This Week in Psychiatry — Sep 21, 2026

Welcome to This Week in Psychiatry. This week we're covering 10 notable papers spanning early identification of serious mental illness, family- and child-focused interventions, and the reward circuitry that underlies depression and eating disorders, plus a set of papers asking harder questions about the words and technologies we are adopting. Let's dive in.

We start with prediction, and with a study that may reframe how we think about where future psychosis comes from. In the Journal of Child Psychology and Psychiatry, Gregersen and colleagues followed nearly two million Danes born from 1990 onward, across more than thirty million person-years of register data, asking a simple question: how many people who go on to develop a psychotic or bipolar disorder had already been seen by child and adolescent mental health services [1]. The answer was fifty-eight percent. Nearly three in five of all such diagnoses in the entire national population occurred in people who had already come through that door as children or adolescents. Attending those services was associated with roughly a fifteenfold increase in risk of any psychotic or bipolar outcome, and the highest-risk group was strikingly concentrated: among young people admitted as psychiatric inpatients in adolescence, the risk was elevated some fortyfold, and half of that inpatient group had received a diagnosis of psychosis or bipolar disorder by age thirty-two. Compare that with conventional clinical high-risk services, which capture only a small fraction of eventual cases, and the implication is uncomfortable but actionable. The high-risk cohort is already sitting in our child services. This is register data, so it cannot tell us that intervention in adolescence changes trajectories, and it cannot separate detection from causation. But it does say that if we want large-scale prediction and prevention, child and adolescent mental health services are where the yield is.

Alongside that, a systematic review and meta-analysis in Molecular Psychiatry from Ang and colleagues looked at the body rather than the brain in first-episode, drug-naive illness [2]. Pooling 105 studies across schizophrenia, major depression and bipolar disorder, they found that peripheral organ dysfunction is present at onset, before any antipsychotic or antidepressant exposure. In schizophrenia this included raised triglycerides, insulin and insulin resistance, higher two-hour glucose, a higher waist-to-hip ratio, raised systolic blood pressure, elevated neutrophil, monocyte and white cell counts, and reduced albumin. In depression, tumour necrosis factor alpha and interleukin-10 were raised, and interleukin-6 and interferon gamma were elevated in both schizophrenia and depression. There were essentially no eligible data on generalized anxiety disorder, and only four studies on bipolar disorder, which is itself a finding. The clinical message is that metabolic risk in first-episode psychosis is not simply iatrogenic — baseline metabolic and inflammatory screening at first presentation is justified on the evidence, not just as a medication safety ritual.

Turning to children and families, two papers in the Journal of Child Psychology and Psychiatry sharpen what we can actually do in clinic. Sim and colleagues report a pragmatic randomized trial of Parenting for Resilience, a non-specialist-delivered intervention co-designed for migrant and displaced families from Myanmar, randomizing 479 caregivers against treatment as usual [3]. This is unusual in that it deliberately targeted caregiver mental health and violence against children together rather than treating parenting skills in isolation, and the effects were substantial and sustained. Physical violence against children was roughly halved at six months post-intervention, psychological violence fell by about a third, and caregiver psychological distress improved at both one and six months. Post-traumatic stress symptoms, emotion regulation, psychological flexibility and family functioning all improved as well. Notably, positive parenting rose only at the earlier timepoint, and there were no detectable effects on parental self-efficacy, social support, or children's own emotional and behavioural difficulties — so the benefit here is demonstrated at the caregiver and family level, not yet at the level of child symptoms. Still, in a low-resource displacement setting delivered by non-specialists, that is a meaningful result.

The second, from Lebowitz and colleagues, does something quietly useful: it gives the Family Accommodation Scale-Anxiety an actual clinical cutoff [4]. Family accommodation — the parental reassurance, the schedule changes, the avoidance we all recognise — is a well-established maintaining factor in childhood anxiety and an explicit treatment target, yet until now clinicians have had no threshold. Across nearly twelve hundred youth aged five to seventeen in clinical and community samples, receiver operating characteristic analyses supported a total score of nine to eleven out of thirty-six as indicating clinically significant accommodation. The authors recommend nine for screening, where sensitivity was about seventy-six percent, and eleven where you want greater specificity. A score of nine sits at the ninetieth percentile of the community distribution. If you use this scale, you now have a number to anchor your formulation and to track.

Our third theme is reward circuitry, and here three papers converge from different directions. In Molecular Psychiatry, Morris and colleagues report a proof-of-principle randomized trial of seven-Tesla functional MRI biofeedback, training sixty-two unmedicated participants to self-modulate the dopaminergic midbrain, targeting motivation regulation [5]. Active biofeedback produced significant clinical improvement in the depressed group compared with sham immediately after training and at twenty-four hours, driven by depressed mood and negative affect rather than positive affect. Crucially, those effects had faded and were no longer significant at seven and thirty days. Active training did engage coupling between the ventral tegmental area and lateral frontal cortex, and better regulation correlated with improvement. So the mechanism looks real and the durability does not — this is a target-engagement study, not a treatment.

Pushing the same circuit pharmacologically, Pizzagalli and colleagues, also in Molecular Psychiatry, present a cross-species programme on the nociceptin orphanin F/Q system [6]. Early-life adversity in mice reduced sucrose preference and increased prepronociceptin expression in the ventral tegmental area, with sex-dependent striatal changes. A nociceptin receptor antagonist enhanced reward learning in rats, but only at the highest dose tested. And in depressed humans, eight weeks of the same antagonist modulated choice consistency on an effort-based decision task relative to placebo. That is an intriguing signal for anhedonia, but it is a behavioural task outcome, not a clinical endpoint — worth watching, not yet worth prescribing.

The third reward paper is the largest coordinated neuroimaging analysis yet in bulimia nervosa, in JAMA Psychiatry. Berner and colleagues pooled seventeen cohorts through the ENIGMA Eating Disorders Working Group, comparing 369 women with bulimia against 417 controls [7]. They found lower nucleus accumbens volume and reduced cortical surface area in superior and transverse temporal cortices, with no differences in cortical thickness, and these held after accounting for body mass index, illness duration, depressive symptoms and medication. Binge-eating frequency, but not compensatory behaviours, tracked with lower surface area across insula, orbitofrontal and cingulate regions. The effect sizes are small — around a fifth of a standard deviation — so this is not a diagnostic biomarker. What it is, is the first reliable structural map implicating reward, interoceptive and cognitive control circuits in a disorder whose neurobiology has been poorly characterised.

Finally, three papers ask us to be more careful about language, technology and hype. In JAMA Psychiatry, Bellato and colleagues conducted a scoping review of 134 studies covering over 150,000 participants who were described as neurodivergent or neurodiverse [8]. Most participants were identified by self-reported diagnosis or simply by self-identification rather than by clinical assessment, the underlying conditions ranged from autism and ADHD through dyslexia, dyspraxia, anxiety and dyscalculia, and about a fifth of the studies specified no clinical diagnosis at all. The authors argue for precise diagnostic descriptors in clinical research, reserving neurodivergent for contexts genuinely about identity and lived experience. In Psychotherapy and Psychosomatics, Qiu and colleagues offer a roadmap for artificial intelligence in mental health, arguing that technical performance is not clinical value, and that these tools should augment rather than replace assessment, with validation, equity and governance as preconditions [9]. And in Molecular Psychiatry, Browning and colleagues review cold exposure for mental health and conclude it remains promising but unproven: small samples, heterogeneous protocols, no active placebo controls, self-selected health-motivated participants, and psychological improvements that have not been consistently tied to biological change [10]. A reasonable answer for the patient who asks about cold plunges is that it is not harmful for most people and not established as treatment.

If you only have time for one paper this week, make it the Danish register study of psychosis and bipolar diagnoses following child and adolescent mental health service contact [1]. It reframes where early detection effort should be concentrated, and it applies to every clinician who sees young people or who receives their referrals a decade later.

Here are the key takeaways from this week in Psychiatry. Nearly three in five future psychotic and bipolar diagnoses occur in people who attended child and adolescent services, and adolescent inpatients are an extraordinarily high-risk group deserving structured follow-up. Metabolic and inflammatory abnormalities are present at first episode before any medication, so screen at presentation rather than after treatment starts. Family accommodation in child anxiety now has an empirical threshold — nine for screening, eleven for specificity. An integrated parenting and caregiver mental health package delivered by non-specialists cut physical violence against children roughly in half in a displacement setting, though child symptom outcomes did not shift. And in depression research, midbrain neurofeedback and nociceptin antagonism both engage reward circuitry, but neither has yet shown durable clinical benefit.

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.