This week in Psychiatry

Nov 16, 2026

Latest episode · 10 min
Editor’s pick

Dialogical therapy plus treatment as usual versus treatment as usual in adults with distressing and persistent auditory hallucinations in the UK (Talking with Voices-II): a randomised controlled trial.

In 309 adults with distressing voices, dialogical therapy added to usual care improved personal recovery at eight months with a moderate effect, which persisted but diminished by fourteen months.

The Lancet Psychiatry · 2026 · PubMed

This week’s highlights

  1. 01Comparative efficacy and acceptability of antidepressants in the long-term treatment of major depressive disorder: a systematic review and network meta-analysis.

    Across 115 trials of long-term antidepressant treatment, escitalopram and duloxetine outperformed fluoxetine on sustained response and dropout, though over half of head-to-head studies had high risk of bias.

    van Haaren PCF et al. · The Lancet Psychiatry · 2026

    PMID 43008763

  2. 02Comparative efficacy and tolerability of pharmacological interventions for the acute treatment of obsessive-compulsive disorder across the lifespan: a systematic review and network meta-analysis.

    In 52 trials of obsessive-compulsive disorder drugs, clomipramine showed the largest symptom benefit in adults and young people but was less tolerable than SSRIs, with paediatric evidence of low certainty.

    Cohen S et al. · The Lancet Psychiatry · 2026

    PMID 43014332

  3. 03Dialogical therapy plus treatment as usual versus treatment as usual in adults with distressing and persistent auditory hallucinations in the UK (Talking with Voices-II): a randomised controlled trial.

    In 309 adults with distressing voices, dialogical therapy added to usual care improved personal recovery at eight months with a moderate effect, which persisted but diminished by fourteen months.

    Longden E et al. · The Lancet Psychiatry · 2026

    PMID 43008764

  4. 04Guided self-help for depression adapted for autistic adults: a multi-centre randomised controlled phase 3 trial (ADEPT-2).

    Autism-adapted guided self-help reduced depression scores by about five points versus usual care at sixteen weeks in 266 autistic adults, with the benefit sustained at one year.

    Russell A et al. · Psychotherapy and Psychosomatics · 2026

    PMID 43024490

  5. 05Psychological and psychosocial interventions for the treatment of antenatal maternal depression: systematic review and network meta-analysis.

    Across ten trials in pregnant women with diagnosed depression, cognitive behavioural therapy gave the largest symptom reduction versus usual care, but certainty of evidence was very low throughout.

    Della Rocca B et al. · British Journal of Psychiatry · 2026

    PMID 43000886

  6. 06Changes in pregabalin use and overdose after government interventions and media coverage: retrospective observational study.

    In Australia, pregabalin dispensing fell by about a quarter and overdose calls by about half after media coverage, prescribing restrictions and boxed warnings, in an observational time-series analysis.

    Cairns R et al. · BMJ · 2026

    PMID 43014818

  7. 07Longitudinal patterns of attention-deficit/hyperactivity disorder medication use and subsequent incident substance use among US youth.

    Among United States adolescents with ADHD, neither starting nor stopping ADHD medication was associated with new substance use one year later after propensity weighting in observational survey data.

    Evans-Polce RJ et al. · Journal of Child Psychology and Psychiatry · 2026

    PMID 43025997

  8. 08Comparative effects of various exercise types and doses to improve executive function in children and adolescents with attention deficit hyperactivity disorder: a systematic review with pairwise, network, and dose-response meta-analyses.

    Across 40 trials in young people with ADHD, mind-body exercise ranked best for working memory and combat sports for inhibitory control and flexibility, with a non-linear dose-response relationship.

    Hou X et al. · Molecular Psychiatry · 2026

    PMID 43020884

  9. 09Rethinking University Mental Health for the AI Era: From Crisis Response to Core Mission.

    With 37 to 58 percent of first-year students meeting criteria for a mental disorder, a review proposes a university-governed digital stepped-care model with regulated artificial intelligence, not yet tested.

    Kessler RC, Ono SJ · JAMA Psychiatry · 2026

    PMID 43013291

  10. 10Should distress and impairment be required for psychiatric diagnosis?

    A conceptual analysis argues that distress and impairment should not be universal requirements for individual psychiatric diagnosis, distinguishing classification, criteria, personal significance and warranted response.

    Vinkers CH, Strijbos DW, Meynen G · British Journal of Psychiatry · 2026

    PMID 43000899

Read the weekly briefing

This AudioScholar briefing is generated by artificial intelligence from PubMed abstracts for healthcare professionals and is not medical advice.

This week: a ranking of antidepressants for the long haul, a therapy that talks directly with distressing voices, and pregabalin harms falling in Australia after media coverage and prescribing restrictions.

Welcome to This Week in Psychiatry. We're covering 10 papers this week: 3 in depth, 4 in brief, 3 quick signals.

In The Lancet Psychiatry, van Haaren and colleagues report a systematic review and network meta-analysis comparing 21 antidepressants for long-term treatment of adult unipolar major depression, meaning beyond twelve weeks [1]. It pooled 115 double-blind randomised trials with about thirty thousand participants, and the primary analysis rested on 35 head-to-head continuation studies with close to ten thousand patients. The main outcomes were sustained response, meaning reaching and then keeping a response, and all-cause discontinuation. With fluoxetine as the reference, escitalopram raised the odds of sustained response by about sixty percent, duloxetine by a similar margin, and agomelatine by about forty percent. Bupropion and reboxetine did worse than several other drugs. Escitalopram and duloxetine also had lower all-cause discontinuation than fluoxetine, while reboxetine and milnacipran had more dropout than several comparators. Sensitivity analyses that added enrichment-design trials pointed the same way. The main limitation is the underlying evidence: just over half of the head-to-head studies carried a high risk of bias, and the authors say certainty is limited. The review adds ranked comparative data for maintenance treatment, which the authors offer as input to shared decision making, and they call for more long-term head-to-head trials.

Also in The Lancet Psychiatry, and sharing an author with the antidepressant review, Cohen and colleagues ran a network meta-analysis of drug treatment for obsessive-compulsive disorder across the lifespan [2]. It included 52 double-blind trials with about seven thousand three hundred participants, analysed separately for nine trials in children and adolescents and 43 trials in adults, with outcomes at eight to sixteen weeks. In adults, clomipramine had the largest effect on symptoms, at about 0.7 of a standard deviation over placebo, rated high certainty, with fluvoxamine close behind, and paroxetine, sertraline, escitalopram and fluoxetine all beating placebo. Clomipramine beat fluoxetine directly, at low certainty. The trade-off was tolerability: adults on clomipramine had more than four times the odds of stopping for adverse events compared with placebo, and it was less tolerable than most active comparators. In young people, clomipramine also had the largest effect and outperformed individual SSRIs, and only sertraline was less tolerable than placebo. The key limitation is the paediatric evidence, which is thin and rated low to very low certainty. The review adds a structured efficacy and tolerability comparison, and people with lived experience shaped it throughout.

The third paper, again in The Lancet Psychiatry, is Talking with Voices-II, a rater-blinded randomised trial led by Longden [3]. It enrolled 309 adults with serious mental health problems and persistent distressing voices across four community National Health Service sites in the United Kingdom. Participants received either up to 26 sessions of dialogical therapy over eight months, in which the therapist speaks directly with the voices, plus usual care, or usual care alone. The primary outcome, self-reported personal recovery at eight months, improved by about seven points more with the therapy, a moderate standardised effect of about 0.6. Twelve of fifteen secondary outcomes also favoured the therapy, including all voice-related measures, negative self-beliefs and post-traumatic stress symptoms, though these were unadjusted for multiple testing. Dissociation and positive self-beliefs did not change. At fourteen months the recovery benefit persisted but was roughly cut by half. Serious adverse events were similar between groups, and none were judged definitely related to therapy. The main limitation is the usual-care comparator, so attention and therapist contact are not controlled for. The trial adds the first multisite randomised evidence for this lived-experience informed approach.

Now, four in brief.

In Psychotherapy and Psychosomatics, Russell and colleagues report ADEPT-2, a multicentre phase 3 randomised trial in 266 autistic adults with at least moderate depressive symptoms recruited from the community [4]. Guided self-help adapted for autism, combining psychoeducation and behavioural activation with nine coach meetings, was compared with usual care. At sixteen weeks, depression scores on the Beck inventory were about five points lower with the adapted programme, an effect close to half a standard deviation, and the difference held at one year. About four in five participants allocated to the programme completed the minimum dose. This is a well-sized trial in an underserved group, though the comparator was usual care and the abstract does not report adverse events.

In the British Journal of Psychiatry, Della Rocca and colleagues ran a network meta-analysis of psychological and psychosocial treatments for clinically diagnosed antenatal depression, pooling ten trials with about eighteen hundred pregnant women [5]. Cognitive behavioural therapy gave the largest and most consistent reduction in depressive symptoms, a large standardised effect of about 0.8 compared with usual care, and it outperformed other active interventions. Dropout did not differ meaningfully across treatments. The authors rate certainty as very low across the network, so the ranking is consistent with current guideline recommendations for first-line psychological care but rests on a small evidence base.

In the BMJ, Cairns and colleagues analysed pregabalin dispensing and poisons centre overdose calls in Australia from 2016 to 2024, an observational study covering a sample of about two hundred and thirty thousand people dispensed the drug [6]. Use peaked in 2018 and then fell, with breakpoints after media coverage in late 2018, prescribing restrictions in 2020, and a later point that may reflect boxed warnings. Combined, dispensing fell by about a quarter and overdose calls by close to half against the projected trend. A high-risk group made up about one in seven users, and two thirds of people in that group were also dispensed opioids. The timing links interventions and declines, but this design cannot separate their individual effects.

In the Journal of Child Psychology and Psychiatry, Evans-Polce and colleagues used a nationally representative longitudinal survey of about two thousand one hundred United States adolescents with parent-reported ADHD [7]. With propensity weighting, starting ADHD medication was not associated with any difference in new alcohol, nicotine, e-cigarette, cannabis or other drug use a year later, and stopping medication showed no difference compared with continuing. This adds to evidence that medication patterns do not raise substance use risk, but it is observational, relies on parent report and follows outcomes for only one year.

Three quick signals.

Staying with ADHD, Hou and colleagues in Molecular Psychiatry pooled 40 randomised trials of exercise in about fifteen hundred young people [8]. Mind-body exercise ranked highest for working memory and combat sports for inhibitory control and cognitive flexibility, with a non-linear dose-response. These rankings are probabilistic, and the abstract gives no certainty grading.

In JAMA Psychiatry, Kessler and Ono review student mental health data showing that between 37 and 58 percent of first-year students worldwide meet twelve-month criteria for a DSM-5 disorder [9]. They propose a university-governed stepped-care model integrating digital tools and regulated artificial intelligence. It is a conceptual proposal, not a tested intervention.

In the British Journal of Psychiatry, Vinkers and colleagues argue that distress and impairment should not be universal requirements for individual psychiatric diagnosis, separating why a condition is classified from whether a person meets criteria [10]. This is a conceptual analysis without new data.

Stepping back across this week's papers, the strongest designs were the randomised trials of psychological treatments, yet each compared an active therapy with usual care, which leaves open how much of the benefit comes from added contact. The network meta-analyses offer useful rankings, but they repeatedly rest on trials at high risk of bias or graded low certainty, and the observational work describes associations whose causes the designs cannot fully separate.

If you only have time for one paper this week, make it Talking with Voices-II [3]. It is the week's most rigorous trial, and it reopens the question of whether a psychological approach can move personal recovery for people with persistent voices, a group for whom existing treatments offer modest gains.

That's your roundup for This Week in Psychiatry. The full transcript and references are available on the episode page. 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 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. Use it to choose which papers to read in full.

Get Psychiatry every week

By email, Spotify, Apple or any podcast app.

Listen on Apple PodcastsListen on Spotify

Spot something worth flagging?