This Week in Psychiatry — Aug 24, 2026

Welcome to This Week in Psychiatry. This week we're covering 10 notable papers spanning substance use and psychosis risk, treatment personalisation across depression, eating disorders and schizophrenia, and the widening world of neurodevelopmental care and digital delivery. Let's dive in.

Let's start with cannabis, because two papers this week converge on the same uncomfortable message from very different directions. In JAMA Psychiatry, Han and colleagues analysed four years of the National Surveys on Drug Use and Health, covering more than 186,000 United States adults between 2021 and 2024. Past-year cannabis use disorder rose in both men and women, and critically, it was the moderate-to-severe end that grew — in men overall, and notably in men aged 35 to 49 and those over 50, and in women across every age band above 21. Among women with any cannabis use disorder, the proportion meeting moderate-to-severe criteria climbed from just over a third to close to half. Meanwhile alcohol use disorder severity stayed essentially flat. And among the youngest adults, those 18 to 20, roughly half to sixty percent of those with cannabis use disorder had the moderate-to-severe form. The authors' framing is blunt: this challenges the idea that cannabis carries low addictive potential [3]. Sitting alongside that, Biological Psychiatry published a large meta-analysis from Stevens and colleagues pooling 181 samples — over 18,000 cannabis users and nearly 53,000 nonusers — on age at onset of psychosis. Overall, cannabis use was associated with psychosis arriving about two and a half years earlier. But the subgroup pattern is the interesting part and it runs counter to intuition. In samples where the mean onset age was under 20, there was no difference at all. The gap widened progressively with later mean onset age — about a year and a half earlier in the 20-to-25 group, four and a half years in the late twenties, and more than six years in samples with mean onset at 30 or above [2]. The practical read: don't restrict your cannabis-and-psychosis counselling to teenagers. The middle-aged patient with escalating high-potency use is not outside the risk window, and the JAMA Psychiatry prevalence data tell you that is exactly the group in which severe cannabis use disorder is growing.

Moving to treatment, and specifically to who gets what. Psychotherapy and Psychosomatics ran two papers this week that both push on personalisation. The first is a systematic review from Wasolua and colleagues on long-term safety of ketamine and esketamine in treatment-resistant depression — 32 studies, 11 of them randomised, with follow-up from three months out to nine years. Across cardiovascular, hepatic and urinary domains they found no evidence of cumulative organ toxicity during maintenance treatment. Dissociation and sedation were common but transient, resolving within one to two hours. No treatment-emergent neurocognitive dysfunction was identified, urologic complications were rare and reversed on stopping, and — notably — no documented cases of iatrogenic addiction, drug-seeking or withdrawal for either formulation. Racemic ketamine and intranasal esketamine looked broadly equivalent. The important caveat the authors themselves raise is heterogeneity in how safety was monitored across studies, so this is reassurance rather than proof. And suicidal ideation scores fell from baseline, but attempts and completed suicides still occurred during follow-up — maintenance ketamine is not a substitute for suicide risk management [1]. The second paper from the same journal is a long-term moderator analysis of the ANTOP trial in anorexia nervosa, led by Wild. In the original trial, focal psychodynamic therapy and enhanced cognitive behavioural therapy did not differ on body mass index. This secondary analysis of 160 randomised patients, looking out to nearly six years, found that baseline psychological burden moderated the outcome. Patients with low depression severity and better mental quality of life did better with enhanced cognitive behavioural therapy, while those with high depression severity and poor mental quality of life did better with focal psychodynamic therapy. The authors are appropriately restrained — across a wide middle range the two treatments performed similarly, and the divergence appears at the extremes. This is exploratory and needs prospective testing, but it is a plausible clinical heuristic: the more depressed and burdened patient with anorexia may be better served by the psychodynamic route [6].

Staying with personalisation but shifting to how we might operationalise it, The Lancet Psychiatry published a Personal View from Garcia-Argibay and colleagues that is essentially a diagnosis of why prediction models fail. More than a hundred prediction models exist for attention deficit hyperactivity disorder, and none is in routine practice. Their argument is that the field reports discrimination metrics like area under the curve but never specifies what a clinician should actually do differently for a patient flagged as high risk. Without that mapping from output to action, an accurate model is clinically inert. They propose prediction-based decision rules, tested through target trial emulation in large observational datasets, and illustrate it with treatment intensity guided by predicted persistence and medication selection guided by predicted response [4]. For the practising clinician, this is a useful lens for appraising any algorithm pitched to your service: ask what action changes, not what the area under the curve is.

On the neurodevelopmental theme, two papers from the Journal of Child Psychology and Psychiatry. Chronis-Tuscano and colleagues ran a hybrid effectiveness-implementation randomised trial in 120 parents with attention deficit hyperactivity disorder and their stimulant-naive children aged three to eight, delivered by telehealth through urban paediatric clinics. Everyone received an integrated behavioural parent training programme; half also received a parent stimulant medication titration first. Children whose parents got medication plus parent training showed significantly steeper declines in clinical global impression ratings — a small-to-moderate effect. Parents improved faster on their own severity ratings, and reported greater gains in consistent discipline and positive parenting. Broader parenting improvements happened in both arms. The message is that treating the parent's attention deficit hyperactivity disorder pharmacologically appears to amplify what parent training can achieve in these multiplex families [5]. Alongside that, Frayman and Mandy analysed England's School Census across eight to eight and a half million pupils annually. Identified autism prevalence more than doubled, from about one and a half percent in 2016 to over three percent in 2024. The male-to-female ratio narrowed from roughly five to one down to under three to one, and the proportion of autistic pupils with identified intellectual disability fell from about fifteen percent to under nine percent. So growth is being driven substantially by recognition of girls and of those without intellectual disability. Autistic pupils were more likely to be White British and from deprived backgrounds, and the cohort structure means further increases are already locked in [9].

Finally, two papers on cognition in schizophrenia and one on scalable digital care. In Molecular Psychiatry, Li and colleagues randomised 69 patients with schizophrenia to a single session of active or sham high-definition transcranial direct current stimulation over the left dorsolateral prefrontal cortex. Working memory accuracy improved and reaction times fell in the active group, and combined transcranial magnetic stimulation with electroencephalography showed reduced excitability in hippocampus and entorhinal cortex, with a mediation model implicating a prefrontal-hippocampal-entorhinal cascade [8]. That is mechanistically interesting but it is one session in 69 patients, and Schizophrenia Bulletin supplies the sobering context. Harvey and colleagues systematically reviewed 104 datasets and found that cognitive performance relates strongly to functional capacity — what a patient can do in a structured assessment — but only weakly-to-moderately to real-world everyday functioning, with just two large effect sizes across 43 studies [10]. So a cognitive gain on a two-back task is a long way from a change in someone's life, and the review argues for multimodal, person-centred approaches to bridge that gap. And in PLOS Medicine, Johnson and colleagues randomised 825 adults with chronic medical conditions across 13 countries to a self-directed digital programme, the same programme with weekly non-clinician telephone check-ins, or waitlist. Anxiety and depression scores improved by about three points with human support against control, and the self-directed arm improved comparably — with no significant difference between the two intervention arms on any outcome. Adding human support did not add measurable benefit, which is encouraging for scalability, though the waitlist design and predominantly female, highly educated sample temper the conclusion [7].

If you only have time for one paper this week, make it the JAMA Psychiatry trends analysis on alcohol and cannabis use disorders [3]. It reframes who you should be screening — middle-aged and older adults, not just the young — and the severity data give you something concrete to say to patients who believe cannabis is not addictive.

Here are the key takeaways from this week in Psychiatry. Severe cannabis use disorder is rising across the adult age range while alcohol use disorder severity holds steady, so routine cannabis screening should extend well beyond young adults. The link between cannabis and earlier psychosis onset is robust and, unexpectedly, largest in later-onset samples — counsel across the lifespan. Long-term ketamine and esketamine maintenance shows no signal of cumulative organ toxicity or iatrogenic addiction, but suicide risk management remains a separate obligation. In anorexia nervosa, baseline depression and mental quality of life may help you choose between cognitive behavioural and psychodynamic therapy at the extremes. Treating a parent's attention deficit hyperactivity disorder with medication before parent training improved child outcomes. And remember that cognitive gains in schizophrenia translate only weakly into real-world functioning — plan interventions accordingly.

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 think of one colleague — in any specialty — who never has time to keep up with the literature. Tell them about AudioScholar: a free ten-minute weekly for every specialty, to listen to in any podcast app, or to read at audioscholar dot C C.

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