Eating disorders, particularly anorexia nervosa and bulimia nervosa, are significant global health challenges.

This study analyses historical trends and forecasts future patterns of eating disorders among young adults aged 15–29 years using machine learning techniques.

Global data on anorexia nervosa and bulimia nervosa from the Global Burden of Disease study 2021 spanning 1990 to 2021 were analysed, examining incidence, prevalence and disability-adjusted life years (DALYs) across age groups, sociodemographic index (SDI) levels and regions. Eight machine-learning models were employed to forecast trends from 2022 to 2050.

Bulimia nervosa showed more pronounced increases compared to anorexia nervosa across all metrics. The 15–19 age group had the highest incidence rates, while the 20–24 age group showed the highest prevalence and DALY rates. Low SDI regions experienced substantial increases, with bulimia nervosa prevalence rising by 179.05%. East Asia demonstrated the most significant rise in age-standardised rates. The Prophet model best forecast anorexia nervosa trends, while ARIMA performed best for bulimia nervosa. Projections indicate continued increases through 2050 for both disorders.

The global burden of eating disorders among young adults is projected to increase significantly by 2050, with bulimia nervosa showing more rapid growth than anorexia nervosa. Substantial variations exist across age groups, SDI levels and regions. These findings highlight the urgent need for enhanced prevention programmes targeting high-risk age groups, strengthened healthcare capacity in rapidly developing regions and evidence-based policy interventions to address the growing global burden of eating disorders.

Benzodiazepine receptor agonists (BZRAs), including benzodiazepines and Z-drugs, are frequently prescribed during pregnancy but their long-term neurodevelopmental safety remains uncertain.

To investigate whether prenatal BZRA exposure is associated with an increased long-term risk of neurodevelopmental disorders (LNDDs) in offspring.

This nationwide, population-based cohort study used Korean National Health Insurance Service data on all live births from 2011 to 2014, followed until 2023. Prenatal BZRA exposure was defined as maternal prescriptions during pregnancy. Propensity score matching (1:10) was applied to balance covariates. Sensitivity analyses in the full cohort evaluated exposure intensity (0, 1–6, 7–29 and ≥30 cumulative days), drug class (benzodiazepines versus Z-drugs), trimester of exposure and discordant sibling comparisons with mother fixed effects.

Among 1 553 505 eligible births, 5949 BZRA-exposed and 55 015 matched unexposed children were analysed. LNDD incidence was 13.9% in the exposed group versus 11.4% in the unexposed (odds ratio 1.25, 95% CI: 1.16, 1.35). In the full cohort, risks increased with exposure intensity: 1–6 days (odds ratio 1.16, 95% CI: 1.05–1.28), 7–29 days (odds ratio 1.19, 95% CI: 1.04–1.36) and ≥30 days (odds ratio 1.18, 95% CI: 1.01–1.38). By trimester, risks were higher with second- (odds ratio 1.30, 95% CI: 1.07–1.59) and third-trimester (odds ratio 1.27, 95% CI: 1.09–1.48) exposure. Class-specific analyses showed stronger associations for benzodiazepines only (odds ratio 1.19, 95% CI: 1.15–1.23) than for Z-drugs only (odds ratio 1.06, 95% CI: 1.04–1.08). In a discordant sibling analysis including 2572 children this association persisted (odds ratio 1.29, 95% CI: 1.05–1.60), indicating that neither familial nor genetic confounding fully explains the observed effects.

Prenatal BZRA exposure was associated with increased long-term risks of LNDDs in offspring, with evidence of dose–response and class-specific effects, and persistence in sibling analyses.

Cue-reactivity responses in addictive behaviours are triggered by cues associated with the addictive activity itself. Although such cues may depict the rewarding aspects of the behaviour, responses may also generalise to more distal cues that do not directly convey this content.

To examine cue reactivity to distal cues (i.e. devices displaying starting or log-in screens of internet applications) in a diagnostically validated sample of individuals with specific problematic usage of the internet (PUIs) and determine whether laboratory-measured cue reactivity predicts real-life behavioural engagement and temptation experiences, in addition to differences across PUI stages and cue types.

In this preregistered study, data were collected from October 2021 to 31 August 2024 from individuals with non-problematic (n = 268), risky (n = 135) and pathological (n = 133) engagement in specific internet activities (gaming, buying and/or shopping, pornography use and social networking). Participants were aged 18–65 years (mean age 26.12 years, s.d. 6.79), and 44.6% were female. A cue-reactivity paradigm with distal cues showing target and non-target internet activities was used. A within–between participants design was used, with repeated measures analyses of variance. Correlations between laboratory cue-reactivity measures and measures from a 14-day end-of-day assessment in the natural environment are reported.

Heightened cue reactivity (arousal, urge and/or craving) was observed in individuals with risky and pathological use compared with those with non-problematic use across all levels of the paradigm. Individuals with pathological use showed elevated levels of urge and craving, along with generalised responses to stimuli showing starting and/or log-in screens not related to their specific (addictive) behaviour. These effects were consistent across different types of PUI and were associated with engagement in the behaviour and temptation experiences in naturalistic settings.

These findings indicate that cue reactivity and craving are central aspects of PUIs. Although different devices may elicit different types of action, our results highlight the challenges of regulating behaviour in environments saturated with unavoidable triggers, such as internet content and devices.

Suicides in children and young people are a major public health concern. Prevention of Future Death (PFD) reports are an underutilised resource detailing coroners’ concerns which, if actioned, are believed to be able to prevent future deaths. Research has investigated common themes for suicide during 2021 and 2022 but there are no published studies that thematically analyse these reports for children alone.

To identify key themes raised by coroners from PFD reports published between 2015 and 2023 for children who have died by suicide.

PFD reports for suicides in children were downloaded from the Courts and Tribunals Judiciary website. Descriptive statistics were collated from reports. Reports (n = 37) were analysed using inductive content analysis to determine primary and sub-themes using QSR NVIVO 14 Qualitative Analysis software.

Reports came from 30 coroners’ areas, with most reports being sent to government departments and NHS Trusts/Clinical Commissioning Groups. The qualitative analysis resulted in six primary themes being identified: service provision, staffing and resourcing, communication, multiple services involved in care, accessing services and access to harmful content and environment. Furthermore, 23 sub-themes were identified such as standard operating procedures/processes not being followed or being inadequate, a lack of specialist services and a disconnect between integrated services. A quarter of reports were on children diagnosed with autism, and there were specific issues highlighted in concerns relating to services and staffing for children with neurodiverse conditions.

The key findings from this report highlight themes raised by coroners relating to deaths of children by suicide. This included themes around service provision, staffing and resourcing of mental health services and communication between services and families. Children with neurodiversity, including autism, appear to be of particular concern.

Ethnic inequalities in compulsory psychiatric hospital detentions are well-documented in the UK and internationally. It is unknown how UK coronavirus disease 2019 (COVID-19) lockdown measures, which led to restrictions in public movement, gatherings, in-person health service delivery and changes to police powers, further impacted inequalities.

In this study, we assessed whether national lockdown measures impacted ethnic inequalities in voluntary and compulsory psychiatric hospital admissions during the COVID-19 pandemic.

Daily counts of admissions and detentions to psychiatric hospitals were extracted from a large population-level sample of secondary mental health service users in South East London. Changes during two COVID-19 lockdown periods, over 2020–2021, were compared with pre-pandemic periods (2016–2019) with the use of a regression discontinuity in time design to assess ethnic inequalities in voluntary and compulsory mental health admissions.

Compared to the pre-pandemic reference (2016–2019), after adjusting for seasonal and weekly trends, overall admissions to mental health units dropped during the first COVID-19 lockdown (incidence rate ratio (IRR) 0.87 (95% CI: 0.75–1.00)), but compulsory detentions rose (IRR 1.25 (1.05–1.54)). This was mostly due to higher compulsory detentions in the Black Caribbean group (IRR 1.54 (1.08–2.19)). During the second COVID-19 lockdown, whereas total daily admissions remained similar to the pre-pandemic reference (IRR 1.03 (0.92–1.15)), total new daily detentions was elevated (IRR 1.28 (1.11–1.49)), specifically in Black Caribbean (IRR 1.53 (1.14–2.06)) and Black African (IRR 1.57 (1.06–2.34)) groups.

COVID-19 lockdown measures exacerbated pre-existing ethnic inequalities in compulsory psychiatric detention, particularly for those from Black Caribbean and Black African backgrounds. There is a need to address ethnic inequalities in compulsory psychiatric detentions and attend to exacerbations of pre-existing inequalities during health emergencies like the COVID-19 pandemic. This cannot be achieved without addressing systemic racism within criminal justice and healthcare systems and tackling inequalities in wider social and economic determinants of mental health.

Clozapine is licensed for treatment-resistant schizophrenia (TRS). Because of the risk of clozapine-induced agranulocytosis, its use requires regular haematological monitoring. Substantive evidence supports revisions of absolute neutrophil counts (ANCs) for clozapine discontinuation and ceasing of indefinite haematological monitoring.

To examine the cost-effectiveness and budget impact of different haematological monitoring schemes compared with the current UK monitoring practice for patients using clozapine.

We performed a cost-effectiveness and budget impact analysis from the healthcare system perspective over a 3-year period, comparing the current UK clozapine monitoring practice with extended haematological monitoring and a revision of ANC criteria. Costs and quality-adjusted life years (QALYs) were estimated using a semi-Markov model that followed a simulated cohort of 100 000 adults with TRS. Sensitivity analyses were conducted.

Extended haematological monitoring would lead to lower mean total costs per patient (6388.34 v. 5569.77 GBP) and not compromise quality of life (in QALYs 795.83 v. 795.79 days). A revision of ANC criteria for clozapine discontinuation would not substantially lower costs (6388.34 v. 6390 GBP), but lead to a slight increase in QALYs (795.83 v. 797.08 days), through patients benefitting from longer clozapine treatment. A combination of extended haematological monitoring and revision of ANC criteria would be the dominant strategy, which means that costs are lower (6388.34 v. 5548.50 GBP) and QALYs slightly increase (795.83 v. 797.03 days) compared with the current UK monitoring practice.

A revision of current UK clozapine monitoring practice would be beneficial from both a clinical and an economic perspective. Adjusting ANC criteria for clozapine cessation avoids unnecessary early discontinuation of clozapine treatment and has a positive impact on quality of life. An extension of monitoring intervals reduces costs borne by the healthcare system. Safety is not compromised by these changes.

The smallest worthwhile difference (SWD) represents the smallest beneficial effect of an intervention that patients deem worthwhile given the harms, expenses and inconveniences of the intervention. The SWD facilitates interpretation of the patient-perceived importance of intervention effects. We previously estimated the SWD for antidepressants for depression, but the SWD for psychotherapy remains unknown.

To estimate the SWD of recommended psychotherapies for depression compared with no treatment.

We estimated the SWD through a patient required difference in response rates between psychotherapy and no treatment after 2 months. We recruited using Prolific, an online cross-sectional survey platform, in the UK and USA in January 2025. We also queried a random subset of respondents to replicate our previous SWD estimation for antidepressants.

In the primary study, we recruited 526 participants (mean age: 36.7 years (s.d. = 12.5); 54% women and 61% White individuals). Of these, 6% reported that they would not initiate psychotherapy with a 100% treatment response. For those willing to initiate psychotherapy, 87 reported moderate-to-severe depressive symptoms but were not in treatment, 184 were in treatment and 131 reported absent-to-mild symptoms with or without previous treatment. The median SWD for people with moderate-to-severe depressive symptoms, not in treatment and willing to consider psychotherapies was a 20% (interquartile range: 10–35%) difference in response rates comparing psychotherapy with no treatment. This was similar to the SWD for antidepressant drugs (SWD = 20%, interquartile range: 10–30%; n = 104). Participant characteristics were not meaningfully associated with the SWD.

Current empirically supported psychotherapy response rates of 15% were sufficient for one in three people to initiate psychotherapy given the burdens, but two in three expected greater treatment benefits or fewer burdens. The SWD for psychotherapy was not materially different from the estimated SWD for antidepressants. Individual patient value judgements and preferences merit greater attention. These findings should be replicated with diverse samples from different geographical locations.

Research suggests there are alterations in the cortisol awakening response (CAR) in patients with premenstrual dysphoric disorder (PMDD), as demonstrated by delayed cortisol peaks and flatter diurnal cortisol slopes compared to healthy controls. While inconsistent, previous work also demonstrates a relation between alterations in CAR, prefrontal serotonin transporter (5-HTT) binding and severity of depressive symptoms.

This longitudinal study explores CAR in relation to midbrain and prefrontal 5-HTT binding and depressive symptoms in patients with PMDD and in healthy controls across the menstrual cycle.

Thirty patients with PMDD and 29 controls each provided 3 saliva samples for assessment of CAR (awakening, +30 min, +60 min) and 5 to assess the diurnal cortisol slope (09.00, 12.00, 15.00, 18.00, 21.00 h) during the periovulatory and premenstrual phases. [11C]DASB positron emission tomography scans were performed to measure 5-HTT non-displaceable binding potential (BPND). Depressive symptoms were assessed using the Hamilton Depression Rating Scale. Associations between cortisol measures, 5-HTT BPND and depressive symptoms were examined using linear mixed-effects models, independent t-tests, mixed analysis of variance (ANOVA) and Spearman rank correlations.

A significant interaction effect between group and cycle phase was found for cortisol peak concentrations (estimate = 0.78, p = 0.05, d = 0.62, 95% CI: [0.01, 1.56]; and corrected for awakening cortisol concentration: estimate = 0.90, p = 0.02, d = 0.77, 95% CI: [0.15, 1.66]). Cortisol peak concentrations correlated negatively with both midbrain 5-HTT BPND (r = −0.34, p < 0.01, R2 = 0.12) and depressive symptoms (r = −0.30, p = 0.02, R2 = 0.09) during the premenstrual phase.

Patients with PMDD showed attenuated cortisol peaks in the periovulatory phase compared with healthy controls, who demonstrated plastic changes across the cycle. Results point towards an interplay between the stress and the serotonergic system, as well as to the severity of depressive symptoms during the premenstrual phase.