Ketamine-Assisted Psychotherapy for Posttraumatic Stress Disorder: A Systematic Review and Individual Participant Data Meta-Analysis of Predictors of Treatment Effects Authors: Judith Rohde, Tyler M. Moore, Kathryn Walker, Shobi S. Ahmed, Henry A. MacConnel, Martin Krsak, Scott Shannon, Vivian W.L. Tsang, Shannon Dames, Marco Buchmann, Birgit Kleim, Erich Seifritz, Sebastian Olbrich, Georgios Schoretsanitis Published: 2026 ABSTRACT AbstractIntroductionKetamine combined with psychotherapy has shown promise in treating posttraumatic stress disorder (PTSD), but predictors of clinical benefit remain unclear. In this systematic review and individual participant data meta-analysis (IPDMA), we examined patient- and treatment-level predictors of PTSD symptom change after combined treatment.MethodsEligible studies combined ketamine with psychotherapy in adults with PTSD. The primary outcome was PTSD symptom change on the PTSD Checklist for DSM-5 (PCL-5). We fit linear mixed-effects models with a random intercept for study and fixed effects for age, sex, baseline PCL-5, number of ketamine sessions, number of psychotherapy sessions, administration route, and treatment duration. Sex-stratified analyses and a quality-restricted sensitivity analysis were conducted.ResultsTwelve studies (533 participants) were included; nine were of poor quality. Greater improvement was associated with more psychotherapy (β = 1.03 per session) and more ketamine sessions (β = 1.15 per session), higher baseline symptom severity (β = 0.50 per PCL-5 point), and shorter treatment duration (β = −0.58 per week). Between-study heterogeneity was nontrivial (intraclass correlation coefficient ≈ 0.15). In sex-stratified models, number of psychotherapy sessions, baseline severity, and shorter treatment duration remained associated with improvement in both sexes. In the sensitivity analysis (n = 63), baseline severity was the only significant predictor of treatment outcome.ConclusionHigher baseline PTSD severity was the most robust predictor of symptom reduction. Session intensity and treatment timing may matter, but associations should be interpreted cautiously given the predominance of poor-quality studies and require replication in prospective trials with standardized protocols and longer follow-up. Introduction Posttraumatic stress disorder (PTSD) is a debilitating disorder associated with a substantial medical and economic burden [1]. Approximately 4% of the global population is affected [2]. PTSD is clinically heterogeneous. Among trauma-exposed individuals, younger age and female sex have been associated with higher risk [2]. Across settings, symptoms typically begin shortly after trauma exposure, although age at onset varies substantially, ranging from early adulthood to midlife in cross-national data [2]. Psychiatric comorbidity is common, with major depression being the most frequent co-occurring condition, while anxiety and substance use disorders are also prevalent [3]. In a meta-analysis, 52% of individuals with current PTSD also met criteria for current major depressive disorder [4]. Clinical guidelines recommend trauma-focused therapy as the first-line treatment [5–8], while pharmacological options are limited and typically modest in effect [9, 10]. For instance, guidelines primarily recommend selective serotonin reuptake inhibitors, with only sertraline and paroxetine approved by the European Medicines Agency and the US Food & Drug Administration [11–13]. On the other hand, psychotherapeutic options have several limitations, including notable dropout rates [14, 15] and insufficient symptom reduction for a substantial subgroup of patients, particularly those with comorbid psychiatric disorders such as depression or substance use disorder [16–18]. Given these challenges, it is evident that existing treatments for PTSD are not universally effective, indicating a pressing need for additional therapeutic options. Due to the complex pathophysiology of PTSD, alternative non-serotonergic pathways have been explored, including the potential of (R,S)-ketamine, hereafter referred to as ketamine. Ketamine is a noncompetitive N-methyl-D-aspartate (NMDA) receptor antagonist, well-known for its longstanding use in anesthesia and pain management [19]. The antidepressant properties of ketamine led the FDA and EMA to approve its enantiomer, S-ketamine, hereafter referred to as esketamine, for the therapy of treatment-resistant depression in 2019 [19, 20]. Moreover, recent reviews highlight ketamine as an emerging intervention for PTSD [21]. As a noncompetitive NMDA-receptor antagonist, ketamine may transiently modulate synaptic plasticity; when administered after trauma-memory retrieval, it has been hypothesized to influence extinction/reconsolidation-related learning processes and thereby potentially augment trauma-focused psychotherapy [22, 23]. While theoretically promising, randomized controlled trials of intravenous (IV) ketamine for PTSD have produced mixed results across heterogenous samples. Feder and colleagues demonstrated ketamine’s efficacy in patients with PTSD through single and repeated IV doses, showing rapid symptom reduction within 24 h, though with limited duration [24, 25]. Others reported reduced symptoms lasting up to 7 days post single ketamine infusion in patients with chronic pain and PTSD [26]. However, the largest RCT to date found no significant dose-related benefit of repeated IV ketamine over placebo on PTSD symptoms in 158 veterans [27]. Notably, these trials largely evaluated ketamine as monotherapy. In depressive and anxiety disorders, outcomes are often improved when pharmacotherapy is combined with psychotherapy [28]. In accordance, ketamine research and clinical practice have increasingly combined ketamine with psychotherapy (“ketamine-assisted psychotherapy”), spanning different routes of administration and varying the timing of psychotherapy relative to dosing [29]. Research on ketamine-assisted psychotherapy for PTSD has increased in recent years. For instance, Pradhan et al. [30] combined a single ketamine dosage with 12 sessions of Trauma Interventions using Mindfulness Based Extinction and Reconsolidation (TIMBER) in a pilot randomized trial involving 10 patients; the results suggested that the combination of ketamine infusions with trauma-focused therapy may enhance the effects on PTSD symptom reduction. Several subsequent studies with considerable methodological heterogeneity assessed the combination of ketamine and psychotherapy reporting encouraging outcomes, [21, 29–31]; with those including follow-up assessments [23, 30–32] showing durable effects. However, findings remain inconsistent: in a large naturalistic sample, Moore et al. did not detect a significant added benefit of clinic-provided psychotherapy over 30 or 180 days, although outside psychotherapy may have occurred and was not captured [33]. Similarly, in a retrospective chart review, MacConnel et al. [34] reported greater PTSD symptom reduction among patients who reported seeing an outside psychotherapist. Taken together, recent reviews and meta-analyses describe ketamine as a promising candidate intervention for PTSD, particularly when integrated with psychotherapy [35–37]. Studies further suggest that combining ketamine with trauma-focused psychotherapy may enhance – and potentially prolong – symptom reduction, but the evidence base remains limited by small samples, heterogeneous methods, uncontrolled study designs, and a lack of standardized protocols [36, 38]. Despite promising initial findings, it remains unclear how robust and generalizable the effects of combining ketamine with trauma-focused psychotherapy are across different study designs and patient populations and, more specifically, which factors predict benefit from the treatment. To address this gap, we conducted a systematic review and an individual participant data meta-analysis of studies combining ketamine with psychotherapy for PTSD. We aimed to identify clinical patient- and treatment-level predictors of change in PTSD symptoms. Methods We applied the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000553144: PRISMA 2020 checklist). The protocol was registered in PROSPERO (CRD420251131773). We followed PRISMA 2020 and PRISMA-IPD reporting guidance [39, 40]. Search Strategy and Study Selection Two researchers (J.R. and M.B.) independently performed a systematic literature search using the databases MEDLINE (via PubMed), PsycINFO, and Embase. The original search was conducted on August 31, 2025, and updated through April 16, 2026. The search strategy combined terms for ketamine, trauma-focused therapy/psychotherapy, and PTSD, adapted as appropriate for each database. No language filters were applied during the search; however, eligibility was restricted to English-language peer-reviewed publications during screening. Full database-specific search strategies are provided in online supplementary Table 2. As a supplementary completeness check during revision, we also screened ClinicalTrials.gov records with posted results using PTSD- and ketamine-related search terms; this did not identify any additional eligible record. Selection of eligible studies was independently performed by two authors (J.R. and M.B.) in the non-AI version of Rayyan based on the predefined eligibility criteria. Search results from each database were imported into Rayyan for screening. No automation tools were used. The study selection process is summarized in the PRISMA flow diagram (Fig. 1). Consensus was reached in all cases, so no additional co-author was involved. Authors were contacted to provide individual participant data, where data were not publicly available. Study flowchart. Inclusion and Exclusion Criteria Studies were included if they: (1) investigated the effects of the combination of ketamine and psychotherapy in adults aged 18 years or older, (2) designs comprised randomized controlled trials (RCTs), open-label trials, case series, or cohort studies, (3) were published in English in peer-reviewed journals, and (4) reported pre-post PTSD symptom severity using the PTSD Checklist for DSM-5. For eligibility, psychotherapy sessions were considered only when sessions involved an explicit psychotherapeutic intervention (individual or group) in addition to ketamine administration. Self-guided or unguided therapeutic activities were not considered psychotherapy sessions. Contacts limited to scheduling, monitoring/safety checks, psychoeducation, or routine preparation that was universally delivered as part of the ketamine protocol were not considered psychotherapy sessions. The PCL-5 (PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders [DSM-5]) is a 20-item self-report measure corresponding to the 20 PTSD symptoms of DSM-5, rated for the past month in a 0–4 Likert scale (total score range: 0–80). It shows strong psychometric properties (e.g., excellent internal consistency and good test-retest reliability and convergent validity) across samples [41–44]. To maximize comparability across studies, we required a uniform, DSM-5-consistent PTSD outcome measure and therefore restricted inclusion to studies reporting pre-post symptom change with the PCL-5. We excluded esketamine studies in order to avoid additional intervention heterogeneity and to keep the pharmacological intervention as consistent as possible across studies. Studies were excluded if they: (1) did not involve psychotherapy sessions, (2) did not assess PTSD-specific outcomes using PCL-5, (3) used esketamine nasal spray, or (4) were reviews, meta-analyses, protocols, dissertations, or conference abstracts. Additionally, studies without access to individual participant data were excluded. Online supplementary Table 3 lists excluded studies. Data Extraction and Coding Individual participant data were extracted and coded in a predesigned Microsoft Excel spreadsheet. Extracted variables included age, sex, route of ketamine administration (IV vs. combination of sublingual and intramuscular [SL/IM] vs. intramuscular [IM]), number of ketamine sessions, number of psychotherapy sessions, total treatment duration (in study-reported units), and baseline and post-treatment PCL-5 (post-treatment defined as the assessment closest to treatment end). Treatment duration referred to the overall study-reported duration of the combined treatment protocol rather than to ketamine dosing alone. Ketamine dose was not included as a predictor because dosing metrics were inconsistently reported across studies and not directly comparable across administration routes. To maximize comparability across studies, we prioritized the PCL-5 assessment closest to treatment end as the post-treatment time point, when multiple follow-up assessments were available; longer follow-up assessments (e.g., 1–6 months) were available only in a minority of studies and were therefore not synthesized quantitatively. Psychotherapy sessions were coded according to the eligibility definition described above. Because psychotherapy content, format, timing relative to ketamine administration, and degree of trauma processing varied substantially across studies, number of sessions was treated as a high-level indicator of therapeutic exposure rather than a standardized or mechanistically uniform intervention. We excluded participants who did not receive both ketamine and psychotherapy sessions (e.g., those allocated to control conditions) and minors (<18 years). Participants were included if a clinical PTSD diagnosis was established using validated instruments (e.g., CAPS-5 or equivalent); when diagnostic information was unavailable, eligibility required a baseline PCL-5 ≥31; participants with a PCL-5 score of <31 were not considered. Participants with missing PCL-5 data on the extracted variables were excluded. Because PCL-5 cutoffs for probable PTSD vary across validation studies (22–49) [45], and recommended thresholds cluster around 31–33 [43], we set eligibility at PCL-5 ≥31. This pragmatic choice aligns with prior studies and targets a clinically meaningful symptom level without unduly restricting inclusion [46–48]. In addition, we considered the psychotherapeutic component as trauma-focused when an evidence-based trauma-focused protocol was explicitly reported or trauma processing was clearly described as the core component; otherwise, psychotherapy was coded as insufficiently specified/unclear. Quality/Risk-Of-Bias Assessment The modified version of the Newcastle-Ottawa scale for nonrandomized studies was used for quality assessment [49]; we removed the item “representativeness of the exposed cohort,” that we judged to be related to applicability, and added ascertainment of PTSD diagnosis using standardized tools as described elsewhere [50]. Ascertainment of exposure (psychotherapy dose) was operationalized as the number of psychotherapy sessions. We deemed a minimum of eight sessions to be sufficient [51, 52]. We judged aspects of treatment-resistance and comorbidities in the comparability domain. Follow-up was considered sufficient if outcomes were assessed ≥4 weeks after treatment completion [53, 54] or, if no post-treatment follow-up was reported, if the overall treatment duration was ≥12 weeks, as a pragmatic approximation of a minimum treatment phase plus a short post-treatment observation window (≈8 weeks + ≈4 weeks). Statistical Analysis We performed an individual participant data meta-analysis using a linear mixed-effects model. The primary outcome was the change score ΔPCL-5 = PCL-5(pre) − PCL-5(post), with higher values indicating greater improvement. Fixed effects were age (per 10 years), sex, number of ketamine sessions, route of administration (SL and IM vs. IV vs. IM), number of psychotherapy sessions, total treatment length, and baseline PCL-5. To account for clustering by study, we included a random intercept. Between-study heterogeneity was quantified using the random-intercept variance (τ00) and the intraclass correlation coefficient (ICC), with the ICC indicating the proportion of total variance attributable to between-study differences. Continuous predictors were kept on their original scales (age scaled per 10 years). For two patients, missing age values were imputed with median values. Analyses were conducted in R (version 2025.05.0); we report Wald 95% confidence intervals and Satterthwaite p values. Cases with missing values on model variables were excluded listwise. Further, analyses were performed in women vs. men separately. In a sensitivity analysis, we repeated the analysis after excluding participants from studies deemed to be of poor quality. In one study treatment duration differed between patients [34]; thus, we included individual patients from this study (whose quality was deemed low), when the overall treatment duration was ≥12 weeks. Results Study Selection The electronic database search identified 127 articles from MEDLINE (via PubMed), 125 from Embase, and 68 from PsycINFO; two additional articles were retrieved through manual reference screening and co-authors’ collections. After removal of duplicates (n = 74), 248 unique articles remained. Following title and abstract screening, 205 records were excluded and 43 full texts were assessed for eligibility (see study flowchart, Fig. 1). Of these, 31 were excluded for the following reasons: missing pre/post-treatment PCL-5 data (n = 13), no access to individual participant data (n = 6), review/meta-analysis (n = 4), no psychotherapy sessions (n = 4), use of esketamine rather than ketamine (n = 1), and other reasons (n = 3) (online suppl. Table 3). For one study, authors provided data on additional patients, not published in the original study [33]. One dataset [55] partially overlapped with another study [56]. Following correspondence with the study authors, we included only the nonoverlapping subset of participants to avoid double counting. Ultimately, 12 studies fulfilled all inclusion criteria and were used for data extraction (Table 1). Table 1.Characteristics of included studiesAuthor, yearDesign n/n included in IPDMAΔ PCL-5Ketamine sessions, nRoute of ketamine applicationPsychotherapy sessions, nTreatment durationPsychotherapy typeaRemarksStudy qualityDames et al. [56] (2021)Case series94/3826.893SL/IM1512 weeks0Group therapy program and 1:1PoorDuek et al. [23] (2023)RCT28/1415.571IV51 week11:1 psychotherapy + ketamine vs. 1:1 psychotherapy + midazolamPoorFlynn et al. [57] (2025)Case series16/1423.363SL/IM66 weeks0Group therapy programPoorFlynn et al. [58] (2025)Case series32/1323.313SL/IM77 weeks1Group therapy programPoorKeizer et al. [59] (2020)Case series11/928.221IV51 week1Continuous ketamine infusion 96 h; 1:1 bedside psychotherapyPoorMacConnel et al. [34] (2025)Retrospective chart review117/11722.972–9IV4–181–45 weeks01:1 psychotherapyPoorbMoore et al. [33] (2025)Nonrandomized comparative202/27318.961–13IV1–101 day-25 weeks0cKetamine with vs. without 1:1 psychotherapyPoorPradhan et al. [60] (2018)RCT20/1049.101IV1210 weeks1TIMBER + ketamine vs. TIMBER + placeboFairShiroma et al. [61] (2020)Open-label single-arm study12/1040.803IV3–1010 weeks1​PoorRobison et al. [62] (2024)Case series10/725.293IV66 weeks1Group therapy programPoorTsang et al. [55] (2024)Nonrandomized comparative57/2119.953SL/IM1512 weeks0Group therapy program with vs. without ketamineFairTsang et al. [32] (2025)Case series8/724.713IM1512 weeks0Group therapy programFairWeeks are rounded to whole weeks; durations <1 week are reported in days.Δ, change; IPMA, individual participant data meta-analysis; PCL-5, post-traumatic stress disorder (PTSD) checklist for DSM-5; RCT, randomized controlled trial; TIMBER, trauma interventions using mindfulness based extinction and reconsolidation. aPsychotherapy type: 0 – not specified, 1 – evidence-based trauma-focused protocol or clear trauma-processing focus. bIndividual participants’ treatment courses were considered fair (overall treatment duration ≥12 weeks). cWith some psychotherapies according to evidence-based trauma-focused protocols. Characteristics of included studies Weeks are rounded to whole weeks; durations <1 week are reported in days. Δ, change; IPMA, individual participant data meta-analysis; PCL-5, post-traumatic stress disorder (PTSD) checklist for DSM-5; RCT, randomized controlled trial; TIMBER, trauma interventions using mindfulness based extinction and reconsolidation. aPsychotherapy type: 0 – not specified, 1 – evidence-based trauma-focused protocol or clear trauma-processing focus. bIndividual participants’ treatment courses were considered fair (overall treatment duration ≥12 weeks). cWith some psychotherapies according to evidence-based trauma-focused protocols. Study Characteristics Key characteristics of the included studies are presented in Table 1. All included studies were conducted in North America, specifically in the USA (n = 9) or Canada (n = 3). Most studies were conducted in outpatient or community-based settings, although settings varied. Among the 12 studies, eight were single-arm studies (case series/open-label/retrospective), two were randomized controlled trials, and two were non-randomized comparative studies. Ketamine sessions ranged from 1 to 13 administrations and were predominantly IV (n = 433, 81.2%), followed by combined SL/IM administration (n = 93, 17.4%) and intramuscular-only administration (n = 7, 1.3%). The number of psychotherapy sessions ranged from 1 to 18, and psychotherapy timing varied (before, during, immediately after, or 24–36 h after ketamine). Mean symptom improvement (ΔPCL-5) ranged from 15.57 to 49.10 across studies. Descriptions of the psychotherapy sessions varied substantially across studies in content, format (individual vs. group), timing relative to ketamine, and degree of trauma processing. Where an evidence-based trauma-focused protocol or clear trauma-processing focus was reported, psychotherapy was coded as trauma-focused; otherwise, it was coded as unclear/insufficiently specified (Table 1). Participant Characteristics The pooled IPDMA sample included 533 participants from 12 studies (mean age 42.8 years, SD 13.4; 60.6% female, 38.5% male, and 0.9% nonbinary). Quality Assessment Online supplementary Table 4 summarizes the quality assessment of individual studies according to the NOS. Of the 12 studies, three were rated as fair and nine as poor quality. Quality concerns were mainly raised due to follow-up being insufficient to adequately evaluate treatment effects, consideration of treatment-resistance or comorbidities among patients, ascertainment of PTSD diagnosis and lack of response/nonresponse analysis. Although two parent studies had randomized designs, our predictor analyses included only participants who actually received both ketamine and psychotherapy, whereas control-arm participants were excluded. We therefore assessed the contributing analytic samples using the modified Newcastle-Ottawa framework, as the present analyses were based on treated cohorts rather than randomized comparisons. Primary Analysis Greater improvement (higher ΔPCL-5) was associated with more ketamine sessions (β = 1.15, 95% CI: 0.02–2.29, p = 0.047), more psychotherapy sessions (β = 1.03, 0.24–1.82, p = 0.011), and more severe baseline PCL-5 ratings (β = 0.50, 0.38–0.62, p < 0.001). Longer treatment length was associated with smaller improvement (β = −0.58, −0.82 to −0.33, p < 0.001). Between-study heterogeneity was non-trivial (τ00 = 44.13; ICC = 0.15). Results are listed in Table 2. Table 2.Linear mixed-effects model results (outcome: ΔPCL-5)PredictorEstimate (β)95% CI p value(Intercept)3.47−19.23 to 26.170.764Age (per 10 years)0.17−0.88 to 1.210.754Sex: male−1.86−4.78 to 1.070.213Sex: nonbinary2.79−11.63 to 17.210.704Ketamine sessions1.150.02 to 2.290.047Application form (IV)4.07−15.72 to 23.850.687Application form (SL, IM)2.55−16.85 to 21.940.797Psychotherapy sessions1.030.24 to 1.820.011Treatment duration (weeks)−0.58−0.82 to −0.33<0.001Baseline PCL-50.500.38 to 0.62<0.001Random effects and model fit: σ2 = 257.34; τ00 (studyID) = 44.13; ICC = 0.15; N(studyID) = 12; observations = 533; marginal R2/conditional R2 = 0.173/0.294. Linear mixed-effects model results (outcome: ΔPCL-5) Random effects and model fit: σ2 = 257.34; τ00 (studyID) = 44.13; ICC = 0.15; N(studyID) = 12; observations = 533; marginal R2/conditional R2 = 0.173/0.294. Sex-Stratified Analyses In 323 females, greater improvement was associated with more psychotherapy sessions (β = 0.78, 95% CI: 0.24–1.32, p = 0.004), and more severe baseline PCL-5 ratings (β = 0.62, 0.47–0.77, p < 0.001) (online suppl. Table 5a); longer treatment length was associated with smaller improvement (β = −0.43, −0.73 to −0.12, p = 0.004). In 205 males, greater improvement was associated with more psychotherapy sessions (β = 1.54, 95% CI: 0.35–2.74, p = 0.011) and more severe baseline PCL-5 ratings (β = 0.42, 0.21–0.62, p < 0.001) (online suppl. Table 5b); longer treatment length was associated with smaller improvement (β = −0.67, −1.12 to −0.23, p = 0.003). Sensitivity Analysis When excluding participants from studies rated as poor quality (9/12 studies), the quality-restricted sample comprised 63 patients, of whom n = 40 were female (63.5%), from three studies rated as fair quality and eligible participants from a fourth study [34] who met the ≥12-week treatment-duration criterion (online suppl. Table 4). Greater improvement was associated with more severe baseline PCL-5 ratings (β = 0.56, 0.14–0.99, p = 0.010) (online suppl. Table 6); no other significant association was found. Discussion Summary of Results and Interpretation In this individual participant data meta-analysis of 12 studies (n = 533), the primary mixed-effects model showed that greater PTSD symptom improvement was associated with more psychotherapy sessions, more ketamine sessions, and higher baseline PCL-5 severity, whereas longer treatment length (in weeks) was associated with smaller improvement. Outcomes showed meaningful clustering by study (ICC ≈ 0.15), indicating nontrivial between-study heterogeneity. Sex-stratified analyses showed a consistent pattern in females (n = 323) and males (n = 205): more psychotherapy sessions and higher baseline severity were associated with greater improvement, and longer treatment length with smaller improvement. In the sensitivity analysis restricted to studies rated as fair quality (n = 63), only baseline severity was a significant predictor, while other associations were attenuated, likely reflecting limited power in this essentially smaller subset. Comparison with Literature Prior evidence shows considerable heterogeneity in combined ketamine-psychotherapy protocols for PTSD (e.g., treatment duration, ketamine regimens, and psychotherapy type, timing, and dose), limiting inference about active components and dose-response relationships [33, 36]. Against this background, our IPDMA across 12 heterogeneous protocols provides an initial signal that higher baseline PTSD symptom severity, a higher number of psychotherapy sessions (likely indexing overall treatment intensity/engagement rather than specific techniques), and shorter overall treatment duration were associated with greater symptom improvement; a higher number of ketamine sessions showed only a small, borderline-significant association in the primary model. The association between higher baseline severity and larger absolute improvement is commonly reported in PTSD outcome research [63, 64] as well as beyond PTSD [65] and likely reflects, at least in part, greater scope for absolute symptom reduction as well as baseline-to-change dependence (including regression to the mean) inherent in change-score analyses. Clinically, it is reassuring that participants with more severe baseline symptoms did not improve less on average, although larger absolute improvement does not necessarily imply higher remission rates. In other words, patients with more severe baseline symptoms are also expected to benefit from the combined ketamine-psychotherapy protocols. Longer treatment windows may reflect maintenance, and/or logistical factors rather than higher therapeutic intensity [36]. This notion may explain the smaller average improvement in patients with longer treatment duration, also taking number of sessions into account. Although this inverse association may appear counterintuitive at first, it may reflect confounding by indication, logistical spacing, or maintenance-oriented treatment protocols rather than a true disadvantage of longer treatment duration. In prior psychotherapy research, fewer sessions have sometimes been associated with better outcomes, which may reflect earlier improvement and earlier treatment termination rather than a true benefit of reduced therapy dose [63]. Our results – shorter duration but more psychotherapy sessions being associated with greater improvement – may therefore point toward a denser, more compact delivery, although confounding by indication remains possible (e.g., more complex or severe cases receiving longer courses). Prior work suggests potential benefits of repeated ketamine dosing for PTSD symptom reduction [24, 38]. In our pooled dataset, ketamine exposure clustered at a few low numbers of sessions, and the per-session association with improvement was modest and less consistent across analytic specifications. Evidence comparing different administration routes remains limited and largely study-confounded; accordingly, our findings are consistent with prior reviews concluding that the current evidence does not favor one route over another [38]. Regarding demographic characteristics, in an exploratory analysis, Moore et al. [33] reported better outcomes with combined ketamine plus psychotherapy in younger females and with ketamine alone in older males. In other studies, such as Karatzias et al. [63], females improved more on the Impact of Event Scale, but no association of age or sex with pre-post change on the PCL or CAPS was reported. In light of the broader trauma-focused psychotherapy evidence, male sex and older age have been associated with poorer response among several baseline characteristics [66]. In our sex-stratified IPDMA, no distinct patterns for moderators of treatment effects between females and males were reported. Limitations and Strengths Several limitations should be considered. First, outcomes were based on self-reported PTSD symptom severity (PCL-5), and standardized diagnostic confirmation (e.g., CAPS-5) was not consistently available. Second, the evidence base was dominated by non-randomized, open-label and case series designs, and reporting of study and participant characteristics was incomplete in some studies. Third, there was considerable clinical and methodological heterogeneity across studies; while we modeled clustering via study-level random intercepts, between-study heterogeneity remained nontrivial (ICC ≈ 0.15), indicating that unmeasured study-level factors (e.g., protocol differences, setting, patient mix, and assessment schedules) may have influenced outcomes and that coefficients represent average associations across heterogeneous studies. Accordingly, findings from the primary model should be interpreted cautiously, particularly as most included studies were rated as poor quality and most treatment-related associations were attenuated in the quality-restricted sensitivity analysis. Post-treatment assessment time points also differed across studies, further contributing to the observed heterogeneity. Moreover, psychotherapy-related benefits may be more strongly reflected in the durability of treatment effects than in immediate post-treatment change, which may not have been fully captured by our primary outcome definition. Given the predominance of observational designs, coefficients should be interpreted as associations rather than causal effects, and residual confounding by unmeasured individual-level factors cannot be excluded. Moreover, because participants were analyzed only within combined ketamine-psychotherapy protocols, and no uniform control condition was modeled across studies, our data do not permit conclusions about the relative contribution of ketamine versus psychotherapy to the observed symptom change. Treatment length was generally more consistently defined than number of psychotherapy sessions, which may vary in definition and format across studies, potentially contributing to differential measurement errors and limiting attribution to specific therapeutic techniques. Route of administration effects are likely study-confounded and statistically unstable due to sparse data in some categories (e.g., a very small IM subgroup compared to other route subgroups), limiting inferences about administration routes. Besides, there may be some collinearity between route of ketamine administration and number of ketamine sessions; however, when we repeated the analysis excluding the route of administration, results did not change (data not shown). Fourth, selection bias cannot be excluded because IPDMA did not include data from all eligible studies due to individual participant data access restrictions. In addition, one large cohort contributed approximately half of all participants, which increases precision but may weight pooled estimates toward that cohort’s treatment setting. Moreover, all included studies were conducted in North America, and no evidence was available from low- or middle-income countries, which may limit the generalizability of the findings to settings outside high-income countries. Fifth, several potentially relevant predictors were not consistently available across studies, including ketamine dosing parameters, psychosocial/clinical factors (e.g., social support, comorbid conditions) that have been linked to response in trauma-focused psychotherapy [66], and process/context variables (e.g., acute subjective experience and its interpretation, therapeutic process factors, set/setting), proposed as potential predictors in adjacent psychedelic-assisted therapy research [67], limiting our ability to evaluate moderators beyond session counts, administration route, and overall treatment duration. This also precluded a meaningful dose-stratified (e.g., high-versus-low dose) analysis. Finally, predictor evidence in available literature often refers to short-term outcomes, highlighting the need for longer follow-up and replication [67]. A key strength of this study is the IPDMA approach, which allowed harmonization of variables across heterogeneous protocols and enabled examination of multiple patient- and treatment-level predictors while accounting for clustering by study through random intercepts. To our knowledge, this is the first IPDMA to examine predictors of PTSD symptom improvement in combined ketamine-psychotherapy interventions. By pooling 12 studies (n = 533), we assembled one of the largest available datasets on combined ketamine-psychotherapy approaches for PTSD to date, increasing statistical efficiency compared with aggregate analyses and allowing sex-stratified and quality-restricted sensitivity analyses. Outlook and Future Directions PTSD remains a highly burdensome condition, and optimizing scalable interventions is a priority [2]. Consistent with calls to move beyond average effects, our IPDMA contributes initial evidence on predictors of symptom improvement after combined ketamine and psychotherapy, supporting efforts to better stratify resource-intensive combined interventions and match patients to treatment [35, 67]. Building on this, prospective, adequately powered studies should standardize and operationalize psychotherapy exposure, harmonize assessment schedules, include clinician-rated outcomes where feasible, and extend follow-up to evaluate durability. To reduce residual heterogeneity and confounding, studies should systematically assess candidate moderators (e.g., comorbidities, social support, and treatment resistance) and capture process/context variables (e.g., therapeutic process factors and set/setting) that were not consistently available in the present IPDMA dataset [66, 67], including potentially the content and dose of preparatory psychoeducation and integration work, as well as related integration-focused outcomes. Qualitative and mixed-methods research – especially among non-responders – may clarify non-pharmacological determinants (e.g., alliance, preparation, and treatment setting) and inform implementation in routine care [68]. Replication across independent cohorts and settings is needed to establish which predictors generalize and to guide optimization of combined ketamine-psychotherapy protocols. At present, these findings should be considered hypothesis-generating and do not warrant changes to clinical guidelines. Nevertheless, where combined ketamine-psychotherapy approaches are already used in practice, they may inform clinical decision making, including careful attention to baseline PTSD severity and to the structure and density of psychotherapy delivery. Overall, this first IPDMA of combined ketamine and psychotherapy for PTSD suggests that higher baseline severity was the most robust predictor of symptom improvement, whereas associations with treatment-related factors were less consistent and require confirmation in prospective studies using more standardized protocols. Conclusion In this individual participant data meta-analysis of combined ketamine and psychotherapy for PTSD, higher baseline PTSD severity emerged as the most robust predictor of symptom improvement. Associations with treatment-related factors were less consistent and should be interpreted cautiously given the predominance of poor-quality and heterogeneous studies. Overall, these findings provide preliminary guidance for future protocol optimization and underscore the need for prospective studies with more standardized interventions and longer follow-up. Statement of Ethics A statement of ethics is not applicable because this study is based exclusively on published literature. Conflict of Interest Statement K.W. is the CEO of Revitalist Lifestyle and Wellness Ltd. T.M.M. and S.S.A. have financial interests, including equity ownership and receipt of consulting fees from Sama Therapeutics. E.S. has received educational grants, consulting fees, and lecture honoraria from Janssen Cilag, Lundbeck, Angelini, Otsuka, Servier, Recordati, Vifor, Sunovion, Schwabe, and Mepha. G.S. has served as a consultant for Dexcel Pharma, HLS Therapeutics, Lundbeck/Otsuka, Roche, Saladax and Thermo Fisher and has received speaker fees from HLS Therapeutics, Lundbeck, OM Suisse and Saladax. He holds shares of Maplight Therapeutics and Medincell. All other authors declare no competing interests. Funding Sources No funding was provided. Author Contributions Conceptualization, investigation, visualization, and methodology: J.R. and G.S.; formal analysis and supervision: G.S.; data curation: J.R., M.B., and G.S.; resources: T.M.M., K.W., S.A., H.M., M.K., S.S., V.W.L.T., and S.D.; writing – original draft: J.R.; writing – review and editing: J.R., T.M.M., K.W., S.A., H.M., M.K., S.S., V.W.L.T., S.D., M.B., B.K., E.S., S.O., and G.S.; project administration: J.R.; funding acquisition: all authors approved the final version of the paper for submission.