Not yet independently audited
This page was written from verified sources but has not yet passed the independent citation audit. Treat its claims as provisional.
Open questions — post-traumatic stress disorder¶
Last curated: 2026-10-23 (OQ-1, OQ-2, OQ-4, OQ-12, OQ-15 and OQ-16 follow-ups checked; stable question IDs retained). All cited PMIDs and NCT IDs were retrieved live from PubMed E-utilities and the ClinicalTrials.gov v2 API in the sessions that wrote them.
17 questions (10 Tier 1 + 2 added by audit, 5 Tier 2) and 21 unconnected-dot junctions. Tier 1 questions could change practice and are designable now. Tier 2 questions require measurement, methods or a preceding answer. Stable IDs are preserved.
Tier 1 — practice-changing and designable now¶
OQ-1. Which therapy maximizes retention-adjusted benefit?¶
Adult networks rank symptom efficacy, while PTSD CBT dropout can exceed controls and routine-care rates of 38–51% motivated a 916-person direct analysis (Mavranezouli 2020, PMID 32063234; Carpenter 2018, PMID 29451967; Harper 2026, PMID 41926191). An IPD network meta-analysis can jointly model assignment, dose, dropout, post-dropout outcome and function. Sharpened 2026-10-09: routine-care data suggest the largest loss may happen before any therapy is assigned. In a post-disaster clinic, 33.5% discontinued within a year, two-thirds of them within 90 days, and only 17.6% ever reached structured psychotherapy (Hori 2026, PMID 42780894). Low-intensity first steps saved clinician time but raised dropout from 26% to 43% (Roberts 2026, PMID 42832930). Retention-adjusted benefit needs a denominator that starts at referral, not at randomization to a therapy.
Sweep 2026-10-16: a full-text meta-regression identifies inconsistent completion definitions across CPT, PE and EMDR and exploratory associations with treatment frequency (Horton 2026, PMID 42504047, Tables 1–2). The proposed IPD comparison also needs a harmonized completion endpoint; study-level pacing associations do not establish a causal retention strategy.
Sweep 2026-10-23: inside the US Military Health System, 51.3% of 3,459 patients with PTSD initiated prolonged exposure, cognitive processing therapy or EMDR, 25.7% received three sessions in 90 days, and 7.5% received eight sessions in 14 weeks (McLean 2026, PMID 42906525). Both session counts are the paper's thresholds for a minimally adequate dosage. These figures are from the PubMed abstract; full text was not retrieved. Loss continues after initiation, which keeps the denominator problem in front of the efficacy ranking. A harmonized completion endpoint is still required.
OQ-2. Do effects transport across DSM-5 PTSD, ICD-11 PTSD and ICD-11 CPTSD?¶
The systems select non-identical populations, and ITQ-defined CPTSD is rarely stratified in treatment trials (Cloitre 2018, PMID 30178492; Brewin 2017, PMID 29029837). Existing item-level trial data could reconstruct strata.
Sweep 2026-10-16: a same-sample comparison in 483 Syrian refugees finds probable PTSD in 26.9% by PCL-5 cutoff versus 21.9% by ITQ ICD-11 algorithm (Alpay 2026, PMID 42438360, Table 1). This supports a measurement crosswalk problem, not a treatment-response interaction: all measures were self-report. Regional coverage remains incomplete because SciELO and LILACS could not be searched (HTTP 403).
Sweep 2026-10-23: in one quota sample of 5,070 adults across five European countries, clinical checks on the ITQ roughly halved the share meeting ICD-11 PTSD or complex PTSD requirements, from 19.2% to 9.8% of all participants (Hyland 2026, PMID 42905693; full text). That is instrument behaviour in a quota sample. It is not a treatment-by-diagnosis interaction, and it is not a national prevalence. SciELO and LILACS remained unreachable (HTTP 403).
OQ-3. Does CPTSD require phase-based treatment before trauma-focused work?¶
Complex-event component evidence supports benefit but does not settle fixed sequencing (Coventry 2020, PMID 32813696). A pragmatic non-inferiority/superiority trial should compare immediate trauma focus with criterion-triggered stabilisation.
OQ-4. Which prospectively defined phenotype predicts prazosin benefit?¶
Three positive trials conflict with the 304-person PACT null result (Raskind 2003, PMID 12562588; Raskind 2007, PMID 17069768; Raskind 2013, PMID 23846759; Raskind 2018, PMID 29414272), and two meta-analyses still rate prazosin effective for trauma-related nightmares (Yücel 2020, PMID 31855732; Zhang 2022, PMID 35661755). The leading candidate marker is pretreatment standing systolic blood pressure, where each 10 mm Hg predicted an additional 14-point CAPS reduction under prazosin but not placebo in a secondary analysis (Raskind 2016, PMID 27320368), contested in a comorbid alcohol-use-disorder sample (Manhapra 2019, PMID 30143236). Status as of the audit's live registry check on 2026-09-02: a phase 3 prazosin-versus-placebo trial designed around noradrenergic biomarkers (NCT03539614, n=87) records an actual primary completion date of 1 January 2026 and is active-not-recruiting, with no results publication indexed in PubMed. The question is therefore no longer "is anyone testing this" but "what did that trial find" — the first item to check at the next sweep. Sweep 2026-10-09: still unreported. The record was last updated on 14 August 2026 and remains active, not recruiting, with no posted results and an estimated study completion of 31 December 2026. PubMed returns nothing for [NCT03539614](https://clinicaltrials.gov/study/NCT03539614){target="_blank" rel="noopener"}. Nine months have passed since primary completion.
Registry re-check 2026-10-16: the prazosin biomarker trial still has no posted results (NCT03539614); the exact-ID PubMed search again returns zero records. This is an absence in those sources only.
Registry re-check 2026-10-23: still unreported in PubMed and on ClinicalTrials.gov. The record remains active, not recruiting, last updated 14 August 2026, with no results posted (NCT03539614). The exact-ID PubMed search returned zero records. WHO ICTRP returned HTTP 502 on every query this session, so this is not a worldwide absence.
OQ-5. Can independent MDMA replication address expectancy and therapist effects?¶
Phase 3 effects were d=.91 and d=.70, but the FDA declined approval in August 2024 and required another phase 3 trial (Mitchell 2021, PMID 33972795; Mitchell 2023, PMID 37709999; Wolfgang 2025, PMID 39741438; Morland 2026, PMID 41820235). A 2026 meta-analysis of eight randomized trials estimates SMD −1.19 (95% CI −1.95 to −0.42) with most studies at high risk of bias in outcome measurement (Fares-Otero 2026, PMID 41825162), and the 2026 state-of-the-science review names blinding, absent active comparators, absent head-to-head therapy-model comparisons, inadequate safety monitoring and narrow generalizability as the specific design debts. Independent funding, expectancy measurement and safety adjudication are designable now. Sweep 2026-10-09: the integrity question is now partly documented, not just alleged. In November 2024 Psychopharmacology retracted three sponsor-authored pooled phase 2 analyses after protocol violations at the MP4 site, which the authors knew of and did not disclose (Mithoefer 2024, PMID 39126501). Two further secondary analyses carry expressions of concern (PMIDs: 39627911, 39627875). A 2026 meta-analysis of seven RCTs again finds benefit, and it names "a common sponsor for all trials" as a limitation (Green 2026, PMID 42708350). A sharper sub-question can be answered from existing reviews: does the pooled MDMA effect survive excluding MP4-site data?
Partly answered 2026-10-16: the new living synthesis excludes the pooled secondary reports containing MP4 data in supplementary Table 10 and still finds a PTSD symptom benefit in its six primary trials (Sevchik 2026, PMID 42860436, Figure 1 and Table 1). Thus MP4-containing pooled reports are not necessary for the signal in this review. Overall certainty remains low and every included trial used the MAPS treatment model. Independent replication, expectancy and safety remain open; the Green and Fares-Otero review-specific inclusion checks remain pending.
OQ-6. Which early intervention prevents PTSD without over-treating recovery?¶
Single-session debriefing is ineffective and potentially harmful, while targeted early CBT has supportive evidence (Rose 2002, PMID 12076399; Bryant 2007, PMID 21352105). A risk-stratified stepped-care trial can estimate absolute benefit and overtreatment. Sharpened 2026-10-09: the pharmacological arm of this question now has two syntheses that disagree. Hydrocortisone shows RR 0.21 (95% CI 0.05–0.89) in severely medically ill samples (Astill Wright 2019, PMID 31819037), but RR 0.47 (95% CI 0.09–2.38, very low certainty) in adults with acute symptoms at trauma centres and emergency departments (Bertolini 2024, PMID 38767196). Whether hydrocortisone prevents PTSD only after critical illness, or not at all, is now a designable population-stratified question.
OQ-7. Do military/civilian differences represent true effect modification?¶
Exposure meta-regression found larger effects in civilian/refugee than military studies, but study-level confounding is substantial (McLean 2022, PMID 34954460). Harmonized IPD can test within-trial interactions.
OQ-8. Does preference matching improve completion and function?¶
Preference matters to uptake, but assignment-by-preference can confound treatment effects (PMID 32074319; Harper 2026, PMID 41926191). A doubly randomized preference design is feasible.
OQ-9. Which pediatric TF-CBT components are necessary?¶
TF-CBT has the largest evidence base—52 of 70 pediatric RCTs—and g=1.06 versus passive control (Hoppen 2025, PMID 39630422). Component or sequential multiple-assignment designs could reduce burden while preserving benefit.
OQ-10. What adverse-event standard should PTSD psychotherapy trials use?¶
Adverse-event data were absent from most trials in a major review (Cusack 2016, PMID 26574151). A consensus taxonomy and mandatory arm-level reporting are achievable now.
OQ-16. Does psilocybin do anything for PTSD under a control condition?¶
The 2022 statement that no psilocybin PTSD trial existed (Khan 2022, PMID 35711024) is no longer current: two uncontrolled trials have since been published — an open-label phase 2 study in 22 participants reporting CAPS-5 change of −29.9 at week 4 (McGowan 2026, PMID 40883964) and an open-label veteran pilot in 12 participants reporting −27.5 points (95% CI 19.9–35.1) at one month (Armstrong 2026, PMID 42533157, NCT05554094). The dated gap as of 2026-09-02 is precise: 34 participants total, no control condition, and no randomized psilocybin PTSD trial published. Given that open-label PTSD trials routinely produce large within-group change, a randomized comparison is the only design that can distinguish drug effect from expectancy and regression. Sweep 2026-10-09: still no randomized psilocybin PTSD trial is published in PubMed. But the design gap has closed in the registry. ClinicalTrials.gov v2 lists randomized trials including a 300-participant phase 2b/3 of COMP360 against a 1 mg comparator (NCT07570654, recruiting) and a 40-participant placebo-controlled phase 2 in PTSD with alcohol use disorder (NCT06853912, active not recruiting, estimated primary completion November 2026). ISRCTN and EU CTIS returned none, and WHO ICTRP could not be queried. The question is now when these trials report, and whether low-dose comparators preserve blinding better than inert placebo.
Registry re-check 2026-10-16: no posted results in the four tracked randomized psilocybin trials; the PubMed randomized-design search returned reviews, protocols and studies in other populations, without a published randomized PTSD efficacy readout identified. ISRCTN psilocybin searches also produced no PTSD-directed trial. WHO ICTRP was partly reachable for a broad query, but drug-specific requests returned HTTP 403; no registry-wide absence is claimed.
Sweep 2026-10-23: still no published randomized psilocybin PTSD efficacy trial in the PubMed search. A media systematic review found no PTSD trial among psilocybin trials published through 10 October 2024, while 74 of 172 media articles (43%) named PTSD (Madden 2026, PMID 42900476; full text, Table 3). That cutoff predates the two uncontrolled 2026 trials already cited. The four tracked randomized records still have no posted results (NCT07570654, last update 21 October 2026, still recruiting, estimated primary completion November 2028; NCT06853912, still active not recruiting, estimated primary completion November 2026). The open-label veteran record is now completed in the registry, actual enrollment 12, last update 16 October 2026, with no results posted (NCT05554094), which agrees with the published pilot and still has no control arm. ISRCTN queries for psilocybin and PTSD returned no PTSD-directed efficacy trial. WHO ICTRP returned HTTP 502 on every query, so it was not searched. No registry-wide absence is claimed.
OQ-17. Can a rapid-acting drug effect be separated from the therapy package it is delivered inside?¶
The MDMA phase 3 programme deliberately bundled drug with a manualized therapy (Mitchell 2021, PMID 33972795; Mitchell 2023, PMID 37709999), which is exactly why attribution has been contested. A phase 2 trial of the neuroplastogen TSND-201 with no psychotherapy component found a CAPS-5 advantage over placebo (LS mean difference 9.64; 90% CI −16.48 to −2.80) in 65 participants (Jones 2026, PMID 41706459, NCT05741710), and ketamine trials without a therapy package disagree with each other (Feder 2021, PMID 33397139; Abdallah 2022, PMID 35046508). A factorial drug × therapy design in one population would answer directly what the current literature only permits inferring across incomparable trials. Sweep 2026-10-09: a 300-participant, two-dose, placebo-controlled phase 3 of TSND-201 was registered on 23 September 2026 (NCT07836647). It is still drug-only, so it will test replication, not the factorial question. The ketamine-plus-PE trial posted registry results with week-22 CAPS-5 of 18.8 against 19.0 for midazolam-plus-PE, no analysis, and heavy attrition (NCT04560660). That is weak evidence that adding ketamine to an effective therapy adds little, and the opposite pattern to the drug-alone TSND-201 signal. A factorial design remains the only clean test.
Tier 2 — depends on methods or prior answers¶
OQ-11. Is moral injury separable enough to warrant a distinct treatment pathway?¶
The 116-study review found no consensus definition or gold-standard measure (Griffin 2019, PMID 30688367). Measurement must separate exposure, appraisal and impairment before comparative treatment testing.
OQ-12. Can biological measures predict individual treatment response?¶
HPA, autonomic and imaging differences are heterogeneous group associations, not clinical predictors (Pitman 2012, PMID 23047775; Campbell 2019, PMID 30779926). Reliable assays and external validation are prerequisites.
Sweep 2026-10-23: a peripheral-blood gene-expression study of veterans in psychotherapy split complete three-timepoint profiles into 6 responders and 11 non-responders, while a trajectory class in the same paper is 8 responders and 11 non-responders (Xu 2026, PMID 42897160; full text). The title says the signatures predict response. The sample that was profiled does not externally validate a clinical predictor.
OQ-13. Are categorical CPTSD and dimensional DSO equally prognostic?¶
Latent-profile studies recover classes across populations but model fit does not establish clinical thresholds (Karatzias 2017, PMID 27723542; Vang 2019, PMID 31984942).
OQ-14. Which outcomes define recovery beyond symptom loss?¶
Function, sleep, relationships, work, parenting and meaning are inconsistently co-primary. Trials and patient-priority studies need a core outcome set (PMID 39514500; PMID 37722204).
OQ-15. Can PTSD–pain treatment target shared avoidance without conflating outcomes?¶
Only 13 studies met criteria in a childhood-trauma/PTSD/CPTSD/chronic-pain review, and four explicitly tested the linkage (Karimov-Zwienenberg 2024, PMID 39213321).
Sweep 2026-10-23: a VA Delphi study reached its pre-set consensus on 48 of 71 items for treating chronic pain and PTSD together (Zaccari 2026, PMID 41851426; PubMed abstract, full text not retrieved). That is expert agreement inside one system. It does not test whether simultaneous care improves either outcome.
Dots not yet connected¶
| # | Dot A | Dot B | Missing junction | Powers |
|---|---|---|---|---|
| D1 | Adult therapy NMA ranks symptoms (PMID 32063234) | Direct dropout evidence exists (PMID 41926191) | Retention-adjusted comparative benefit | OQ-1 |
| D2 | ITQ identifies CPTSD (PMID 30178492) | Most trials used DSM definitions | Reconstructed diagnostic-stratum effects | OQ-2, OQ-3 |
| D3 | Complex-trauma interventions can help (PMID 32813696) | Phase-based care is widely proposed | Randomized sequencing by measured need | OQ-3 |
| D4 | Prazosin meta-analyses positive (PMIDs: 31855732, 35661755) | PACT null (PMID 29414272) | Published result of the completed biomarker trial NCT03539614 (still unreported in PubMed and ClinicalTrials.gov on 2026-10-23) | OQ-4 |
| D5 | MDMA phase 3 effects positive (PMIDs: 33972795, 37709999) | FDA required another trial (PMID 39741438) | Independent design resolving expectancy/safety | OQ-5 |
| D6 | Debriefing can harm (PMID 12076399) | Targeted early CBT can help (PMID 21352105) | Calibrated triage threshold | OQ-6 |
| D7 | Military studies dominate | Civilian/refugee effects appear larger (PMID 34954460) | Patient-level transport model | OQ-7 |
| D8 | Preferences shape uptake (PMID 32074319) | Treatments have similar active-comparator effects | Randomized preference-sensitive pathway | OQ-8 |
| D9 | Pediatric TF-CBT evidence is strongest (PMID 39630422) | Caregiver availability/safety varies | Component-by-caregiver configuration | OQ-9 |
| D10 | Psychotherapy benefit is established (PMID 26574151) | Adverse-event reporting is sparse | Core harm taxonomy | OQ-10 |
| D11 | Moral injury correlates with PTSD (PMID 33475402) | No gold-standard moral-injury outcome (PMID 30688367) | Discriminant measurement before treatment | OQ-11 |
| D12 | Autonomic group difference exists (PMID 30779926) | Therapy selection remains preference/availability-led | Prospective validated treatment predictor | OQ-12 |
| D13 | CPTSD profiles replicate (PMID 27723542) | Prognosis is continuous | Category-versus-dimension prediction | OQ-13 |
| D14 | Symptom endpoints dominate | Patient sources prioritize function and relationships | Co-designed core outcome set | OQ-14 |
| D15 | PTSD and chronic pain co-occur (PMID 39213321) | Avoidance is targeted in both fields | Factorial integrated intervention | OQ-15 |
| D16 | Psilocybin produces large uncontrolled PTSD change (PMIDs: 40883964, 42533157) | No randomized psilocybin PTSD trial is published; four tracked trials still have no posted results (NCT07570654, NCT06853912; re-checked 2026-10-23). A media review through 2024-10-10 found no PTSD trial (PMID 42900476) | The readout of a controlled comparison | OQ-16 |
| D17 | A drug-only trial shows a PTSD effect (PMID 41706459); phase 3 now registered (NCT07836647) | Ketamine added to PE shows no registry-level separation (NCT04560660) and MDMA is measured only inside a therapy package (PMIDs: 33972795, 37709999) | Factorial drug × therapy design in one population | OQ-17 |
| D18 | Pre-trauma cohorts identify predisposing neural and fear-learning markers (PMIDs: 40024495, 41103636) | Treatment selection uses none of them | Prospective marker-stratified treatment assignment | OQ-12 |
| D19 | Sequencing CBT-I before CPT helps (PMID 34265777) | Integrating CBT-I with PE does not (PMID 40488726) | Head-to-head test of sequence vs integration | OQ-3 |
| D20 | Massed and telehealth delivery are non-inferior (PMIDs: 29362795, 26243685) | Routine care still delivers weekly in-person courses with 38–51% dropout (PMID 41926191) | Implementation trial of the non-inferior format as default | OQ-1 |
| D21 | Hydrocortisone prevents PTSD after severe medical illness (PMID 31819037) | Cochrane finds hydrocortisone inconclusive in acute-symptom ED/trauma-centre adults (PMID 38767196) | Population-stratified hydrocortisone prevention trial | OQ-6 |
Pooling control¶
GAD and PTSD are neighbours, not subsets. A mixed-anxiety trial or pooled anxiety-disorder effect does not answer a PTSD question unless a PTSD stratum is reported. Comorbid depression is cross-linked to the depression condition and measured separately.