This Week in Psychiatry — Aug 10, 2026

Welcome to This Week in Psychiatry. This week we're covering seven notable papers spanning psychedelic and trauma-focused psychotherapy trials, the hard problem of scaling and prevention when interventions are simplified or delivered universally, and new biological work on psychosis in the brain. Let's dive in.

We start with two randomised trials that test intensive, targeted therapies in difficult populations. In Nature Medicine, Rucker and colleagues report a double-blind, placebo-controlled feasibility trial of a single twenty-five milligram dose of psilocybin with psychological preparation, dosing support and integration, delivered inside a single National Health Service site in England [1]. Sixty adults with major depressive disorder who had not responded adequately to at least two antidepressant treatments were randomised one to one, and fifty-nine of sixty completed all follow-up depression ratings, which is remarkable retention for this population. The primary outcomes were feasibility measures — recruitment, retention and variance estimation — but the between-group difference on the Montgomery-Åsberg scale at three weeks favoured psilocybin by roughly ten points, a large effect that was still present at six weeks. Adverse events were non-serious in both arms, though more numerous with psilocybin. The important framing for practice is that this is a feasibility study supporting a future confirmatory trial, not a licensing trial — but it does demonstrate that psilocybin-assisted therapy can be delivered within a public healthcare system rather than only in specialist research units. Alongside that, JAMA Psychiatry publishes a multisite trial of Impact of Killing therapy, an individual psychotherapy for veterans whose post-traumatic stress is bound up with having killed in war or feeling responsible for others' deaths [2]. One hundred veterans across three Veterans Affairs systems were randomised to Impact of Killing or present-centred therapy. Both groups improved in psychosocial functioning, and on the World Health Organization quality of life measure there was no significant difference between the two treatments. But on the Sheehan Disability Scale, Impact of Killing produced significantly greater improvement in work, social and family functioning, a moderate effect, and it also outperformed present-centred therapy on post-traumatic stress and moral injury symptoms. So for the veteran who describes guilt about killing rather than fear of danger, a moral-injury-focused protocol appears to add something beyond good supportive care.

The second theme is less comfortable: what happens when we try to scale psychological treatment or push it upstream into prevention. The Lancet Psychiatry reports a Bayesian component network meta-analysis of individual participant data from thirty trials and more than ten thousand participants receiving task-shared psychosocial interventions for depression and anxiety, delivered by non-specialists [3]. By dismantling interventions into their constituent elements, Papola and colleagues found that strengthening social support carried the largest incremental benefit, followed by behavioural activation and problem management. More provocatively, relaxation was associated with a clearly detrimental effect, and cognitive reframing with a less certain but still detrimental signal, when added to these brief non-specialist packages. Component effects also varied by baseline severity and sociodemographic characteristics, and the authors provide a free web application for personalised estimates. That doesn't mean relaxation or cognitive restructuring are harmful in expert hands — it means that in lay-delivered, brief formats, loading on social support, activation and practical problem-solving may be the better bet. Two other trials this week show scaling failing outright. In Psychotherapy and Psychosomatics, the GiVE3 trial randomised one hundred and thirty-five patients with psychosis across three National Health Service sites to guided self-help cognitive behavioural therapy for distressing voices delivered by briefly trained assistant psychologists, plus usual care, versus usual care alone [4]. Voice-related distress differences at sixteen and twenty-eight weeks favoured the intervention numerically but were not statistically significant, and the authors conclude the intervention was not clinically effective when delivered by briefly trained staff. And in the British Journal of Psychiatry, the AWARE cluster randomised trial enrolled over twelve thousand Year Nine pupils across one hundred and fifty-three English secondary schools to compare Youth Aware of Mental Health, the Mental Health and High School Curriculum Guide, and usual practice [5]. Youth Aware of Mental Health produced no improvement in depressive symptoms, and a notable minority of schools could not deliver it at all. The Guide did modestly increase intended help-seeking. But at longer-term follow-up, depressive symptom scores were higher in both intervention arms than in control — a possible iatrogenic signal that led the investigators to advise further investigation before wider rollout in English schools. Taken together with the component analysis, the message is that fidelity, therapist training and content selection are not administrative details; they determine whether a scaled intervention works, does nothing, or possibly harms.

Finally, two papers in Molecular Psychiatry reframe psychosis biologically. Kubota and colleagues scanned thirty-seven patients with late-onset psychosis and forty-seven age-matched controls using amyloid and tau positron emission tomography with a tracer sensitive to a broad range of tau pathologies [6]. Roughly a third of patients were amyloid-positive and about two-thirds were tau-positive, compared with two percent and fifteen percent of controls. Increased tau retention in the parietal cortex persisted even when analysis was restricted to amyloid-negative patients, and among amyloid-positive patients greater parietal tau burden tracked with worse frontal executive performance. Practically, this strengthens the case for taking cognitive assessment and neurodegenerative work-up seriously in first-episode psychosis in later life. In the same journal, Connolly and colleagues targeted the default mode network with theta burst stimulation in schizophrenia [7]. In a five-session accelerated course of continuous theta burst stimulation, processing speed improved in the schizophrenia group, reductions in default mode network connectivity correlated with those gains, and younger participants benefited more. The samples are small — twenty-nine participants in the multi-session study — so this is early proof-of-concept for personalised network-targeted stimulation of cognitive deficits, not a clinical service yet.

If you only have time for one paper this week, make it the Lancet Psychiatry component network meta-analysis of task-shared psychosocial interventions [3]. It is the paper most likely to change what you actually put inside a brief intervention or supervise a non-specialist to deliver, and its detrimental signal for relaxation deserves your attention.

Here are the key takeaways from this week in Psychiatry. First, psilocybin-assisted therapy was feasible and showed a large symptom difference in a public healthcare setting, but this was a feasibility trial and confirmatory data are still pending. Second, for veterans distressed by killing, a moral-injury-focused therapy beat present-centred therapy on disability, post-traumatic stress and moral injury, though not on quality of life. Third, in brief non-specialist interventions, prioritise social support, behavioural activation and problem management, and be cautious about relaxation. Fourth, two scaling attempts failed: guided self-help for voices delivered by briefly trained assistants was not effective, and universal school programmes produced no depression benefit with a longer-term worsening signal. Fifth, in late-onset psychosis, tau pathology is common enough that neurodegeneration belongs in your differential.

That's your roundup for This Week in Psychiatry. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if you have a paper you have been meaning to read, upload the PDF, or paste any link, at audioscholar dot C C. We will turn it into audio like this one, in any of thirty-one languages.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.