ecological validity.

Results indicated a significant decrease in intrusive memory frequency over time in both treatment groups, consistent with findings from the parent trial and prior research demonstrating the efficacy of trauma-focused interventions in reducing PTSD symptoms, including intrusions [14,26]. Importantly, the inclusion of participants with co-occurring BPD symptoms underscores the robustness of these improvements in a particularly challenging clinical population. By capturing symptom fluctuations in daily life, this analysis complements the parent trial by illustrating how therapeutic changes manifest in patients’ everyday environments.

While intrusion frequency decreased over time in both treatment groups, no significant group×time interaction emerged. This pattern indicates that both DBT-PTSD and CPT were associated with reductions in daily intrusive memories, without evidence of differential treatment effects. In contrast, the parent trial reported small but statistically significant overall advantages for DBT-PTSD over CPT in clinician-rated PTSD symptom reduction [14]. Notably, the parent trial examined overall symptom severity without analyzing differences at the level of specific symptom clusters, such as intrusions. In addition, this study assessed intrusive memories through self-initiated, momentary reports, whereas the parent trial relied on retrospective clinician ratings encompassing both frequency and intensity. These conceptual and methodological distinctions limit the direct comparability of findings and underscore the value of examining both aggregated and momentary symptom dynamics to better understand treatment-related change. Future research should extend this approach by examining how DBT-PTSD and CPT may differentially influence distinct PTSD symptom clusters in daily life, thereby clarifying how their respective therapeutic mechanisms translate into specific patterns of symptom change.

The most pronounced difference between treatments emerged in the reduction of inner tension reported at the time of intrusive memories, with significantly greater improvements observed in the DBT-PTSD group. This finding may reflect DBT-PTSD’s strong focus on emotion regulation and distress tolerance, which specifically target the heightened arousal and emotional reactivity characteristic of PTSD with comorbid BPD features [1]. Accordingly, these findings extend prior evidence indicating that DBT-PTSD is particularly well suited for this subgroup, while generalization to individuals with PTSD without BPD features should be made with caution. Supporting this interpretation, findings from the parent trial demonstrated greater reductions in dissociation-related tension (Dissociation Tension Scale [27]) and borderline symptoms (Borderline Symptom List 23 [28]) in the DBT-PTSD group compared with the CPT group [14]. Another potential contributor is DBT-PTSD’s use of repeated imaginal exposure, which promotes fear extinction through the activation of trauma-related networks and the integration of corrective experiences [15]. In contrast, CPT primarily targets maladaptive cognitions such as self-blame or distorted worldviews [29], which may be less directly related to immediate reductions in tension. Together, these findings underscore the importance of tailoring interventions to individual symptom profiles, such as the presence of BPD features, to optimize treatment outcomes.

While these findings provide important insights, several limitations should be acknowledged. First, the sample consisted exclusively of cisgender women with PTSD related to childhood abuse and comorbid BPD features, limiting generalizability to other sexes, trauma types, and PTSD without BPD symptoms. Second, although e-diaries allowed for real-time symptom capture, reliance on self-reports may introduce biases. Complementing self-reports with physiological or behavioral measures of arousal or distress could strengthen future research. Third, potential assessment-related reactivity cannot be fully ruled out, although no indication of such effects was observed in the present data. Fourth, the inner tension item may have been interpreted differently across treatments, as “tension” is explicitly addressed in DBT-PTSD but not in CPT, potentially affecting reporting. Fifth, the conceptual relationship between inner tension and general psychological distress requires further empirical investigation to clarify their overlapping and unique aspects. Finally, this study focused on intrusive memories and related inner tension, leaving other PTSD symptoms, such as avoidance and hyperarousal, unexplored. Extending e-diary assessments to these domains could provide a more comprehensive understanding of treatment effects.

This study is the first to embed e-diary assessments within a large randomized trial of DBT-PTSD and CPT, providing ecologically valid insights into daily-life symptom trajectories in a clinically complex PTSD population with co-occurring BPD features. Both treatments were associated with significant reductions in intrusive memory frequency and momentary inner tension experienced during such episodes, with DBT-PTSD showing greater reductions in the latter. These findings underscore the clinical relevance of both treatments, while suggesting that DBT-PTSD may be particularly beneficial for managing emotional dysregulation and distress in daily life. They further demonstrate that meaningful therapeutic change extends beyond questionnaire-based symptom scores into everyday functioning. Future research should continue to leverage e-diaries to examine how distinct treatment mechanisms translate into real-world recovery across a broader range of PTSD symptoms and in more diverse clinical populations, ultimately helping to refine and individualize treatment approaches for complex trauma presentations.