A pattern I’ve been tracking across my patient panel for years now crystallized again last month. A man in his late forties, Spanish-speaking, chronic schizophrenia well-controlled on clozapine, referred to me primarily for alcohol use disorder. Three sessions in, the trauma history surfaces. Childhood violence, a near-death experience as a young adult. Classic PTSD symptom cluster running quietly underneath the psychosis like a second engine. When I reached out to coordinate trauma-focused care, the answer from the referring team was familiar: he’s not a good candidate, the psychosis makes it too risky. That response is common. It has also, as of this month, lost its clinical justification.

Published in The Lancet Psychiatry this week, the STAR trial is the largest multicentre randomized controlled trial of integrated trauma-focused CBT for people with co-occurring psychosis and PTSD. The signal is unambiguous: the intervention was safe, highly acceptable, and effective in this population. Therapy disengagement ran at just 6.5 percent, a figure that would be remarkable in any psychiatric trial and is striking in a population that clinical lore has always treated as inherently difficult to retain. The field has been making an assumption about these patients’ fragility. The data says that assumption was wrong.

A Systematic Exclusion With No Evidentiary Basis

The exclusion of psychosis patients from trauma-focused therapy trials predates any formal evidence that such exclusion was warranted. A systematic review of PTSD psychotherapy RCTs identified psychosis as the single most common exclusion criterion across CBT, EMDR, and group psychotherapy trials. Not because trauma-focused interventions were tested and found harmful in psychosis patients. Because they were never tested at all. The field built a wall around a population and called the wall safety.

The consequence is a treatment gap that sits on top of a substantial disease burden. Analysis of the 2007 Adult Psychiatric Morbidity Survey found that over 30 percent of individuals with PTSD in England showed co-occurring psychotic symptoms, with 78.5 percent of PTSD cases carrying at least one comorbidity. These are not rare edge cases being seen in academic centers. They are a significant portion of the patients cycling through community mental health services, showing up in substance use programs, and getting triaged in emergency departments. In my addiction psychiatry practice, dual-diagnosis presentations involving psychosis and trauma history are the rule, not the exception.

What made clinicians so cautious? The working theory was that trauma-focused therapy, particularly exposure-based approaches, would destabilize psychotic symptomatology, worsen positive symptoms, or precipitate crisis. A 2015 randomized trial comparing trauma-focused treatment to a waitlist in patients with lifetime psychotic disorder and chronic PTSD found no symptom exacerbation and no adverse events in the treatment arm. That signal was available. It did not substantially move practice.

What the Trial Design Tells Us About Sequencing

STAR’s pragmatic, multicentre design is as instructive as its outcome. The trial did not wait for psychosis to fully remit before initiating trauma work. The integration of trauma-focused therapy with CBTp across a nine-month treatment course directly challenges the implicit sequencing model that has governed dual-diagnosis care: stabilize the psychosis first, address the trauma later. In practice, “later” rarely arrives. Patients disengage, life circumstances shift, and the trauma layer gets managed symptomatically with medication rather than therapeutically.

From a neuropharmacological standpoint, this sequencing assumption has always been fragile. Trauma and psychosis share overlapping dysregulation in the hypothalamic-pituitary-adrenal axis and dopaminergic circuits. Treating psychotic symptoms in isolation while leaving the trauma architecture intact means working against a system that keeps re-activating the very stress responses antipsychotics are trying to modulate. The STAR results are consistent with that mechanistic picture. Treating both simultaneously, in an integrated protocol, produced low dropout and meaningful outcomes, not destabilization.

Having treated thousands of patients at the intersection of trauma, mood disorders, and psychosis, including through guided ketamine therapy where trauma content frequently emerges in the therapeutic process, I recognize the clinical caution that shaped the old exclusion logic. These patients do require careful titration, close monitoring, and coordinated care. The STAR protocol accounts for that. The 6.5 percent disengagement rate across a multicentre pragmatic trial reflects a protocol that was built by people who understood the population, not one designed for convenience.

The Design Gap the Field Now Needs to Close

The immediate implication for trial designers is this: psychosis as a blanket exclusion criterion in PTSD intervention trials is no longer defensible as a safety rationale. It is a recruitment convention, inherited from an era when nobody had tested the alternative. Future PTSD trials that exclude active psychotic disorder owe their sponsors and their IRBs an affirmative argument for why, grounded in mechanism or prior harm signal, neither of which the literature currently supports.

The harder design question involves endpoints. Standard PTSD trial endpoints, PCL-5 scores, CAPS-5 clinician ratings, were developed and validated largely in populations without active psychosis. In dual-diagnosis patients, the symptom overlap between PTSD hyperarousal and psychotic activation, between dissociation and negative symptoms, creates measurement noise that most power calculations do not account for. The STAR trial’s acceptability data, particularly that dropout figure, suggests these patients are highly motivated and retain well when they feel the protocol is built for them. That retention profile makes them a valuable population for longer-duration trials. Getting the endpoint measurement right is the next methodological problem the field needs to solve.

Spanish-speaking patients with co-occurring psychosis and trauma, a population I see regularly across my telehealth practice spanning 36 states, remain almost entirely absent from the evidence base. STAR was conducted in UK NHS services. The generalizability question to minoritized and linguistically isolated populations is open, and it is not a small question.

I will be watching for the STAR team’s secondary outcome data and any follow-up analyses on which psychosis subgroups drove the strongest treatment response. That subgroup signal will matter enormously for how sponsors think about eligibility criteria in the next generation of integrated trauma-psychosis trials.

References

  1. The Lancet Psychiatry — “Trauma-focused therapy integrated with cognitive behavioural therapy for psychosis for people with post-traumatic stress disorder and psychosis (the STAR trial)”
  2. EurekAlert — STAR trial results: 6.5% therapy disengagement rate in psychosis/PTSD population
  3. PubMed — Psychosis as the most common exclusion criterion in PTSD psychotherapy RCTs
  4. PMC — Prevalence of psychotic symptoms in PTSD: analysis of the 2007 Adult Psychiatric Morbidity Survey
  5. PubMed — Trauma-focused treatment in lifetime psychotic disorder and chronic PTSD: no symptom exacerbation or adverse events
+ posts

Dr. Leonardo Vando, MD is Principal Investigator of Psychiatry at Vando Medical Services.