The COVID-19 pandemic has affected perinatal mental health. Reliable tools are needed to assess perinatal stress during pandemic situations.

To assess the psychometric properties of the Greek versions of the Pandemic-Related Pregnancy Stress Scale (PREPS) and the Pandemic-Related Postpartum Stress Scale (PREPS-PP) and to explore the associations between women's characteristics and perinatal stress during the second pandemic wave.

The PREPS and PREPS-PP were completed by 264 pregnant and 188 postpartum women, respectively, who also completed the State-Trait Anxiety Inventory (STAI) and the Edinburgh Perinatal Depression Scale (EPDS).

The internal consistency was similar for PREPS and PREPS-PP. It was good for preparedness stress (a = 0.77 and α = 0.71, respectively) and infection stress (α = 0.83 for both scales) but low for positive appraisal (α = 0.46 and α = 0.41, respectively). Of the pregnant women, 55.33% and 55.27%, respectively, reported scores of ≥40 on STAI-S and STAI-T, and the respective percentages for the postpartum women were 47.34% and 46.80%. In addition, 14.39% of the pregnant women and 20.74% of the postpartum women scored ≥13 on the EPDS. Higher preparedness stress on PREPS and PREPS-PP was associated with primiparity (P = 0.022 and P = 0.021, respectively) and disrupted perinatal care (P = 0.069 and P = 0.007, respectively). In postpartum women, higher infection stress was associated with chronic disease (P = 0.037), primiparity (P = 0.02) and perceived risk of infection (P = 0.065). Higher score on infection stress was associated with disrupted perinatal care in both groups (P = 0.107 and P = 0.010, respectively).

The Greek versions of PREPS and PREPS-PP are valid tools for the assessment of women at risk of perinatal stress during a health crisis.

Prenatal selective serotonin reuptake inhibitor (SSRI) antidepressant exposure is associated with increased internalising and anxious behaviours in young children; whether this continues into early adolescence is unknown. Also, it is not well established whether it is the in utero exposure to SSRIs or the underlying maternal mood that contributes more to these associations.

To examine associations between maternal depressive symptoms, prenatal SSRI antidepressant treatment and internalising and anxiety behaviours from childhood into pre-adolescence.

From a prospective longitudinal cohort, measures of maternal depressive symptoms and SSRI use and child outcomes (n = 191 births) were obtained from the second trimester to 12 years. Maternal reports of internalising and anxiety behaviours in children were obtained at 3, 6 and 12 years.

Multilevel mixed-effects models revealed that maternal depressed mood at the third trimester assessment, not prenatal SSRI exposure, was associated with longitudinal patterns of higher levels of internalising and anxiety behaviours across childhood from 3 to 12 years of age. At each age, hierarchical regressions showed that maternal mood at the third trimester, compared with current maternal depression or prenatal SSRI exposure, explained a greater proportion of the variance in internalising and anxiety behaviours.

Even with prenatal SSRI treatment, maternal depressed mood during the third trimester still had an enduring effect as it was associated with increased levels of internalising and anxiety behaviours across childhood and into early adolescence. Importantly, we found no evidence of a ‘main effect’ association between prenatal SSRI exposure and internalising and anxiety behaviours in children.

Clinicians working in mental health in-patient settings may have to use nasogastric tube feeding under physical restraint to reverse the life-threatening consequences of malnutrition when this is driven by a psychiatric condition such as a restrictive eating disorder.

To understand the decision-making process when nasogastric tube feeding under restraint is initiated in mental health in-patient settings.

People with lived experience of nasogastric tube feeding under restraint and parents/carers were recruited via the website of the UK's eating disorder charity BEAT. Eating disorder clinicians were recruited via an online post by the British Eating Disorders Society. Semi-structured interviews were administered to all participants.

Themes overlapped between the participant groups and were integrated in the final analysis. Two main themes were generated: first, ‘quick decisions’, with the subthemes of ‘medical risk’, ‘impact of not eating’ and ‘limited discussions’; second, ‘slow decisions’, with subthemes of ‘threats’, ‘discussions with patient’, ‘not giving up’ and ‘advanced directives’. Benefits and harms of both quick and slow decisions were identified.

This research offers a new perspective regarding how clinical teams can make best practice decisions regarding initiating nasogastric feeding under restraint. In-patient mental health teams facilitating this clinical intervention should consider discussing it with the patient at the beginning of their admission in anticipation of the need for emergency intervention and in full collaboration with the multidisciplinary team.

Liaison psychiatry provision for children and young people in England is poorly evaluated.

We sought to evaluate paediatric liaison psychiatry provision and develop recommendations to improve practice.

The liaison psychiatry surveys of England (LPSE) cross-sectional surveys engage all liaison psychiatry services in England. Services are systematically identified by contacting all acute hospitals with emergency departments in England. Questions are developed in consultation with NHS England and the Royal College of Psychiatrists’ Faculty of Liaison Psychiatry, and updated based on feedback. Responses are submitted by email, post or telephone. Questions on paediatric services were included from 2015 (LPSE-2), and we analysed data from this and the subsequent four surveys.

The number of acute hospitals with access to paediatric liaison psychiatry services increased from 29 (15.9%) in 2015 to 46 (26.6%) in 2019, compared with 100% provision for adults. For LPSE-4, only one site met the Core-24 criteria of 11 full-time equivalent mental health practitioners and 1.5 full-time equivalent consultants, and for LPSE-5, just two sites exceeded them. Acute hospitals with access to 24/7 paediatric liaison psychiatry services increased from 12 to 19% between LPSE-4 and LPSE-5. The proportion of paediatric liaison psychiatry services based offsite decreased from 30 to 24%.

There is an unacceptable under-provision of paediatric liaison psychiatry services compared with provision for adults. Number of services, staffing levels and hours of operation have increased, but continued improvement is required, as few services meet the Core-24 criteria.

Research into neural mechanisms underlying cue-induced cigarette craving has attracted considerable attention for its significant role in treatments. However, there is little understanding about the effects of exposure to smoking-related cues on electroencephalogram (EEG) microstates of smokers, which can reflect abnormal brain network activity in several psychiatric disorders.

To explore whether abnormal brain network activity in smokers on exposure to smoking-related cues would be captured by EEG microstates.

Forty smokers were exposed to smoking and neutral imagery conditions (cues) during EEG recording. Behavioural data and parameters for microstate topographies associated with the auditory (A), visual (B), salience and memory (C) and dorsal attention networks (D) were compared between conditions. Correlations between microstate parameters and cigarette craving as well as nicotine addiction characteristics were also analysed.

The smoking condition elicited a significant increase in the duration of microstate classes B and C and in the duration and contribution of class D compared with the neutral condition. A significant positive correlation between the increased duration of class C (smoking minus neutral) and increased craving ratings was observed, which was fully mediated by increased posterior alpha power. The increased duration and contribution of class D were both positively correlated with years of smoking.

Our results indicate that smokers showed abnormal EEG microstates when exposed to smoking-related cues compared with neutral cues. Importantly, microstate class C (duration) might be a biomarker of cue-induced cigarette craving, and class D (duration and contribution) might reflect the relationship between cue-elicited activation of the dorsal attention network and years of smoking.

Current psychiatric diagnoses, although heritable, have not been clearly mapped onto distinct underlying pathogenic processes. The same symptoms often occur in multiple disorders, and a substantial proportion of both genetic and environmental risk factors are shared across disorders. However, the relationship between shared symptoms and shared genetic liability is still poorly understood.

Well-characterised, cross-disorder samples are needed to investigate this matter, but few currently exist. Our aim is to develop procedures to purposely curate and aggregate genotypic and phenotypic data in psychiatric research.

As part of the Cardiff MRC Mental Health Data Pathfinder initiative, we have curated and harmonised phenotypic and genetic information from 15 studies to create a new data repository, DRAGON-Data. To date, DRAGON-Data includes over 45 000 individuals: adults and children with neurodevelopmental or psychiatric diagnoses, affected probands within collected families and individuals who carry a known neurodevelopmental risk copy number variant.

We have processed the available phenotype information to derive core variables that can be reliably analysed across groups. In addition, all data-sets with genotype information have undergone rigorous quality control, imputation, copy number variant calling and polygenic score generation.

DRAGON-Data combines genetic and non-genetic information, and is available as a resource for research across traditional psychiatric diagnostic categories. Algorithms and pipelines used for data harmonisation are currently publicly available for the scientific community, and an appropriate data-sharing protocol will be developed as part of ongoing projects (DATAMIND) in partnership with Health Data Research UK.

Increased rates of visual impairment are observed in people with schizophrenia.

We assessed whether genetically predicted poor distance acuity is causally associated with schizophrenia, and whether genetically predicted schizophrenia is causally associated with poorer visual acuity.

We used bidirectional, two-sample Mendelian randomisation to assess the effect of poor distance acuity on schizophrenia risk, poorer visual acuity on schizophrenia risk and schizophrenia on visual acuity, in European and East Asian ancestry samples ranging from approximately 14 000 to 500 000 participants. Genetic instrumental variables were obtained from the largest available summary statistics: for schizophrenia, from the Psychiatric Genomics Consortium; for visual acuity, from the UK Biobank; and for poor distance acuity, from a meta-analysis of case–control samples. We used the inverse variance-weighted method and sensitivity analyses to test validity of results.

We found little evidence that poor distance acuity was causally associated with schizophrenia (odds ratio 1.00, 95% CI 0.91–1.10). Genetically predicted schizophrenia was associated with poorer visual acuity (mean difference in logMAR score: 0.024, 95% CI 0.014–0.033) in European ancestry samples, with a similar but less precise effect that in smaller East Asian ancestry samples (mean difference: 0.186, 95% CI –0.008 to 0.379).

Genetic evidence supports schizophrenia being a causal risk factor for poorer visual acuity, but not the converse. This highlights the importance of visual care for people with psychosis and refutes previous hypotheses that visual impairment is a potential target for prevention of schizophrenia.

Timely provision of aftercare following self-harm may reduce risks of repetition and premature death, but existing services are frequently reported as being inadequate.

To explore barriers and facilitators to accessing aftercare and psychological therapies for patients presenting to hospital following self-harm, from the perspective of liaison psychiatry practitioners.

Between March 2019 and December 2020, we interviewed 51 staff members across 32 liaison psychiatry services in England. We used thematic analyses to interpret the interview data.

Barriers to accessing services may heighten risk of further self-harm for patients and burnout for staff. Barriers included: perceived risk, exclusionary thresholds, long waiting times, siloed working and bureaucracy. Strategies to increase access to aftercare included: (a) improving assessments and care plans via input from skilled staff working in multidisciplinary teams (e.g. including social workers and clinical psychologists); (b) supporting staff to focus on assessments as therapeutic intervention; (c) probing boundaries and involving senior staff to negotiate risk and advocate for patients; and (d) building relationships and integration across services.

Our findings highlight practitioners’ views on barriers to accessing aftercare and strategies to circumvent some of these impediments. Provision of aftercare and psychological therapies as part of the liaison psychiatry service were deemed as an essential mechanism for optimising patient safety and experience and staff well-being. To close treatment gaps and reduce inequalities, it is important to work closely with staff and patients, learn from experiences of good practice and implement change more widely across services.

Previous studies have not investigated response rates after electroconvulsive therapy (ECT) in patients with non-psychotic treatment-resistant depression (TRD).

To assess and compare the response rate of ECT for patients with TRD and non-TRD, in a large and clinically representative patient sample.

Patients aged ≥18 years, who were treated for a unipolar, non-psychotic depressive episode with at least one ECT session as part of a first-time, index ECT series between 1 January 2011 and 31 December 2017 were included from the Swedish National Quality Register for ECT. Patients who had initiated a third consecutive trial of antidepressants or add-on medications before start of ECT were classified as having TRD. Patients not meeting criteria for TRD were classified as non-TRD. The main outcome was response to ECT according to the Clinical Global Impressions – Improvement Scale (CGI-I), scored as 1 or 2 (‘very much’ or ‘much improved’ after ECT, respectively). Logistic regression was used to compare outcome measures between TRD and non-TRD, adjusting for potential confounders.

A total of 4244 patients were included. Of these, 1121 patients had TRD and 3123 patients had non-TRD. The CGI-I response rate was 65.9% in the TRD group compared with 75.9% in the non-TRD group (adjusted odds ratio 0.64, 95% CI 0.54–0.75). Older age and more severe depression were predictors of response in patients with TRD.

A clear majority of patients with TRD, as well as patients with non-TRD, responded to ECT, although the response rate was somewhat lower for TRD.

Mental health services are encouraged to use language consistent with principles of recovery-oriented practice. This study presents a novel approach for identifying whether clinical documentation contains recovery-oriented rehabilitation language, and evaluates an intervention to improve the language used within a community-based rehabilitation team.

This is a pilot study of training to enhance recovery-oriented rehabilitation language written in care review summaries, as measured through a text-based analysis of language used in mental health clinical documentation.

Eleven case managers participated in a programme that included instruction in recovery-oriented rehabilitation principles. Outcomes were measured with automated textual analysis of clinical documentation, using a custom-built dictionary of rehabilitation-consistent, person-centred and pejorative terms. Automated analyses were run on Konstanz Information Miner (KNIME), an open-source data analytics platform. Differences in the frequency of term categories in 50 pre-training and 77 post-training documents were analysed with inferential statistics.

The average percentage of sentences with recovery-oriented rehabilitation terms increased from 37% before the intervention to 48% afterward, a relative increase of 28% (P < 0.001). There was no significant change in use of person-centred or pejorative terms, possibly because of a relatively high frequency of person-centred language (22% of sentences) and low use of pejorative language (2.3% of sentences) at baseline.

This computer-driven textual analysis method identified improvements in recovery-oriented rehabilitation language following training. Our study suggests that brief interventions can affect the language of clinical documentation, and that automated text-analysis may represent a promising approach for rapidly assessing recovery-oriented rehabilitation language in mental health services.

Little is known about the reasons for suicidal behaviour in Africa, and communities’ perception of suicide prevention. A contextualised understanding of these reasons is important in guiding the implementation of potential suicide prevention interventions in specific settings.

To understand ideas, experiences and opinions on reasons contributing to suicidal behaviour in the Coast region of Kenya, and provide recommendations for suicide prevention.

We conducted a qualitative study with various groups of key informants residing in the Coast region of Kenya, using in-depth interviews. Audio-recorded interviews were transcribed and translated from the local language before thematic inductive content analysis.

From the 25 in-depth interviews, we identified four key themes as reasons given for suicidal behaviour: interpersonal and relationship problems, financial and economic difficulties, mental health conditions and religious and cultural influences. These reasons were observed to be interrelated with each other and well-aligned to the suggested recommendations for suicide prevention. We found six key recommendations from our thematic content analysis: (a) increasing access to counselling and social support, (b) improving mental health awareness and skills training, (c) restriction of suicide means, (d) decriminalisation of suicide, (e) economic and education empowerment and (f) encouraging religion and spirituality.

The reasons for suicidal behaviour are comparable with high-income countries, but suggested prevention strategies are more contextualised to our setting. A multifaceted approach in preventing suicide in (coastal) Kenya is warranted based on the varied reasons suggested. Community-based interventions will likely improve and increase access to suicide prevention in this study area.

A critical step in research on the epidemiology of post-traumatic stress disorder (PTSD) in low-resource settings is the validation of brief self-reported psychometric tools available in the public domain, such as the Impact Event Scale – Revised (IES-R).

We aimed to investigate the validity of the IES-R in a primary healthcare setting in Harare, Zimbabwe.

We analysed data from a survey of 264 consecutively sampled adults (mean age 38 years; 78% female). We estimated the area under the receiver operating characteristic curve and sensitivity, specificity and likelihood ratios for different cut-off points of the IES-R, against a diagnosis of PTSD made using the Structured Clinical Interview for DSM-IV. We performed factor analysis to evaluate construct validity of the IES-R.

The prevalence of PTSD was 23.9% (95% CI 18.9–29.5). The area under the curve for the IES-R was 0.90. At a cut-off of ≥47, the sensitivity of the IES-R to detect PTSD was 84.1 (95% CI 72.7–92.1) and specificity was 81.1 (95% CI 75.0–86.3). Positive and negative likelihood ratios were 4.45 and 0.20, respectively. Factor analysis revealed a two-factor solution, with both factors showing good internal consistency (Cronbach's factor-1 α = 0.95, factor-2 α = 0.76). In a post hoc analysis, we found the brief six-item IES-6 also performed well, with an area under the curve of 0.87 and optimal cut-off of 15.

The IES-R and IES-6 had good psychometric properties and performed well for indicating possible PTSD, but at higher cut-off points than those recommended in the Global North.

The quality of the relationship between mother and infant may have profound implications for the development of a child. Early indicators of psychological vulnerability may allow targeting of support for the child's cognitive, emotional and social development. A challenging mother–infant relationship could be one indicator of risk.

This study examined variations in psychological well-being and psychopathology among boys and girls according to early maternal perception of the mother–infant relationship.

This study is based on 64 663 mother–infant pairs from the Danish National Birth Cohort, for which data on the mother–infant relationship were collected at 6 months postpartum. Behavioural problems were assessed with the Danish version of the Strengths and Difficulties Questionnaire (SDQ) at child ages 7, 11 and 18 years, and we retrieved information on diagnosed childhood and adolescent psychiatric disorders and prescriptions of psychotropic drugs from Danish registries.

Children in the challenging mother–infant relationship group had higher odds of behavioural problems at age 7 among both boys and girls. The same pattern of elevated estimates was identified for boys across all SDQ domains and for girls in three of five SDQ domains. All associations were attenuated at age 18, but increased odds of behavioural problems still existed. A challenging early mother–infant relationship increased the offspring's risk of being diagnosed with a psychiatric disorder or being prescribed a psychotropic drug before the age of 18.

A challenging self-reported mother–infant relationship was associated with later psychopathological difficulties. Routine clinical enquiry may be useful in identification of future vulnerability.

Psychological and pharmacological therapies are the recommended first-line treatments for common mental disorders (CMDs) but may not be universally accessible or utilised.

To determine the extent to which primary care patients with CMDs receive treatment and the impact of sociodemographic, work-related and clinical factors on treatment receipt.

National registers were used to identify all Stockholm County residents aged 19–64 years who had received at least one CMD diagnosis (depression, anxiety, stress-related) in primary care between 2014 and 2018. Individuals were followed from the date of their first observed CMD diagnosis until the end of 2019 to determine treatment receipt. Associations between patient factors and treatment group were examined using multinomial logistic regression.

Among 223 271 individuals with CMDs, 30.6% received pharmacotherapy only, 16.5% received psychological therapy only, 43.1% received both and 9.8% had no treatment. The odds of receiving any treatment were lower among males (odds ratio (OR) range = 0.76 to 0.92, 95% CI[minimum, maximum] 0.74 to 0.95), individuals born outside of Sweden (OR range = 0.67 to 0.93, 95% CI[minimum, maximum] 0.65 to 0.99) and those with stress-related disorders only (OR range = 0.21 to 0.51, 95% CI[minimum, maximum] 0.20 to 0.53). Among the patient factors examined, CMD diagnostic group, prior treatment in secondary psychiatric care and age made the largest contributions to the model (R2 difference: 16.05%, 1.72% and 1.61%, respectively).

Although over 90% of primary care patients with CMDs received pharmacological and/or psychological therapy, specific patient groups were less likely to receive treatment.

Music therapy can lift mood and reduce agitation for people living with dementia (PwD) in community and residential care settings, potentially reducing the prevalence of distress behaviours. However, less is known about the impact of music therapy on in-patient psychiatric wards for PwD.

To investigate the impact of music therapy on two in-patient psychiatric wards for PwD.

A mixed-methods design was used. Statistical analysis was conducted on incidents involving behaviours reported as ‘disruptive and aggressive’ in 2020, when music therapy delivery varied because of the COVID-19 pandemic. Semi-structured interviews conducted online with three music therapists and eight ward-based staff were analysed using reflexive thematic analysis.

Quantitative findings showed a significant reduction in the frequency of behaviours reported as disruptive and aggressive on days with in-person music therapy (every 14 days) than on the same weekday with no or online music therapy (every 3.3 or 3.1 days, respectively). Qualitative findings support this, with music therapy reported by music therapists and staff members to be accessible and meaningful, lifting mood and reducing agitation, with benefits potentially lasting throughout the day and affecting the ward environment.

We identified a significant reduction in the occurrence of distress behaviours on days with in-person music therapy when compared with no music therapy. Music therapy was reported to be a valuable intervention, supporting patient mood and reducing agitation. Interventional studies are needed to investigate the impact of music therapy and its optimum mode of delivery.

People with severe mental illness (SMI) die earlier than the general population, primarily because of physical disorders.

We estimated the prevalence of physical health conditions, health risk behaviours, access to healthcare and health risk modification advice in people with SMI in Bangladesh, India and Pakistan, and compared results with the general population.

We conducted a cross-sectional survey in adults with SMI attending mental hospitals in Bangladesh, India and Pakistan. Data were collected on non-communicable diseases, their risk factors, health risk behaviours, treatments, health risk modification advice, common mental disorders, health-related quality of life and infectious diseases. We performed a descriptive analysis and compared our findings with the general population in the World Health Organization (WHO) ‘STEPwise Approach to Surveillance of NCDs’ reports.

We recruited 3989 participants with SMI, of which 11% had diabetes, 23.3% had hypertension or high blood pressure and 46.3% had overweight or obesity. We found that 70.8% of participants with diabetes, high blood pressure and hypercholesterolemia were previously undiagnosed; of those diagnosed, only around half were receiving treatment. A total of 47% of men and 14% of women used tobacco; 45.6% and 89.1% of participants did not meet WHO recommendations for physical activity and fruit and vegetable intake, respectively. Compared with the general population, people with SMI were more likely to have diabetes, hypercholesterolemia and overweight or obesity, and less likely to receive tobacco cessation and weight management advice.

We found significant gaps in detection, prevention and treatment of non-communicable diseases and their risk factors in people with SMI.

Prisoners with an intellectual disability are overrepresented in custody and more likely to reoffend and be reincarcerated compared with the general prison population. Although prisoners with intellectual disability have many of the same risk factors for recidivism as the general prison population, the high rates of mental illness experienced by this group are key drivers of recidivism.

We aimed to assess the impact of provision of post-release disability and community mental health support on rates of reincarceration in a cohort with identified intellectual disability and serious mental illness diagnosis.

We conducted a historical cohort study using linked administrative data-sets, including data on hospital admissions, community mental health, disability support and corrections custody in New South Wales, Australia (n = 484). To assess the time to return to adult custody, we used survival analysis on multiple failure-time data.

Over the median follow-up period of 7.4 years, 73.7% (357) received community mental health support, 19.8% (96) received disability support and 18.6% (85) received a combination of supports during a post-release period from prison. Lower hazards of reincarceration in a post-release period were associated with receipt of community mental health support (hazard ratio [HR] = 0.58, CI 0.49–0.69, P < 0.001), or a combination of community mental health and disability support (HR = 0.46, CI 0.34–0.61, P < 0.001).

High rates of reincarceration for prisoners with intellectual disability and history of serious mental illness may be modifiable by provision of appropriate mental health and disability supports.

Little is known about the effect of ethnicity on the response to antipsychotic medication in patients with schizophrenia.

To determine whether ethnicity moderates the response to antipsychotic medication in patients with schizophrenia, and whether this moderation is independent of confounders.

We analysed 18 short-term, placebo-controlled registration trials of atypical antipsychotic medications in patients with schizophrenia (N = 3880). A two-step, random-effects, individual patient data meta-analysis was applied to establish the moderating effect of ethnicity (White versus Black) on symptom improvement according to the Brief Psychiatric Rating Scale (BPRS) and on response, defined as >30% BPRS reduction. These analyses were corrected for baseline severity, baseline negative symptoms, age and gender. A conventional meta-analysis was performed to determine the effect size of antipsychotic treatment for each ethnic group separately.

In the complete data-set, 61% of patients were White, 25.6% of patients were Black and 13.4% of patients were of other ethnicities. Ethnicity did not moderate the efficacy of antipsychotic treatment: pooled β-coefficient for the interaction between treatment and ethnic group was −0.582 (95% CI −2.567 to 1.412) for mean BPRS change, with an odds ratio of 0.875 (95% CI 0.510–1.499) for response. These results were not modified by confounders.

Atypical antipsychotic medication is equally effective in both Black and White patients with schizophrenia. In registration trials, White and Black patients were overrepresented relative to other ethnic groups, limiting the generalisability of our findings.

Various effective psychotherapies exist for the treatment of depression; however, only approximately half of patients recover after treatment. In efforts to improve clinical outcomes, research has focused on personalised psychotherapy – an attempt to match patients to treatments they are most likely to respond to.

The present research aimed to evaluate the benefit of a data-driven model to support clinical decision-making in differential treatment allocation to cognitive–behavioural therapy versus counselling for depression.

The present analysis used electronic healthcare records from primary care psychological therapy services for patients receiving cognitive–behavioural therapy (n = 14 544) and counselling for depression (n = 4725). A linear regression with baseline sociodemographic and clinical characteristics was used to differentially predict post-treatment Patient Health Questionnaire (PHQ-9) scores between the two treatments. The benefit of differential prescription was evaluated in a held-out validation sample.

On average, patients who received their model-indicated optimal treatment saw a greater improvement (by 1.78 PHQ-9 points). This translated into 4–10% more patients achieving clinically meaningful changes. However, for individual patients, the estimated differences in benefits of treatments were small and rarely met the threshold for minimal clinically important differences.

Precision prescription of psychotherapy based on sociodemographic and clinical characteristics is unlikely to produce large benefits for individual patients. However, the benefits may be meaningful from an aggregate public health perspective when applied at scale.