This Week in Psychiatry — Sep 7, 2026

Welcome to This Week in Psychiatry. This week we're covering 10 notable papers spanning treatment-resistant mood disorders and neuromodulation, addiction and behavioural neuroscience, and the developmental and social determinants that shape who gets ill and who gets better. Let's dive in.

We start with treatment-resistant depression, where the most clinically consequential paper of the week comes from JAMA Psychiatry. Orsini and colleagues report the Ket-BD trial, a double-blind, midazolam-controlled randomized trial across three sites in Ontario, testing adjunctive intravenous ketamine in treatment-resistant bipolar depression — a population deliberately excluded from most of the ketamine literature [1]. Sixty-eight adult outpatients with bipolar one or two disorder, in a moderate to severe depressive episode and having failed at least two adequate pharmacotherapy trials, received four flexibly dosed forty-minute infusions over two weeks, on top of a stable mood stabiliser or antipsychotic. At day fourteen, the ketamine group was about seven points lower on the Montgomery-Åsberg scale than midazolam, a moderate-to-large effect. Just as important for the practising clinician, there were no cases of mania, hypomania, psychosis, or suicide attempts in either arm; one participant in each group developed subthreshold mixed features. Two caveats are worth carrying with you: the sample is small, and just under half of participants correctly guessed their allocation after the first infusion, so functional unblinding remains a real concern in psychedelic-adjacent trials. Still, this is the strongest randomized evidence to date that ketamine can be used in bipolar depression without provoking the switch clinicians most fear.

Molecular Psychiatry contributes a mechanistic companion piece. Momi and colleagues report neurophysiological signatures of Stanford Neuromodulation Therapy — the high-dose accelerated intermittent theta-burst protocol — in twenty-four TMS-naive patients with treatment-resistant depression, randomized to active or sham, with fourteen separate TMS-EEG sessions [3]. Active stimulation progressively reduced cortical excitability specifically at the left dorsolateral prefrontal target, with a significant drop by day three and roughly a quarter reduction in the early evoked component, while the vertex control site showed nothing. Estimated medial prefrontal source activity, consistent with the subgenual cingulate, also fell selectively and the effect persisted at one month. An exploratory correlation suggested that patients with higher baseline subgenual activity improved more — the authors themselves flag this as preliminary and in need of replication before anyone treats it as a biomarker for patient selection.

Against those two carefully controlled trials, Molecular Psychiatry also publishes a cautionary tale about the softer end of the evidence base. Fornaro and colleagues conducted a network meta-analysis of nutraceuticals and phytoceuticals in major depressive disorder, spanning 163 randomized trials and nearly sixteen thousand participants [2]. On the face of it, several compounds — eicosapentaenoic acid, vitamin D3, St John's wort extract — beat placebo with effect sizes the authors themselves describe as unrealistically large. The telling analysis is the meta-regression: larger effect sizes tracked with lower methodological quality scores. Heterogeneity was substantial, and overall confidence in the evidence was rated low to very low. The practical message is that when a patient asks about omega-3s or St John's wort, the honest answer is that the signal is real enough to warrant rigorous trials but too fragile to build a treatment plan on.

Turning to addiction, the American Journal of Psychiatry offers a broad review from Mariani and Levin on new horizons in substance use disorder treatment, and it frames the problem starkly: we have approved medications for alcohol, nicotine, and opioid use disorder, and none at all for cannabis, cocaine, or methamphetamine use disorder, while behavioural treatments remain limited by both effectiveness and access [4]. The authors argue that the next generation of treatments will come from new drug classes, new technologies, and new delivery models rather than a single breakthrough. Molecular Psychiatry supplies an example of exactly that. Guo and colleagues built a closed-loop transcranial alternating current stimulation system for methamphetamine use disorder, using cue-evoked parietal-occipital theta power as a real-time biomarker to trigger stimulation of the left dorsolateral prefrontal cortex [8]. Across three cohorts, forty hertz closed-loop stimulation was the only condition that both reduced cue-related electroencephalographic activity and improved inhibitory control. Random-timed stimulation reduced cue reactivity but did nothing for inhibition, and continuous open-loop stimulation did neither — so both frequency and timing mattered. A five-day open-label course in severe methamphetamine use disorder reduced craving and improved inhibitory control, but note that this final step was single-arm and uncontrolled, so it establishes feasibility rather than efficacy.

The third theme concerns who becomes depressed, who responds to psychotherapy, and how our services account for the patients in their care. In Psychotherapy and Psychosomatics, Völker and colleagues analysed baseline data from more than 159,000 adults in the German National Cohort, and found that childhood maltreatment was associated with roughly two and a half times the odds of lifetime depression and close to three times the odds of current depression [5]. What makes this worth your time is the attributable-fraction analysis: maltreatment accounted for around a fifth to a quarter of lifetime depression cases and about a third of current depression, and it explained roughly a fifth of the entire sex gap in depression. The subtype patterns diverged by sex — physical abuse and neglect carried stronger associations in women, sexual abuse a stronger association in men — and emotional subtypes contributed the largest population attributable fractions. This is cross-sectional and relies on retrospective self-report, so causal claims should be cautious, but it argues for routine, sex-aware maltreatment histories.

The same journal carries a small preliminary study on the timing of psychotherapy. Zorzini and colleagues studied forty-five women with spider phobia undergoing group exposure therapy, grouped by follicular phase, luteal phase, or hormonal contraceptive use [7]. Everyone improved on phobic fear and avoidance, and there was no difference between menstrual cycle groups — a negative primary comparison. However, higher progesterone measured on the day of exposure was associated with less improvement in behavioural avoidance. With this sample size, that is hypothesis-generating only, but it raises the question of whether the day we schedule exposure matters. Alongside these, the Journal of Child Psychology and Psychiatry reports work from Cassart and colleagues on 288 adolescent boys with conduct disorder, splitting them by the DSM-5 limited prosocial emotions specifier plus self-reported anxiety [9]. The high-anxiety, so-called secondary variant scored higher on psychopathic traits, affective empathy, depressive symptoms, negative urgency, emotional maltreatment, and sexual abuse — yet the groups did not differ in antisocial behaviour, aggression, or treatment engagement. In other words, a simple anxiety questionnaire flags a subgroup with far more complex needs who look identical on behaviour alone. Molecular Psychiatry adds a genomic angle from Lu and colleagues, a pleiotropic analysis of eleven sleep traits and suicide attempts that identified substantial polygenic overlap, prioritised thirty-two shared genes enriched in synaptic, immune, and metabolic pathways, and found that genetically predicted morning chronotype was associated with lower odds of suicide attempt [10]. And in the British Journal of Psychiatry, Barbui, Sunkel and Thornicroft argue that accountability in acute inpatient mental health care — answerability, responsibility and responsiveness — rests on evidence that is strongest for reducing coercion and for robust measurement systems, and that structural change without cultural change will not hold [6].

If you only have time for one paper this week, make it the Ket-BD trial in JAMA Psychiatry [1]. It addresses a question clinicians face weekly and have had to answer from extrapolation — whether ketamine is safe and effective in bipolar depression — and it does so with a randomized, active-comparator design.

Here are the key takeaways from this week in Psychiatry. First, adjunctive intravenous ketamine produced a meaningful antidepressant effect in treatment-resistant bipolar depression over two weeks, with no observed manic or psychotic switches, though the sample was small and blinding imperfect. Second, accelerated theta-burst stimulation produces site-specific, progressive cortical changes that persist a month out, but baseline subgenual activity is not yet a usable predictor of who will respond. Third, be sceptical about nutraceutical effect sizes — the largest benefits in the literature come from the weakest studies. Fourth, closed-loop, biomarker-guided brain stimulation for methamphetamine use disorder is feasible and frequency- and timing-dependent, but the clinical data remain uncontrolled. And fifth, childhood maltreatment explains a substantial share of depression burden and a meaningful part of the sex gap, with different subtypes mattering differently in men and women — which makes a careful trauma history a population-level intervention, not just an individual one.

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.