This Week in Psychiatry — Sep 14, 2026

Welcome to This Week in Psychiatry. This week we're covering 10 notable papers spanning severe mental illness treatment selection, personalised and pharmacogenomic prescribing, and the shifting landscape of child and adolescent diagnosis. Let's dive in.

We start with treatment of psychotic and bipolar illness, where three papers push in useful and slightly uncomfortable directions. In The Lancet Psychiatry, Montag and colleagues report a single-centre, assessor-blinded randomised trial from Berlin testing a manualised modified psychodynamic psychotherapy for schizophrenia added to treatment as usual, against treatment as usual alone, in 130 outpatients with schizophrenia or schizoaffective disorder [1]. Participants in the therapy arm received a minimum of thirty sessions, and the primary outcome was psychosocial functioning on the Mini-ICF-APP, assessed at 24 and 36 months. Functioning favoured the psychotherapy group at both time points, and notably the advantage was slightly larger at 36 months than at the end of treatment, suggesting benefit that persisted after therapy ended. Adverse event rates were similar between groups; there were three serious adverse events, including two deaths by suicide, one in each arm. Effect sizes on the functioning scale were modest, this was a single centre, and patients and therapists were unblinded, so we should be measured. But for a modality with almost no randomised evidence base in psychosis, this is the strongest signal we have that psychodynamic work tailored to interpersonal vulnerability can sit alongside, rather than instead of, established psychosocial treatments.

Staying with severe mental illness, also in The Lancet Psychiatry, Szmulewicz and colleagues used the BD-CAUSAL Collaboration to emulate a pragmatic target trial in young people presenting with a first manic episode with psychotic features, drawing on first-episode psychosis clinics in North America, Chile, and Spain [2]. Among 371 eligible patients with a median age of 22, they compared mood stabiliser monotherapy, second-generation antipsychotic monotherapy, and combination therapy over two years, with psychiatric hospitalisation or emergency department attendance as the outcome. Roughly two in five patients were hospitalised. Both mood stabiliser monotherapy and combination therapy carried about a seven percentage point lower two-year risk than antipsychotic monotherapy — around an eighteen percent relative reduction — while mood stabiliser alone and combination therapy performed similarly. This is observational, with all the residual confounding that implies, and the authors are explicit that randomised confirmation is needed. Still, the practical message is that after a first psychotic mania, defaulting to an antipsychotic alone may be the weakest of the three options, and adding a mood stabiliser bought nothing extra over the mood stabiliser by itself. Set that against Deng and colleagues in the American Journal of Psychiatry, who used a self-controlled case series in Hong Kong electronic health records covering nearly eighteen thousand people with bipolar disorder, of whom about one in eight had received both long-acting injectable and oral antipsychotics [3]. Within-person, injectable periods were associated with about a fifth fewer all-cause hospitalisations, a third fewer psychiatric hospitalisations, and roughly half the rate of admission for mania. There was no significant difference for depressive episodes, and no excess of non-psychiatric or cardiovascular admissions. Extrapyramidal symptoms were about three times more frequent early on, with the difference no longer significant beyond ninety days. So antipsychotics clearly retain a role in bipolar disorder — particularly the injectable route in patients with adherence problems and mania-predominant relapse — but the first-episode data argue against them as a standalone maintenance strategy.

Our second theme is precision and equity in prescribing. Park and colleagues, again in The Lancet Psychiatry, examined CYP2D6 and CYP2C19 genotypes and venlafaxine treatment failure across two independent naturalistic cohorts: over five thousand patients from a Norwegian therapeutic drug monitoring service and over three and a half thousand from the United Kingdom Biobank [4]. Treatment failure was defined pragmatically as switching to another antidepressant within a year. CYP2D6 poor metabolisers had roughly double the odds of switching in the Norwegian cohort and about a sixty percent increase in the United Kingdom cohort. Intermediate and ultra-rapid metabolisers also showed elevated odds, and patients who were poor metabolisers at both enzymes had around six times the odds of switching, though that dual-poor-metaboliser group was small and the estimate imprecise. Switching is an indirect proxy for non-response or intolerance, but the replication across two very different data sources strengthens the case that pre-emptive genotyping, where available, could spare patients a failed venlafaxine trial. Equity of a different kind is addressed by Smith and colleagues in The Lancet Psychiatry, who convened an international multidisciplinary panel including people with lived experience to build an evidence-based consensus prioritising antipsychotics for schizophrenia in Ethiopia, Nigeria, Rwanda, and South Africa [5]. They produced a prioritised drug list to guide implementation, research, and market shaping, while flagging that almost none of the underlying trial evidence comes from these settings and that trial populations are highly selected. It is a reminder that our evidence base is geographically narrow, and a template for contextualising it elsewhere.

The third theme is developmental psychiatry, and here the headline is diagnostic drift. In JAMA Psychiatry, Tarp and colleagues used Danish national registries covering more than two million children, comparing over seventy thousand children diagnosed with ADHD or autism between 2012 and 2022 against ten matched controls each, across nineteen prediagnostic characteristics [7]. Children with these diagnoses still differed from their peers on every characteristic, but the differences narrowed steadily over the decade, with socioeconomic and perinatal factors attenuating most. The association with low birth weight, for instance, fell from roughly a fifty percent excess risk early in the period to under twenty percent by 2020 to 2022. Attenuation was greater for ADHD than autism and most pronounced in those diagnosed between ages ten and seventeen. The authors are careful, and so should we be: this suggests rising diagnostic rates reflect changing practice and service capacity, not that these conditions have become less impairing for the individual child. That dovetails with a review by Chiang and colleagues in Molecular Psychiatry on the global rise in youth mental health diagnoses, which argues that in high-income countries the increases in depression, self-harm, and suicide mortality are supported by objective severity indicators and so are not purely diagnostic inflation, while identifying sleep disruption as a central transdiagnostic risk factor alongside digital engagement, cyberbullying, and academic pressure [8]. Their proposed three-tiered framework is worth a look, particularly the emphasis on proactively monitoring sleep in high-risk youth. Two further papers refine developmental risk. Kopal and colleagues, also in Molecular Psychiatry, modelled temperament trajectories at eighteen months, three years, and five years in more than fifty thousand children from the Norwegian Mother, Father and Child Cohort, and found that departures from a child's own predicted trajectory mapped onto two transdiagnostic dimensions, with higher scores associated with roughly a fifty percent higher hazard of later ADHD and a comparable increase for Asperger syndrome, sharing identifiable genetic loci whose effects shifted with age [10]. And in the Journal of Child Psychology and Psychiatry, Bernhard and colleagues review sex differences in conduct disorder, integrating the FemNAT-CD study, and conclude that girls and boys with conduct disorder are far more similar than different — neurobiological profiles largely overlap, with differences confined mainly to clinical phenotype such as age of onset, callous-unemotional traits, aggression subtype, and comorbidity [9].

One more paper on mechanism. In the American Journal of Psychiatry, Rötzer and colleagues studied 231 non-treatment-seeking adults with mostly mild to moderate alcohol use disorder, combining functional MRI alcohol cue reactivity with ecological momentary assessment of daily craving [6]. Higher childhood trauma scores predicted greater prefrontal and anterior cingulate activation to alcohol cues in women but not men, and that activation predicted subsequent daily craving only in women, with the mediation running in opposite directions by sex. It is cross-sectional imaging in a modest sample, but it supports taking a trauma history seriously and sex-sensitively in early alcohol use disorder.

If you only have time for one paper this week, make it the target trial emulation of first-episode psychotic mania in The Lancet Psychiatry [2]. It speaks directly to a decision most of us make repeatedly and where guidance has been thin, and it argues for putting the mood stabiliser first.

Here are the key takeaways from this week in Psychiatry. After a first manic episode with psychotic features, mood stabiliser-based maintenance outperformed antipsychotic monotherapy, and adding an antipsychotic to the mood stabiliser added nothing. Long-acting injectables still earn their place in established bipolar disorder for mania-predominant relapse, with close monitoring for extrapyramidal symptoms in the first three months. Manualised psychodynamic psychotherapy improved psychosocial functioning in schizophrenia-spectrum disorders, with gains sustained a year after treatment ended. CYP2D6 poor metaboliser status roughly doubled the likelihood of abandoning venlafaxine, strengthening the case for pre-emptive genotyping. And in child psychiatry, the profile of children receiving ADHD and autism diagnoses is converging on that of their peers — interpret rising rates as a change in practice, not a change in the disorders.

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 a quick rating on Apple Podcasts or Spotify helps other physicians discover the show.

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