This analysis of a clinical trial (n=22) examines the role of therapeutic alliance in MDMA-assisted psychotherapy (MDMA-AP) for treating chronic PTSD. It reports that after controlling for baseline PTSD severity, a strong therapeutic alliance at the mid and late stages of treatment (sessions 4 and 9) significantly predicts lower clinician-assessed and self-reported PTSD severity post-treatment.
Papers cited by this study that are also in Blossom
Agin-Liebes, G. I., Zeifman, R. J., Luoma, J. B. et al. · Journal of Psychopharmacology (2022)
Background
MDMA-assisted psychotherapy (MDMA-AP) is a combined psychotherapeutic and pharmacologic intervention that shows promise in the treatment of posttraumatic stress disorder (PTSD). Although therapeutic alliance has been established as a key predictor across psychotherapies and is emphasised within MDMA-AP treatment manuals, research has not yet examined the relationship between therapeutic alliance and MDMA-AP treatment outcomes.
Objective
Examine whether therapeutic alliance predicts changes in PTSD symptoms following MDMA-AP.
Method
Twenty-three individuals with chronic PTSD participated in a MDMA-AP clinical trial that included a randomised (MDMA vs. placebo) and open-label phase. The present analyses focused on participants who were administered MDMA over the course of the randomised and open-label phases (n = 22). Therapeutic alliance was assessed using the Working Alliance Inventory at sessions baseline (pre-session 3) and sessions 4 and 9. PTSD symptoms were assessed using the Clinician Administered PTSD Scale and the Impact of Events Scale-Revised.
Results
Controlling for baseline clinician-assessed PTSD severity, therapeutic alliance at sessions 4 and 9 (but not baseline) significantly predicted post-MDMA-AP clinician-assessed PTSD severity. Controlling for baseline self-reported PTSD severity, therapeutic alliance at baseline (although this did not survive correction for multiple comparisons) and sessions 4 and 9 predicted post-MDMA-AP self-reported PTSD severity.
Conclusions
The present results provide the first preliminary evidence for the relationship between the therapeutic alliance and treatment outcomes within MDMA-AP for PTSD. These findings highlight the important role of psychotherapy, and common psychotherapeutic factors, within MDMA-AP. Replication in studies with larger and more diverse clinical samples remain necessary.
Post-traumatic stress disorder (PTSD) remains a chronic and disabling condition, and although several evidence-based psychotherapeutic and pharmacological treatments exist, there is still a need to improve outcomes and reduce dropout. MDMA-assisted psychotherapy (MDMA-AP) has shown promise in earlier trials, including a recent Phase III study, but the field has had limited evidence about how its psychotherapeutic components contribute to benefit. Zeifman and colleagues note that therapeutic alliance—the collaborative relationship between patient and therapist—is widely recognised as an important predictor in psychotherapy and is emphasised in MDMA-AP manuals, yet it had not been formally tested as a predictor of MDMA-AP outcomes. The study therefore aimed to examine whether therapeutic alliance predicts changes in PTSD symptoms after MDMA-AP, using data from a previously published clinical trial. The paper is presented as an initial investigation of a possible mechanism of change within MDMA-AP, with relevance both to clinical practice and to ongoing debates about the relative importance of psychotherapy versus medication effects in psychedelic-assisted treatments.
The parent study was a clinical trial in adults with chronic PTSD that included a double-blind randomised phase followed by an open-label phase. Participants were recruited through internet advertisements and letters to psychotherapists, and eligibility required, among other criteria, age at least 18 years, chronic DSM-IV PTSD, moderate to severe symptom severity, prior unsuccessful PTSD treatment, and no major medical contraindications or current psychiatric comorbidity beyond specified exceptions. The extracted text indicates that the trial aimed to enrol 21 individuals with replacement of drop-outs, but the present analyses included 22 participants who received MDMA at least once. In Phase I, participants were randomised to MDMA (125 mg) or inactive placebo, alongside non-drug psychotherapy. Following a protocol amendment, some later participants received an additional half-dose of MDMA or placebo (62.5 mg) two hours into the session. In Phase II, placebo-assigned participants could cross over to open-label MDMA. The therapy model followed the MDMA-AP manual and used a fixed male-female co-therapist dyad for all sessions. Treatment included preparatory sessions, long experimental sessions lasting 8-10 hours with overnight nursing observation, and integration sessions afterwards. The same dyad delivered all treatment sessions. Therapeutic alliance was assessed using the 36-item Working Alliance Inventory (WAI), which captures agreement on tasks and goals and the emotional bond between patient and therapist. Alliance was measured at baseline before session 3 and again after sessions 4 and 9. PTSD symptoms were measured with the clinician-rated Clinician Administered PTSD Scale (CAPS) and the self-report Impact of Event Scale-Revised (IES-R). The main analyses used hierarchical regression to predict post-treatment PTSD severity from baseline severity, time since baseline, and alliance at each timepoint; separate models were run for clinician-rated and self-reported outcomes. The authors also conducted sensitivity analyses excluding three participants omitted from the original outcomes paper, and supplemented the frequentist analyses with Bayesian generalised linear models, using a Region of Practical Equivalence to judge whether effects were negligible.
The clinical trial aimed to enrol a total of 21 individuals with PTSD with replacement of drop-outs (clinicaltrials.gov; NCT00090064). To be eligible for the clinical trial, participants were required to meet the following criteria: (a) aged ≥18; (b) chronic PTSD (based on the DSM-IV) resulting from a crime-related or military related traumatic experience; (c) moderate to severe PTSD severity (based on a Clinician Administered PTSD Scale [CAPS;] score of ≥ 50); (d) at least one unsuccessful PTSD treatment with a selective serotonin reuptake inhibitor (SSRI) or serotonin-norepinephrine reuptake inhibitor (SNRI) and one course of psychotherapy or unwillingness to engage in conventional PTSD treatment; (e) absence of current psychiatric comorbidities including borderline personality disorder or DSM-IV Axis I disorders (with the exception of anxiety disorders, affective disorders other than bipolar I disorder, substance abuse or dependence disorder in remission for ≥60 days, and eating disorder without active purging); (f) absence of major medical conditions; (g) absence of pregnancy; (h) use of effective birth control if female and able to have children; (i) weigh less than 50 kg; (j) proficiency in the English language. The study specifically aimed to recruit 20 individuals with treatment-resistant (failed pharmacotherapy and psychotherapy) PTSD (of ≥ 5 years in duration) and one veteran with military-related PTSD (≥1 and ≤5 years in duration) with failed PTSD treatment or unwillingness to engage in conventional PTSD treatment. Participants were recruited through internet advertisements and by sending letters to psychotherapists. Participants completed an initial telephone screen to determine eligibility and then an in-person visit to review and sign informed consent, followed by an interview with an independent rater and physician to further determine eligibility. Participants were obligated to taper off of and abstain from using psychotropic medications throughout the study. However, use of study physician approved sedative hypnotics or anxiolytics was permitted between experimental sessions. Ethical approval for the study was received from the Copernicus Group Independent Review Board (IRB), Research Triangle Park, NC, USA. MDMA was produced by David E Nichols, PhD for the study sponsor (Multidisciplinary Association for Psychedelic Studies [MAPS]).
Create a free account to open full-text PDFs.
Bershad, A. K., Mayo, L. M., Van Hedger, K. et al. · Neuropsychopharmacology (2019)
Carhart-Harris, R. L., Roseman, L., Haijen, E. C. H. M. et al. · Journal of Psychopharmacology (2018)
Davis, A. K., Streeter Barrett, F., Griffiths, R. R. et al. · Journal of Psychopharmacology (2021)
Feduccia, A. A., Jerome, L., Yazar-Klosinski, B. et al. · Frontiers in Psychiatry (2019)
Gukasyan, N., Nayak, S. · Preprints (2020)
Haijen, E. C. H. M., Kaelen, M., Roseman, L. et al. · Frontiers in Pharmacology (2018)
Jardim, A. V., Jardim, D. V., Chaves, B. D. R. et al. · brazilian Journal of Psychiatry (2021)
Kettner, H., Rosas, F. E., Timmermann, C. et al. · Frontiers in Pharmacology (2021)
Mitchell, J., Bogenschutz, M. P., Lilienstein, A. et al. · Nature Medicine (2021)
Grob, C. S., Mithoefer, M. C., Brewerton, T. D. · Lancet Psychiatry (2016)
Mithoefer, A. T., Mithoefer, M. C., Feduccia, A. A. et al. · Lancet Psychiatry (2018)
Mithoefer, M. C., Wagner, M. T., Mithoefer, A. T. et al. · Journal of Psychopharmacology (2010)
Molla, H. M., Lee, R., Lyubomirsky, S. et al. · Scientific Reports (2023)
Monson, C. M., Wagner, A. C., Mithoefer, A. T. et al. · European Journal of Psychotraumatology (2020)
Murphy, R., Murphy-Beiner, A., Kettner, H. et al. · Frontiers in Pharmacology (2022)
Oehen, P., Traber, R., Widmer, V. et al. · Journal of Psychopharmacology (2012)
Ot'alora G, M., Grigsby, J., Poulter, B. et al. · Journal of Psychopharmacology (2018)
Peill, J. M., Trinci, K. E., Kettner, H. et al. · Journal of Psychopharmacology (2022)
Sessa, B., Higbed, L., Durant, C. et al. · Journal of Psychopharmacology (2021)
Traynor, J. M., Roberts, D. E., Ross, S. et al. · Focus (2022)
Wagner, A. C. · Frontiers in Psychology (2021)
Wagner, A. C., Liebman, R. E., Mithoefer, A. T. et al. · Frontiers in Psychiatry (2021)
Wardle, M. C., De Wit, H. · Psychopharmacology (2014)
Wolfson, P. E., Andries, J., Feduccia, A. A. et al. · Scientific Reports (2020)
Zeifman, R. J., Kettner, H., Pagni, B. A. et al. · Scientific Reports (2023)
Zeifman, R. J., Wagner, A. C., Monson, C. M. et al. · Journal of Affective Disorders (2023)
Papers in Blossom that reference this study
Ramaekers, J. G., Kuypers, K. P. C., Vollenweider, F. X. · Molecular Psychiatry (2026)
Cherniak, A. D., Granqvist, P., Mikulincer, M. et al. · Journal of Psychoactive Drugs (2026)
The sample had a mean age of 57.50 years (SD 7.51), was entirely Caucasian, and was predominantly female (77.3%). Mean PTSD duration was 18.2 years, and baseline clinician-rated PTSD severity was high (mean CAPS 82.9, SD 21.77). Most participants had crime-related trauma (90.9%), and all had failed at least one SSRI or SNRI treatment. There were no significant sex differences in therapeutic alliance at baseline, session 4, or session 9. For clinician-rated PTSD outcomes, baseline severity and days since baseline did not significantly predict post-treatment severity. Adding therapeutic alliance at baseline did not significantly improve the model. However, alliance at session 4 significantly increased explained variance in post-treatment CAPS scores, accounting for an additional 29% of the variance, and alliance at session 9 also significantly increased explained variance, accounting for an additional 24%. Bayesian analyses were consistent, with session 4 and session 9 alliance showing evidence of meaningful effects, whereas baseline alliance did not. For self-reported PTSD outcomes, baseline self-reported severity was not significantly associated with post-treatment severity. Baseline alliance initially appeared to improve prediction and accounted for an additional 20% of the variance, but this did not survive correction for multiple comparisons. Alliance at session 4 significantly predicted post-treatment self-reported PTSD severity, accounting for an additional 40% of the variance, and alliance at session 9 also significantly predicted outcomes, accounting for an additional 26%. Bayesian analyses again supported the session 4 and session 9 findings, while baseline alliance was not robust. Sensitivity analyses excluding the three participants omitted from the primary outcomes paper showed the same overall pattern.
Zeifman and colleagues interpret these findings as the first preliminary empirical evidence that therapeutic alliance is related to outcomes in MDMA-assisted psychotherapy for PTSD. They emphasise that alliance measured during treatment, particularly at sessions 4 and 9, was associated with both clinician-rated and self-reported symptom improvement, whereas baseline alliance was weaker and did not remain significant after correction for multiple comparisons in the self-report analysis. The authors describe the observed effects as fairly strong, with alliance explaining roughly 11-40% of the variance in PTSD outcomes beyond baseline severity and time since baseline. The paper places these results in the context of broader psychotherapy research showing that therapeutic alliance is linked to better outcomes, including in PTSD treatment. The authors argue that the findings support the view that psychotherapy is an important component of MDMA-AP and that MDMA should not be understood as a simple stand-alone drug treatment. They also relate the results to a broader debate about common psychotherapeutic factors in pharmacologically enhanced therapies, including therapies involving psilocybin. Several limitations are acknowledged. The sample was small, predominantly female, and exclusively Caucasian, limiting generalisability. Therapeutic alliance was measured only by patient self-report, and the PTSD measures used DSM-IV criteria rather than more recent DSM-5-TR instruments. The study was also limited to one MDMA-AP model, one therapist dyad, and only three alliance timepoints, so the authors say the findings cannot establish a unique causal role for alliance or determine whether MDMA itself changes alliance. They note that future research should include larger and more diverse samples, therapist- and observer-rated alliance measures, session-by-session assessment, and examination of whether the findings extend to other MDMA-AP indications and treatment models. In terms of practical implications, the authors suggest that attention to building and repairing alliance before MDMA sessions may be worthwhile, and that training manuals may need more explicit guidance on alliance-enhancing and other common-factor techniques. They also suggest extra care may be needed when working with populations in whom strong alliance is harder to establish.
The parent study was a clinical trial (clinicaltrials.gov; NCT00090064) that included a double-blind randomised controlled phase (Phase 1) followed by an openlabel phase (Phase 2). Participants were initially randomised to two experimental sessions in which they were administered either (a) MDMA (125 mg) or (b) inactive placebo (lactose) (Phase 1). Following a protocol amendment, the final 9 participants were allowed to receive a supplement half-dose of MDMA (62.5 mg) or placebo in all of their experimental sessions (administered 2 h into the experimental session). Four participants in the MDMA group and 4 participants in the placebo group received supplemental doses in this manner. Individuals in both conditions also received non-drug psychotherapy. In Phase 1, participants received two preparatory psychotherapy sessions (sessions 1 and 2), two experimental sessions (sessions 3 and 8), and eight additional non-drug psychotherapy sessions (sessions 4-7 and 9-12), including one the day after each experimental session (sessions 4 and 9). Additional sessions were permitted, if needed, on a case-by-case basis. In Phase 2, participants in the placebo condition were eligible to receive two open-label MDMA sessions (sessions 13 and 17) each of which was followed by three psychotherapy sessions (sessions. Following a protocol amendment, four participants initially in the placebo condition and five participants initially in the MDMA condition received an additional (i.e. a third) MDMA session. Sixty percent of participants were randomised to receive MDMA and 40% of participants were randomised to placebo (with replacement of dropouts). Thirteen individuals (including one individual with military-related PTSD that was not required to be treatment-resistant) were initially randomised to the MDMA condition and 8 individuals were randomised to placebo. Two individuals assigned to the MDMA condition dropped out of treatment prior to their second experimental session (due to travelling difficulties for one individual and resumption of medication for depression relapse for the other individual) and two additional individuals were subsequently randomised to the MDMA condition. The two participants who dropped out of treatment and one individual with military-related PTSD (that was not required to be treatment-resistant) were excluded from the primary outcomes paperbut are included here due to: (a) meeting criteria for PTSD; (b) receiving psychotherapy and at least one MDMA-assisted psychotherapy session; and (c) to ensure that findings are not biased by treatment dropout (e.g. due to poor therapeutic alliance). One individual in the placebo arm opted not to receive open-label MDMA and is therefore excluded from the present analysis. Due to limited sample size, all participants (n = 22) administered MDMA (whether in Phase 1 or 2) are included in the present analyses. For study flow, see Figure. For study procedures, see Figure. The study was conducted in accordance with the MDMA-AP treatment manual (), utilising a psychotherapeutic platform administered by a dyad therapy team, and inclusive of preparatory sessions, administration of MDMA/placebo, and post-dosing integration sessions. Preparatory sessions focused on orienting participants to the therapeutic approach, the structure of experimental sessions, and the acute effects of MDMA. Experimental sessions occurred in a comfortable outpatient office, participants were offered the option to use eyeshades while reclining in a comfortable position, and listening to a pre-set music playlist. Experimental sessions lasted 8-10 h after which participants remained overnight with a nurse on site to ensure safety. When needed, zolpidem or lorazepam were prescribed for difficulty sleeping following experimental sessions. Integration sessions focused on emotional processing and reviewing the experimental session, as well as identifying and reflecting on new insights and perspectives related to their life and PTSD. Additional integration sessions occurred when judged to be necessary by the study investigators. The same male and female co-therapist team were present for all treatment sessions. For further details, see. Subjects, investigators, nurses, and independent raters were blinded to condition during Phase 1. Outcome measures were repeated and the blind was broken two months after participants' second MDMA/placeboassisted psychotherapy session. PTSD symptoms were measured at baseline (prior to session 3) and 3-7 days after each MDMA/placebo session, 2 months after the second experimental session, and (among individuals that were initially assigned to receive placebo and then enrolled in Phase 2) 6-8 weeks after their final MDMA session. Clinician-assessed PTSD severity ratings were completed by a blinded independent rater. PTSD symptoms at participants' final assessment (with the majority occurring after session 12 and as late as after session 24) are used as post-treatment outcomes in the present analyses. Participants completed ratings of therapeutic alliance at baseline (prior to session 3) and at sessions 4 and 9.
Therapeutic alliance was measured using the WAI. The WAI is a 36 item self-report measure of the client's perception of the therapeutic alliance that includes three subscales (Tasks [agreement on the task of therapy], Goals [agreement on the goals of therapy], and Bonds [affective bond between the therapist and patient]) and a Total score. Items (e.g. 'I believe _____ is genuinely concerned for my welfare' and '___ and I trust one another') are rated on a Likert scale from 1 (never) to 7 (always). Fourteen items are reverse scored. The total WAI score is the mean of all items with higher scores representing a stronger therapeutic alliance. The WAI shows good psychometric properties, including strong internal consistency. In line with previous research with co-therapists (e.g., participants were instructed to complete the WAI once at each timepoint with reference to both of their co-therapists.
Clinician-assessed PTSD symptoms were measured using the Clinician Administered PTSD Scale (CAPS;. The CAPS assesses the 17 PTSD symptoms in the DSM-IV, including identification of a criterion A trauma, symptoms of re-experiencing, avoidance, and hyperarousal, symptom duration, and level of distress/impairment. For each symptom, the CAPS includes measures of frequency (rated on a scale 0-4) and intensity (rated on ) indicates a visit with assessments but no therapy sessions occurred. Sessions with striped backgrounds ( ) represent those added following protocol amendments. *Supplemental half doses were available after a protocol amendment, and the final nine participants enrolled were allowed this supplement. Four patients in the MDMA group and four in the placebo group received supplemental doses. #A protocol amendment allowed four participants initially in the placebo condition and five participants initially in the MDMA condition to receive an additional (i.e. a third) MDMA session and three integration sessions. a scale 0-4), which are then summed to calculate the severity for each symptom (ranging from 0 to 8). A total PTSD severity is then calculated by summing each of the 17 severity scores. The CAPS was considered a gold-standard measure of PTSD at the time the parent trial was conducted and has excellent psychometric properties. Self-reported PTSD symptoms were measured using the revised Impact of Event Scale-Revised (IES-R;. The IES-R is a 22-item selfreport measure developed to assess PTSD symptoms based on the DSM-IV. It includes three subscales (intrusion [8 items], avoidance [8 items], and hyperarousal [6 items]), and a total score. Participants identify a stressful life event and rate event-related items (e.g. 'I tried not to think about it' and 'I had waves of strong feelings about it') based on the amount of distress they caused them over the previous 7 days. Items are rated on a scale from 0 (not at all) to 4 (extremely). The IES-R has shown good psychometric properties, including good internal consistency and test-retest reliability.
Independent samples t-tests examined potential differences in therapeutic alliance based on participant sex. Correlation coefficients were calculated for therapeutic alliance (baseline and sessions 4 and 9) and self-reported and clinician-assessed PTSD severity (baseline and post-treatment). Hierarchical regression was used to examine the contribution of therapeutic alliance in explaining the variance in post-treatment PTSD severity (controlling for baseline PTSD severity and the number of days since baseline that post-treatment PTSD severity was measured at). Separate analyses were run for self-reported (IES-R) and clinician-assessed (CAPS) PTSD treatment outcomes. In the first stage, pre-treatment PTSD severity and days since baseline were entered as independent variables. Post-treatment PTSD severity was entered as the dependent variable. For the second stage, therapeutic alliance was entered as an independent variable with separate analyses conducted for therapeutic alliance at baseline (Stage 2a) and sessions 4 (Stage 2b) and 9 (Stage 2c). We set the alpha level for significance as p < .05 (two-tailed) and further examined statistical significance following FDR correction for multiple comparisons using the Benjamini and Hochberg method (p-FDR < .05). Analyses were conducted using SPSS version 28. Sensitivity analyses were conducted to confirm results were consistent when excluding the three participants excluded from the primary outcomes paper (due to not having treatment-resistant PTSD [n = 1] or only receiving single dose of MDMA [n = 2]). The above frequentist hierarchical regression analyses were supplemented with Bayesian generalised linear modelling, which is known to produce less biased results compared to maximum likelihood estimation methods when sample sizes are small. The Region of Practical Equivalence (ROPE) was defined as the range from -0.1-0.1, corresponding to a negligible effect size according toafter z-standardization of the outcome (CAPS and IES-R) and predictor (WAI at stage 2a, 2b, and 2c) variables (to achieve scale-invariance of the posterior coefficient estimates in relation to the ROPE;. Significance of an effect was defined as <2.5% of the posterior distribution falling into the ROPE. Bayesian analyses were conducted in R using the rstanarmand bayestestR packages. Following reporting guidelines recommended by Makowski, Ben-Shachar, Chen, et al. (), three indices are provided for each model: (1) existence of effect, indicated by the probability of direction (pd; i.e. the percentage probability that an effect of the predictor on the outcome exists in a given direction); (2) significance of the effect, indicated by the percentage of the full posterior distribution within the ROPE; and (3) effect size, indicated by the standardised median coefficient.
The sample (Mean age = 57.50; SD = 7.51) was exclusively Caucasian (100%) and predominantly female (77.3%). There were no statistically significant differences between males and females in therapeutic alliance at baseline (female mean = 220.94, SD = 27.99; male mean = 224.20, SD = 19.69; t(20) = .24, p = .812), session 4 (female mean = 226.75, SD = 24.46; male mean = 227.00, SD = 19.60; t(18) = .02, p = .985), or session 9 (female mean = 227.59, SD = 22.19; male mean = 223.20, SD = 11.65; t(20) = -.42, p = .679). The mean duration of participants' PTSD was 18.2 years with a baseline clinician-assessed PTSD severity of 82.9 (SD = 21.77). Twenty individuals (90.9%) had a crime-related index trauma and two individuals (9.1%) had a military-related index trauma. All individuals had at least one unsuccessful PTSD treatment with an SSRI or SNRI. Twenty-one individuals (95.5%) had previously received at least one course of psychotherapy. For means, standard deviations, and correlations for therapeutic alliance and PTSD severity measures, see Table.
In Stage 1, baseline clinician-assessed PTSD severity and days since baseline were not significantly associated with post-treatment clinician-assessed PTSD severity (F[2, 19] = .14, p = .869, R 2 = .12) and accounted for 1% of the variance. At Stage 2(a), adding therapeutic alliance at baseline to the model did not result in a significant change in explained variance (R 2 = .13; F change = 2.42, p = .137, p-FDR = .137; r 2 change = .11). At Stage 2(b), adding therapeutic alliance at session 4 to the model resulted in a significant increase in explained variance (R 2 = .29; F change = 6.77, p = .019, p-FDR = .038) and accounted for an additional 29% of the variation in post-treatment PTSD severity, above and beyond baseline PTSD severity and days since baseline. At Stage 2(c), adding therapeutic alliance at session 9 to the model resulted in a significant increase in explained variance (R 2 = .26; F change = 5.89, p = .026, p-FDR = .039) and accounted for an additional 24% of the variation in post-treatment PTSD severity, above and beyond baseline PTSD severity and days since baseline. Bayesian analyses were consistent with frequentist analyses, with therapeutic alliance at baseline not resulting in a significant change in explained variance (12.3% in ROPE), while therapeutic alliance at sessions 4 and 9 accounted for a significant amount of variance (both 0% in ROPE) above and beyond baseline clinicianassessed PTSD severity and days since baseline. See Supplementary Materials for further details.
At Stage 1, baseline self-reported PTSD severity was not significantly associated with post-treatment selfreported PTSD severity (R 2 = .03; F[1, 19] = .27, p = .767) and accounted for 3% of the variance. At Stage 2(a), adding therapeutic alliance at baseline to the model resulted in a significant increase in explained variance (R 2 = .22; F change = 4.54, p = .047), although this increase did not survive correction (p-FDR = .056), and accounted for an additional 20% of the variation in post-treatment PTSD severity, above and beyond baseline PTSD severity and days since baseline. At Stage 2(b), adding therapeutic alliance at session 4 to the model resulted in a significant increase in explained variance (R 2 = .40; F change = 10.56, p = .005, p-FDR = .030) and accounted for an additional 40% of the variation in post-treatment PTSD severity, above and beyond baseline PTSD severity and days since baseline. At Stage 2(c), adding therapeutic alliance at session 9 to the model resulted in a significant increase in explained variance (R 2 = .29; F change = 6.70, p = .019, p-FDR = .038) and accounted for an additional 26% of the variation in post-treatment PTSD severity, above and beyond baseline PTSD severity and days since baseline. Bayesian analyses were consistent with frequentist analyses, with therapeutic alliance at baseline not resulting in a significant change in explained variance (3.1% in ROPE), while therapeutic alliance at sessions 4 and 9 accounted for a significant amount of variance (0% and 0.01% in ROPE, respectively) above and beyond baseline selfreported PTSD severity and days since baseline. See Supplementary Materials for further details.
Supplementary frequentist and Bayesian analyses confirmed that the above results were consistent when excluding the three participants who had been excluded from the primary outcomes paper (see Supplementary Material).
In this examination of therapeutic alliance as a putative mechanism underlying MDMA-AP's therapeutic effect on PTSD, therapeutic alliance at baseline (presession 3) and sessions 4 and 9 significantly predicted self-reported PTSD outcomes. Notably, the relationship between therapeutic alliance at baseline and self-report PTSD outcomes did not survive correction for multiple comparisons. Additionally, therapeutic alliance at sessions 4 and 9 (but not baseline) significantly predicted clinician-assessed PTSD outcomes. The relationship between therapeutic alliance and treatment was fairly strong with therapeutic alliance accounting for 11-40% of the variance in PTSD treatment outcomes, above and beyond baseline PTSD and days since baseline. Therapeutic alliance at baseline accounted for 11% of the variance in clinician-assessed PTSD outcome, which is a medium sized effect that is consistent with previous research on the relationship between therapeutic alliance and PTSD treatment outcomes (for a meta-analysis, see. Importantly however, this effect was not statistically significant and should therefore be interpreted with strong caution. All results were consistent when examined in the full sample of individuals that had received at least one MDMA session, as well as among those that had received at least two MDMA sessions and had a history of both failed psychotherapeutic and pharmacological PTSD treatment. These findings provide the first preliminary empirical support for the importance of therapeutic alliance within MDMA-AP or the relationship between therapeutic factors and MDMA-AP treatment outcomes. These findings are generally consistent with research indicating the therapeutic alliance is predictive of treatment outcomes across psychotherapeutic interventions, including treatments for PTSD, and suggest that the importance of therapeutic alliance may extend to MDMA-AP. There is a growing debate regarding the importance of psychotherapy in MDMA-AP and interventions with related pharmacological properties (e.g. psilocybin therapy;. The results of the present study provide preliminary support for the importance of the psychotherapeutic component of MDMA-AP and against MDMA being a traditional stand-alone pharmacological intervention. They also provide early support for suggestions that such pharmacologically-enhanced interventions lead to change via common factors of psychotherapy, including the therapeutic alliance. While further research remains necessary due to the limited sample size and correlational nature of present findings, several studies have similarly found support for a relationship between the therapeutic alliance and one's bond with individuals present during one's acute experience and subsequent improvements in mental health. For instance, a recent study found that therapeutic alliance was predictive of reductions in depressive symptoms following psilocybin therapy for major depressive disorder. Although further replication remains necessary to draw strong conclusions, the present findings have several potentially important clinical implications. First, they suggest that it is worthwhile to pay specific attention to developing and enhancing therapeutic alliance prior to providing MDMA sessions. It may therefore be worthwhile for MDMA-AP training manuals and treatment development provide specific guidance on enhancing the therapeutic alliance (e.g., resolving ruptures in the alliance, and to draw more explicitly from common factor techniques. Additionally, the present results suggest that if MDMA-AP is extended to samples of individuals that exhibit complex comorbidities or greater difficulties with establishing a strong therapeutic alliance or where the strength of the therapeutic alliance plays an especially important role (e.g. individuals with borderline personality disorder [BPD];, further attention may be needed to ensure that a strong therapeutic alliance is established. The present study has several important limitations. The sample was fairly small and non-diverse (i.e. mostly female and exclusively Caucasian), which is a key issue in MDMA-AP research to date. Further research in larger samples with greater diversity will be necessary to identify the generalizability of these findings. Therapeutic alliance was only measured via patient-rated self-report. Therefore, additional research that includes therapist and observer-rated therapeutic alliancewill help to further elucidate the role of therapeutic alliance within MDMA-AP. The PTSD symptom measures used in the present study were based on DSM-IV PTSD symptoms. Therefore, it will be important for future research to examine the relationship between therapeutic alliance and measures of DSM-5-TR PTSD symptoms (e.g. the CAPS-5;. The present findings were only examined in the context of MDMA-AP for PTSD. As research begins to expand the application of MDMA-AP, including to alcohol use disorder,, distress related to life-threatening illness, and couples-based therapy, it will be necessary to examine whether the role of therapeutic alliance within MDMA-AP extends across clinical samples. It will also be important to examine the relationships between therapeutic alliance and outcomes including: (a) experiences during MDMA sessions (e.g. cognitive reappraisal, insight, and acceptance/ avoidance); (b) trait-level changes (e.g. insightand avoidance); and (c) treatment outcomes that extend beyond PTSD severity (e.g. quality of life and relational improvements [e.g. see). Although the therapy model used in the present study is the most commonly researched MDMA treatment model, there remains a need for examining the role of therapeutic alliance within alternative MDMA-AP models (e.g. MDMA-assisted cognitive behavioural conjoint therapy or exposure therapy;. Additionally, a single therapist dyad provided MDMA-AP for all participants in the present study. Further research will therefore be necessary to determine the generalizability of the present findings across MDMA-AP providers. Finally, the present study only assessed therapeutic alliance at three timepoints, had a small sample (particularly within the placebo condition), and was not designed to establish a unique causal role of therapeutic alliance or the effect of MDMA-AP on therapeutic alliance itself. Given research suggesting that MDMA may enhance feelings of closeness and alters processing of physiological and behavioural responses to social stimuli, larger studies that measure therapeutic alliance at each session (including during MDMA sessions) will be important for examining the effect of MDMA on therapeutic alliance, differences with therapy that does not include MDMA, and the potential causal role of therapeutic alliance within MDMA-AP.