As generative AI chatbots become embedded in everyday life, their inherent risks are manifesting in unprecedented ways. Emerging reports describe cases of AI Psychosis in which intensive chatbot use triggers delusional thinking and altered behaviour. This paper proposes a mechanism wherein baseline user vulnerabilities and specific engagement patterns interact with generative AI attributes, such as sycophancy and hallucination, to induce cognitive distortions. Subsequently, it outlines clinical, design, and regulatory strategies that can help mitigate risks.

Background Healthcare workers (HCWs) face ongoing mental health challenges, which were aggravated during the COVID-19 pandemic due to increased workloads, exposure to trauma and heightened infection risks. Although self-reported data have indicated elevated mental health symptoms among HCWs, there is limited evidence based on official workers' compensation claims to assess the true extent of work-related mental disorders in this group.Aims To examine changes in workers' compensation claims for mental disorders among HCWs and other workers before and throughout the first 2 years of the pandemic.Methods Mental disorder compensation claims data from January 2019 to December 2021 among HCWs and other workers were obtained from the Workers' Compensation Board of British Columbia. Monthly claim rates were calculated using denominator data derived from Statistics Canada's Labour Force Survey. Changes in claim rates among HCWs were assessed through interrupted time series analysis using other workers as control and the onset of the pandemic as the event.Results The analysis found no significant shifts in the incidence of mental disorder claims among HCWs during the pandemic. In contrast, a marginal increase in claim rates during the same period was observed in other workers.Conclusions Unlike prior research based on self-reported data, this study found no evidence of an increase in mental disorder claims among HCWs during the pandemic. These results suggest that workers' compensation claims data may not fully capture the broader mental health challenges experienced by HCWs, potentially due to underreporting or barriers to accessing claims.

BACKGROUND:Generative AI (GenAI) mental health chatbots are increasingly being developed to help address persistent barriers to mental healthcare. Unlike earlier rule-based and retrieval-based systems, GenAI chatbots generate open-ended outputs that can be inaccurate and unsafe. Documented harms from general-purpose GenAI chatbots have highlighted the need for purpose-built interventions with dedicated safeguards, yet how safety is implemented in such interventions remains poorly understood. METHODS:This scoping review followed the Joanna Briggs Institute methodology and PRISMA-ScR guidelines, with a prospectively registered and peer-reviewed protocol. A systematic search of seven academic databases and search engines including MEDLINE, Scopus, PsycINFO, ACM Digital Library, IEEE Xplore, Google Scholar and Consensus was conducted in July 2025. Two reviewers independently screened records and extracted data. Safety mechanisms and risk mitigation strategies were narratively synthesised across three pre-specified domains: technical safeguards, pre-deployment safety considerations, and delivery-phase risk mitigation strategies. RESULTS:Twenty-one studies across 11 countries were included. Most interventions incorporated at least one technical safety mechanism, most commonly fine-tuning and prompt engineering. A smaller subset implemented layered safety architectures combining retrieval systems, content filters or risk classifiers, and rule-based algorithms. Pre-deployment safeguards included clinical expert and user co-design approaches, research ethics procedures, and data privacy measures. During intervention delivery, detailed onboarding with role clarification was common, but human oversight was limited. Crisis referral protocols varied in rigour but were mostly underdeveloped, and systematic adverse event monitoring was sparse. Documented safety failures included missed suicidal ideation and provision of inaccurate clinical information. CONCLUSIONS:GenAI chatbot interventions require a robust sociotechnical approach that integrates technical safeguards with user co-design, procedural controls, and human oversight. Future research is needed to evaluate efficacy, improve safeguards and standardise safety outcome measurement. Regulatory oversight proportional to the risks these systems carry is required to enable integration into stepped or blended mental healthcare.

Generative AI (GenAI) mental health chatbots offer scalable, on-demand support with the potential to address persistent gaps in mental healthcare access. Yet evidence on how these interventions are designed and how users experience them remains fragmented. To our knowledge, this scoping review represents the first systematic and integrated mapping of conversational agent features, intervention design characteristics and user experience (UX) outcomes in purpose-built GenAI mental health chatbot interventions. A systematic search of seven databases identified 1899 articles, from which 21 studies across 11 countries were included. Most interventions were early-stage, cognitive behavioural therapy-based, and delivered through non-embodied text chatbots. Target conditions included depression, anxiety, dementia, eating disorders, and post-traumatic stress disorder. UX outcomes indicated moderate-to-high usability, therapeutic alliance, and user satisfaction, driven by convenience, personalisation, and perceived empathy. However, engagement commonly declined over time, attributed to limited interactivity, and erosion of trust following inaccurate or contextually misaligned outputs. Design approaches including multi-modal interaction, domain knowledge grounding, structured delivery formats, and co-design processes emerged as important influences on UX outcomes. Future work should prioritise efficacy trials that incorporate standardised UX outcome measures and researchers should adopt co-design approaches with diverse end-users to enable equitable, human-centred interventions.

INTRODUCTION:Mental health problems constitute a significant global health challenge due to their rising prevalence and substantial treatment gap. Digital Mental Health Interventions (DMHIs) including mental health chatbots have emerged as promising solutions due to their effectiveness and scalability. Recent advances in Generative Artificial Intelligence (GenAI) have improved the conversational abilities of these chatbots, further amplifying their potential. However, despite instances of inadvertent harm stemming from the unpredictable nature of GenAI, little attention has been paid to user experience and safety of these chatbots. OBJECTIVE:This proposed review will explore existing research on GenAI-based mental health chatbots. Specifically, it aims to identify and describe current chatbots, focusing on user experience, safety and risk mitigation strategies. METHODS:The review will follow the Joanna Briggs Institute (JBI) guidelines for conducting scoping reviews. It will also adhere to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Review (PRISMA-ScR). A systematic database search of Medline (PubMed), Scopus, PsycINFO, ACM Digital Library, and IEEE Xplore will be conducted. The database search will be complimented by research-based search engines (Google Scholar and Consensus). Studies focusing on the development, evaluation or implementation of GenAI-based mental health chatbots will be included without limitations to specific disorders or population groups. Two independent reviewers will perform screening and data extraction. The analysis will include descriptive summary and thematic analysis, with results presented in tabular, graphical, and narrative formats. CONCLUSION:This review will provide a comprehensive overview of GenAI-based mental health chatbots while identifying innovative practices and knowledge gaps relating to user experience and safety. Findings will inform the ethical development, evaluation and implementation of GenAI-based mental health interventions.

Generative Artificial Intelligence (GenAI) mental health chatbots promise scalable, on-demand support, yet GenAI-related harms have raised concerns about user safety. To our knowledge, this scoping review represents the first systematic mapping of chatbot architecture, delivery modalities, user experience (UX) outcomes, and risk mitigation strategies in purpose-built chatbot interventions. A systematic search of seven databases identified 1899 articles, from which 21 studies across 11 countries were included. Most interventions were early-stage, cognitive behavioural therapy-based, and delivered through non-embodied chatbots. Target conditions included depression, anxiety, dementia, eating disorder and post-traumatic stress disorder. UX outcomes indicated moderate-to-high usability, therapeutic alliance, and user satisfaction, driven by convenience, personalization, and perceived empathy. However, engagement commonly declined, attributed to trust concerns and limited interactivity. Risk mitigation strategies incorporated technical controls, most frequently fine-tuning and prompt engineering, followed by risk classifiers and content filters, retrieval-augmented generation, and rules-based hybrid integrations. These safeguards were often implemented alongside human oversight, data security measures, and crisis referral pathways. Adverse event monitoring was rare, but some studies documented missed crisis cues and inaccurate content. Future research is needed to evaluate efficacy, standardize safety and UX outcome measurement, and develop layered safeguards to enable integration into stepped or blended care.

Mental healthcare utilization in the US remains low, with persistent disparities observed across population groups. However, little is known about how sex and race/ethnicity jointly shape access to care. Intersectionality theory highlights the need to examine these dimensions together, as their combined influence may produce unique disadvantages not captured in single-axis analyses. This study utilized data from the 2009-2018 cycles of the National Health and Nutrition Examination Survey (NHANES). The relative differences (prevalence ratios) of mental healthcare utilization across intersecting sex and racial/ethnic groups were estimated using design-based log-binomial models. The absolute measure (prevalence differences) across these intersectional groups were obtained using linear probability regression models. Stratified analyses were conducted to examine how socioeconomic and need-related factors modified disparities. Overall, 9.1% of adults reported accessing mental health services in the preceding year. Marked disparities were observed across the intersectional groups. Hispanic males had the lowest utilization rates compared to Non-Hispanic (NH) White males, with an adjusted prevalence ratio (aPR) of 0.59 [95% CI: 0.47-0.73]. Among females, all minority racial/ethnic groups reported lower utilization compared with NH White females with aPRs ranging from 0.73 to 0.81. Within racial/ethnic groups, women generally accessed care more than men, though the magnitude of sex differences varied. Stratified analyses showed that disparities were magnified among those without insurance and attenuated at higher income levels. These results show that sex and race/ethnicity jointly shape patterns of mental healthcare utilization in the United States, producing compounded disadvantages for specific groups such as Hispanic men. Stratified analyses suggest that socio-economic status may modify these disparities, pointing to the role played by systemic inequities. These findings underscore the importance of intersectional approaches in population mental health research and policy. Future research should consider additional intersecting identities including sexual orientation and disability.

Abstract Background Estimating the indirect mortality due to COVID-19 is of the utmost importance to develop adequate public health policy during future outbreaks. Methods From province-wide administrative datasets, we identified British Columbians who tested negative for COVID-19 during the first wave and never tested positive throughout 2020. We obtained a pre-pandemic (2018) cohort matched on age, sex, history of non-communicable disorders (NCDs), multimorbidity, and severity/acuity, and implemented a doubly robust estimation of the effect of the first pandemic wave on mortality. Results The adjusted odds ratio (AOR) of death was 3.2 times higher for a 2020 cohort who tested negative for COVID-19 (n = 123,133), compared to matched pre-pandemic controls. In both cohorts, a majority (72.5%) experienced at least one pre-existing NCD. Stratification by NCD shows an AOR of death ranges between 2–for people with substance use disorders– and 7–for people previously undiagnosed with NCDs (e.g., incident cases that went untreated). The largest subgroup was composed of people with mental disorders (47,413 people), with an AOR of death of 2.5. Though the COVID-19 direct mortality in the general population remained low (1.9 per 10,000), the excess mortality in this COVID-negative cohort was extremely high − 4,085 of the 123,133– which entails a minimum indirect excess mortality death rate of 6.5 per 10,000 in the general population. Conclusions During the first pandemic year, mortality in COVID-negative adults was several times greater than before COVID-19, in people with matched NCD distribution and severity. Our findings suggest that low direct COVID-19 mortality was accompanied by less visible–but much higher– indirect mortality due to undiagnosed and/or untreated NCDs, highlighting the need to focus not only on mitigating the harms of new agents, but also of continuing service delivery for treatable conditions.

BACKGROUND:Non-suicidal self-injury (NSSI) is associated with mental disorders, yet work regarding the direction of this association is inconsistent. We examined the prevalence, comorbidity, time-order associations with mental disorders, and sex differences in sporadic and repetitive NSSI among emerging adults. METHODS:We used survey data from n = 72,288 first-year college students as part of the World Mental Health-International College Student Survey Initiative (WMH-ICS) to explore time-order associations between onset of NSSI and mental disorders, based on retrospective age-of-onset reports using discrete-time survival models. We distinguished between sporadic (1-5 lifetime episodes) and repetitive (≥6 lifetime episodes) NSSI in relation to DSM-5 mood, anxiety, and externalizing disorders. RESULTS:We estimated a lifetime NSSI rate of 24.5%, with approximately half reporting sporadic NSSI and half repetitive NSSI. The time-order associations between onset of NSSI and mental disorders were bidirectional, but mental disorders were stronger predictors of the onset of NSSI (median RR = 1.94) than vice versa (median RR = 1.58). These associations were stronger among individuals engaging in repetitive rather than sporadic NSSI. While associations between NSSI and mental disorders generally did not differ by sex, repetitive NSSI was a stronger predictor for the onset of subsequent substance use disorders among females compared to males. Most mental disorders marginally increased the risk for persistent repetitive NSSI (median RR = 1.23). CONCLUSIONS:Our findings offer unique insights into the temporal order between NSSI and mental disorders. Further work exploring the mechanism underlying these associations will pave the way for early identification and intervention of both NSSI and mental disorders.

High unmet need for treatment of mental disorders exists throughout the world. An understanding of barriers to treatment is needed to develop effective programs to address this problem. Data on barriers were obtained from face-to-face interviews in 22 community surveys across 19 countries (n = 102,812 respondents aged ≥ 18 years, 57.7

Data from the World Mental Health (WMH) surveys on the coverage cascade has underscored the importance of perceived need for seeking treatment of mental disorders. However, little research has focused on treatment contact after adjusting for perceived need. We do so here in analysis of WMH data. The WMH data considered here come from 25 community surveys implemented between 2001 and 2019 across 21 countries. n = 12,508 of the n = 117,739 respondents in these surveys aged 18 and older met criteria for one or more 12-month DSM-IV anxiety, mood, or substance use disorders assessed across all these surveys. Information was obtained about 12-month treatment of each disorder. The predictors considered were disorder type, socio-demographics, and history of prior treatment. Twelve-month treatment was obtained for 17.7

Obstructive sleep apnea (OSA) affects up to 936 million adults globally and is linked to significant health risks, including neurocognitive impairment, cardiovascular diseases, and metabolic conditions. Despite its prevalence, OSA remains largely underdiagnosed. This study aimed to enhance OSA awareness and risk assessment using the STOP-Bang questionnaire in a telemedicine format. During a six-week campaign on a popular Latin American news portal, 5,966 adults completed the STOP-Bang questionnaire. Participants reporting moderate or severe OSA risk were advised to seek clinical evaluation. Among respondents, 44.7

Importance Accurate baseline information about the proportion of people with mental disorders who receive effective treatment is required to assess the success of treatment quality improvement initiatives. Objective To examine the proportion of mental and substance use disorders receiving guideline-consistent treatment in multiple countries. Design, Setting, and Participants In this cross-sectional study, World Mental Health (WMH) surveys were administered to representative adult (aged 18 years and older) household samples in 21 countries. Data were collected between 2001 and 2019 and analyzed between February and July 2024. Twelve-month prevalence and treatment of 9 DSM-IV anxiety, mood, and substance use disorders were assessed with the Composite International Diagnostic Interview. Effective treatment and its components were estimated with cross-tabulations. Multilevel regression models were used to examine predictors. Main Outcomes and Measures The main outcome was proportion of effective treatment received, defined at the disorder level using information about disorder severity and published treatment guidelines regarding adequate medication type, control, and adherence and adequate psychotherapy frequency. Intermediate outcomes included perceived need for treatment, treatment contact separately in the presence and absence of perceived need, and minimally adequate treatment given contact. Individual-level predictors (multivariable disorder profile, sex, age, education, family income, marital status, employment status, and health insurance) and country-level predictors (treatment resources, health care spending, human development indicators, stigma, and discrimination) were traced through intervening outcomes. Results Among the 56 927 respondents (69.3% weighted average response rate), 32 829 (57.7%) were female; the median (IQR) age was 43 (31-57) years. The proportion of 12-month person-disorders receiving effective treatment was 6.9% (SE, 0.3). Low perceived need (46.5%; SE, 0.6), low treatment contact given perceived need (34.1%; SE, 1.0), and low effective treatment given minimally adequate treatment (47.0%; SE, 1.7) were the major barriers, but with substantial variation across disorders. Country-level general medical treatment resources were more important than mental health treatment resources. Other than for the multivariable disorder profile, which was associated with all intermediate outcomes, significant predictors were largely mediated by treatment contact. Conclusions and Relevance In addition to the gaps in treatment quality, these results highlight the importance of increasing perceived need, the largest barrier to effective treatment; the importance of training primary care treatment clinicians in recognition and treatment of mental disorders; the need to improve the continuum of care, especially from minimally adequate to effective treatment; and the importance of bridging the effective treatment gap for men and people with lower education.

OBJECTIVES:To explore the relationship between socioeconomic and health-related changes during the COVID-19 lockdown and sleep quality. METHODS:A panel study was conducted with 667 participants from the Argentine Social Debt Survey in 2019 (pre lockdown), 2020 (during lockdown), and 2021 (post lockdown). Generalized linear mixed-effects models were performed to explore the following predictors of self-reported sleep quality over time: age, educational level, living in poverty, employment status, place of residence, psychological distress, and health status. RESULTS:Reporting poor health and residing in Buenos Aires were associated with poor sleep quality, independent of the lockdown. Advanced age emerged as a significant predictor of poor sleep quality after the lockdown. Differences in sleep quality associated with living in poverty and psychological distress disappeared during lockdown and resumed post lockdown. CONCLUSIONS:This work highlights the importance of the dynamic interplay between socioeconomic and health-related factors when assessing sleep quality. In this urban Argentine panel study, the COVID-19 lockdown appeared to mitigate poverty-related disparities in sleep quality, underscoring the need to refocus attention on these vulnerable subpopulations in the post-lockdown period, when such disparities re-emerged.

PURPOSE:Investigate the sociodemographic disparities in mental health treatment among Canadian college students through quantifying disparities at 2 critical stages of the treatment-seeking pathway: (1) perceived need for treatment and (2) service use, conditional on perceived need. METHODS:Survey data collected at 4 Canadian universities under the World Mental Health International College Student initiative were analyzed. Students meeting the criteria for 12-month mental disorder, substance use disorder, suicidal thoughts and behaviors, and/or non-suicidal self-injury were included in the analyses (N = 8,581). Multivariable logistic regression models were run to evaluate the associations between sociodemographic characteristics and (1) perceived need for mental health treatment and (2) service use, conditional on perceived need. Gender, sexual orientation, race and ethnicity, age, international student status, parental education, and financial stress were investigated as covariates. RESULTS:Students meeting the criteria for a 12-month mental health condition comprised 30.4% of the total sample. Of these students, 78.7% reported a perceived need for mental health treatment and 40.2% reported service use. Gender and sexual orientation disparities appeared to predominately arise from differences in perceived need. Conversely, disparities based on age, international student status, and financial stress appeared to predominately arise from differences in structural barriers. Disparities based on race/ethnicity and parental education appeared to arise from both attitudinal and structural barriers. DISCUSSION:Findings identify key blockage points for further investigation and highlight heterogeneity in the causes underlying sociodemographic disparities in service use, emphasizing the importance of targeted efforts to promote equitable access to mental health treatment for students.

Background. Although melatonin is widely used in Sleep Medicine for its chronobiological action, its potent antioxidant and mitochondrial regulatory effects, as well as its immunomodulatory and anti-inflammatory functions, make it of interest as a cytoprotective agent in several chronic pathologies. These actions are evident at doses higher than those used for sleep disorders. Even at high doses, melatonin’s adverse effects are few, mild, and self-limited or resolve quickly after discontinuation of treatment. Based on its safety profile, we treated melatonin for sleep disorders in the presence of comorbidities with doses ≥ 40 mg daily. Methods. This was a retrospective mixed observational analytical design comprising a retrospective uncontrolled cohort analysis and a cross-sectional study. Eighty-one patients (57 female) with sleep disorders ranging in age from 55 to 98 years (mean 74.4 years) were treated with melatonin 40 to 200 mg daily (mean 72.7 mg) were examined. Fifty-six percent of patients received treatment for more than 4 years. The control group for the cross-sectional analysis included 81 patients over 52 years of age, matched by age and sex and not receiving melatonin but having sleep disorders within the same period. Results. A significant decrease was observed in arterial hypertension, ischemic heart disease and diabetes mellitus after melatonin administration. Analysis of clinical laboratory variables indicated no changes in the treated group versus the untreated group, except for a lower alkaline phosphatase concentration in patients who received melatonin. Conclusions. These findings suggest a beneficial effect of cytoprotective doses of melatonin on the cardiovascular and metabolic profile in an aged population.

BACKGROUND:Alzheimer's disease and other dementias (ADODs) severely threaten the wellbeing of older people, their families, and communities, especially with projected exponential growth. Understanding the macroeconomic implications of ADODs for policy making is essential but under-researched. METHODS:We used a health-augmented macroeconomic model to calculate the macroeconomic burden of ADODs for 152 countries or territories, accounting for: the effect on labour supply of reduced working hours of informal caregivers; the effect on labour supply of ADODs-related mortality and morbidity; age-sex-specific differences in education, work experience, labour market participations, and informal caregivers; and treatment and formal care costs diverting from savings and investments. FINDINGS:ADODs will cost the world economy 14 513 billion international dollars (INT$, measured in the base year 2020; 95% uncertainty interval [UI] 12 106-17 778) from 2020 to 2050, equivalent to 0·421% (95% UI 0·351-0·515) of annual global GDP. Japan incurs the largest annual GDP loss at 1·463% (1·225-1·790). China (INT$2961 billion [2507-3564]), the USA (INT$2331 billion [1989-2829]), and Japan (INT$1758 billion [1471-2150]) face the largest absolute economic burdens. The economic burden of informal care ranges from 60·97% in high-income countries to 85·45% in lower-middle-income countries, and treatment and formal care costs range from 10·50% in lower-middle-income countries to 30·80% in high-income countries. INTERPRETATION:The macroeconomic burden of ADODs is substantial and unequally distributed across countries and regions. Global efforts to reduce the burden, especially with regard to informal care, are urgently needed. FUNDING:National Institute on Aging, National Institutes of Health; Chinese Academy of Engineering; Chinese Academy of Medical Sciences; Bill & Melinda Gates Foundation; Davos Alzheimer's Collaborative through Data for Decisions.

Background Mental health service providers are increasingly interested in patient perspectives. We examined rates and predictors of patient-reported satisfaction and perceived helpfulness in a cross-national general population survey of adults with 12-month DSM-IV disorders who saw a provider for help with their mental health. Methods Data were obtained from epidemiological surveys in the World Mental Health Survey Initiative. Respondents were asked about satisfaction with treatments received from up to 11 different types of providers (very satisfied, satisfied, neither satisfied nor dissatisfied, somewhat dissatisfied, very dissatisfied) and helpfulness of the provider (a lot, some, a little, not at all). We modelled predictors of satisfaction and helpfulness using a dataset of patient-provider observations ( n = 5,248). Results Most treatment was provided by general medical providers (37.4%), psychiatrists (18.4%) and psychologists (12.7%). Most patients were satisfied or very satisfied (65.9-87.5%, across provider) and helped a lot or some (64.4-90.3%). Spiritual advisors and healers were most often rated satisfactory and helpful. Social workers in human services settings were rated lowest on both dimensions. Patients also reported comparatively low satisfaction with general medical doctors and psychiatrists/psychologists and found general medical doctors less helpful than other providers. Men and students reported lower levels of satisfaction than women and nonstudents. Respondents with high education reported higher satisfaction and helpfulness than those with lower education. Type of mental disorder was unrelated to satisfaction but in some cases (depression, bipolar spectrum disorder, social phobia) was associated with low perceived helpfulness. Insurance was unrelated to either satisfaction or perceived helpfulness but in some cases was associated with elevated perceived helpfulness for a given level of satisfaction. Conclusions Satisfaction with and perceived helpfulness of treatment varied as a function of type of provider, service setting, mental status, and socio-demographic variables. Invariably, caution is needed in combining data from multiple countries where there are cultural and service delivery variations. Even so, our findings underscore the utility of patient perspectives in treatment evaluation and may also be relevant in efforts to match patients to treatments.

Abstract Objective The standard method of generating disorder‐specific disability scores has lay raters make rankings between pairs of disorders based on brief disorder vignettes. This method introduces bias due to differential rater knowledge of disorders and inability to disentangle the disability due to disorders from the disability due to comorbidities. Methods We propose an alternative, data‐driven, method of generating disorder‐specific disability scores that assesses disorders in a sample of individuals either from population medical registry data or population survey self‐reports and uses Generalized Random Forests (GRF) to predict global (rather than disorder‐specific) disability assessed by clinician ratings or by survey respondent self‐reports. This method also provides a principled basis for studying patterns and predictors of heterogeneity in disorder‐specific disability. We illustrate this method by analyzing data for 16 disorders assessed in the World Mental Health Surveys (n = 53,645). Results Adjustments for comorbidity decreased estimates of disorder‐specific disability substantially. Estimates were generally somewhat higher with GRF than conventional multivariable regression models. Heterogeneity was nonsignificant. Conclusions The results show clearly that the proposed approach is practical, and that adjustment is needed for comorbidities to obtain accurate estimates of disorder‐specific disability. Expansion to a wider range of disorders would likely find more evidence for heterogeneity.