Psychiatric Morbidity and Violence in England and Wales

This extension agreement is to allow the Queen Mary University of London to continue to hold and process the Adult Psychiatric Morbidity Survey (APMS) 2014 data disseminated under the previous version of the agreement.

Covid-related disruption has impacted the research fellow time planned for this work. The substantive impacts were:

1) Some universities, including Queen Mary University of London, elected not to renew fixed-term contracts during the coronavirus pandemic due to income uncertainty. One of the intended researchers on the project was affected by this and

2) Universities in China placed restrictions on working with foreign nationals, which directly affected one of our investigating Professors, who had a Chinese academic scheduled to come to the UK as a university fellow to complete a specific piece of analysis.

3) The above restriction has also impacted an investigating Professor, who is currently unable to return to his substantive academic position in West China.

********************************

v0.4

The Centre for Psychiatry at Queen Mary University of London (QMUL) is seeking to use the Adult Psychiatric Morbidity Survey (APMS) 2014 data to extend and augment existing academic work around the epidemiology of violence that uses the 2000 and 2007 datasets. This work will ensure up-to-date knowledge of the intricate link between mental disorder and violence with the population of England and Wales.

This application is to make use of the Adult Psychiatric Morbidity Survey, 2014, a research dataset held by UKDS relating to prevalence of mental disorder in the UK population for the purposes of conducting research into the epidemiology of violence and mental disorder in the UK, suitable for publication in academic peer-reviewed journals. It is justified under two sections of the GDPR:

a) Article 6 (1)(e) - a legitimate public interest, on the basis that a greater understanding of the causes and predictors of violence can aid preventive measures to reduce the burden of violence on the population; and

b) Article 9 (2)(j) - in the public interest for scientific research purposes.

The study team confirm that there are no moral or ethical issues in the use of this dataset as the data has already been collected and epidemiological research published based upon it, and also that there will make no efforts to link the data to other datasets or to de-anonymise individuals.

The data-set requested is the fourth repetition of a population-level investigation of mental disorder that has taken place every 7 years since 1993. It represents one of the largest and most inclusive data-sets relating to the prevalence of mental disorder, service use and negative outcomes, including violence, in the world.

This work is a continuation of a programme of work begun by two of the research team in a National Institute for Health Research (NIHR) funded Programme of research (Improving risk management for violence in mental health services: a multimethods approach: RP-PG-0407-10500) that ran between 2008 and 2014.

Several of the authors have already published work relating to the key research questions based on the 2000 and 2007 APMS datasets (Coid et al., 2006; Bhui et al., 2009 - fuller reference list available at request), but there is a public interest in updating this work to ensure the findings are robust over time and relevant to the current social context.

Data Subjects

This data-set comprises demographic and mental health screening questionnaire data on 7,528 individuals sampled from the UK household population, with additional clinical data on a subset of 630 individuals who met criteria for at least one mental disorder.

Data Analysis

The processing activities will be conducted along five work streams: i) longitudinal trends in violence ; ii) alcohol use and violence; iii) affective symptoms of borderline personality disorder, iv) risk factors within psychosis in the general population; and v) intimate partner violence and bullying.

Workstream 1: Longitduinal Trends in Violence

Evidence from other sources suggests that violence in England and Wales has declined over the past decade. This study aims to investigate hypothesised risk factor variables associated with self-reported violence in each survey and to observe their decline over time based on a combination of the 2000, 2007 and 2014 data-sets. These factors will include measures of psychiatric morbidity and social and demographic risk factors, as used in previous studies.

Workstream 2: Alcohol and Violence

Previous research using the 2000 and 2007 APMS data-sets (Freestone et al., 2019) has shown a linear relationship between alcohol use and likelihood of both perpetration of violent incidents and victimisation. This finding duplicates previous work by Public Health England suggesting that annually up to 20% of the population are negatively impacted by drinking (Benyon et al, 2019). This study will seek to test two core hypotheses:

1. Does the linear relationship between alcohol use and the likelihood of violence perpetration observed in previous cohorts hold true in the 2014 study?

2. Do demographic risk factors such as age, social status and income affect the relationship between alcohol and violence?

Workstream 3: Affective symptoms of borderline personality disorder

This study will use the categorical diagnosis and criteria items of borderline personality disorder to investigate the associations with dependent continuous variables of affective and anxiety disorder measures using the Clinical Interview Schedule-Revised (CIS-R) and psychotic symptoms in the Psychosis Screening Questionnaire (PSQ). It will require access to these diagnostic screen variables at both phases of interview as well as demographic variables that may affect this relationship and negative outcome measures such as violence, homelessness and financial difficulties.

Workstream 4: Modelling risk factors across the Spectrum of Psychosis in the General Population

This study will utlise both the phase 1 and phase 2 data. Participants will be divided into 5 subgroups using the (PSQ), and those who receive a diagnosis of clinical psychosis in phase 2 to create the population “spectrum”. Subgroups will range from the following: 5 (clinical diagnosis of psychosis), 4 (scores of 4-5 PSQ, excluding any from 5), 3 (scores of 2-3 PSQ), 2 (score 1), 1 (score 0). A statistical model will be used in which the spectrum is the dependent variable and a range of putative risk factors for psychosis are the independent variables to test linearity and extra-linearity across the spectrum. Independent variables will include demography, comorbid psychopathology including substance misuse, social environment, and traumatic risk factors.

Workstream 5: Violence exposure, physical health and mental health

It is well reported that the potential impact of exposure to incidents of violence (i.e. intimate partner violence, child maltreatment), in all its forms, on the individual and society is a serious public health issue. Exposure to victimisation is well documented as increasing the risk of mental and physical health problems (Arseneault 2018; Hughes et al 2017; Simmons, Knight, & Menard, 2018). Furthermore, studies also suggest that there is an association between mental and physical health (Nabi et al., 2008, Surtees et al., 2008).

This work-stream aims to investigate the relationship between exposure to violence (i.e. trauma, neglect and intimate partner violence) and physical health and mental health problems. It will address this via four research questions:

1. Do common mental health disorder symptoms (i.e. disturbances in sleep, fatigue and worry) mediate the relationship between violence (neglect, trauma, and intimate partner violence) and physical health problems?

2. Does childhood trauma and neglect moderate the relationship between mental health problems (i.e. bipolar disorder) and physical health problems?

3. Does intimate partner violence contribute to the relationship between childhood trauma and neglect and physical health?

4. Does intimate partner violence contribute to the relationship between childhood trauma and neglect and mental health?

To answer these research questions, the study will conduct an analysis of the APMS 2014 data-set, usually in combination with the 2000 and/or 2007 to provide a longitudinal analysis of trends at the population level. This data-set is pseudonymised and no linkage with other data-sets or personal information will be possible or necessary as part of this project. Usage of summary or aggregate data relating to this data-set is unfortunately not possible as several of the analyses proposed require the use of 'comparable' variables across data-sets (e.g. 2000 and 2014) and/or inferential statistical methods such as regression or path analysis.

The study team anticipate that the analysis stage of this project will take approximately three years, after which academic archiving policies require that the data be retained in an archive format for a further 7 years in compliance with the need to retain academic data for further scrutiny following publication of results.

Processing will be conducted by the Centre for Psychiatry, Wolfson Institute for Preventive Medicine, School of Medicine and Dentistry, Queen Mary University of London. Data will be held securely by the Barts Cancer Centre, which is also a part of the medical school and provides secure network and data storage services to the Centre of Psychiatry within the Wolfson Institute as well as Barts Cancer Institute. Data will be stored using the remote server that allows secure storage of data for remote analysis.

There are no funders or commissioners involved in this research.

The sole data controller who also process data for this purpose will be Queen Mary University of London.

This extension agreement is to allow the Queen Mary University of London to continue to hold and process the APMS 2014 data disseminated under the previous version of the agreement.

********************************

v0.4

This study will involve no data flow into NHS digital.

This study requires use of the Adult Psychiatric Morbidity Survey 2014, which is a cross-sectional dataset containing pseudonymised information about participants to a large population-level study with no participant-identifiable information.

After receipt by QMUL, data will be securely held on a secure server. After the conclusion of this project, data will be archived by the University for a period of seven years, then securely deleted - subject to appropriate agreements with NHS Digital being in place to allow for retention of this data.

Data will be stored on a secure server operated by the Barts Cancer Centre at Queen Mary University of London which operates on a single-user, dual-authentication password protected system.

Data processing will be conducted by academic or research staff holding substantive, honorary or emeritus contracts with Queen Mary University of London only. Processing will be restricted to the statistical analysis of data using an approved list of software packages, including: R, STATA and SPSS.

At no stage will raw data leave the secure server: only aggregate-level data (summary, descriptive data and outputs of statistical analysis) in Microsoft Word or Excel format will be produced as a result of the analysis and stored outside the server.

No data linkage is being proposed.

The study team confirm there is no requirement here to re-identify participants for any purpose.

All those involved with data processing will be substantive employees of the data processor and will be expected to have completed the NHS Health for England Data Security Awareness (NHSD) Level 1 Training course at minimum before access is granted.

All those accessing the data must use two-factor authentication via their work email address

The APMS dataset will be received in a pseudo-anonymised format that will make identification of individuals impossible to those not involved with the original data collection. No linking of the dataset to 'live' health records or any other data sources is proposed.

The original outputs still stand with amended timelines as below:

All workstreams will be produced as a result of the data processing.

The data processing will result in research outputs comprising:

a. Submissions to peer reviewed journals

b. Presentations at national and international academic conferences

c. Discussions through national and international associations and clinical interest groups

The study team anticipate that each workstream will produce at least one academic paper relating to the key research question(s). The findings will also be disseminated via academic seminars, at national and international conferences (including the Royal College of Psychiatrists Division of Forensic Psychiatry conference and the British Psychological Society's annual Forensic Psychology conference) and the annual East London Research Day held in November of each year.

A summary findings report can be provided to stakeholders at the end of the project (May 2023)

All work streams will contain only high-level aggregate data (i.e. whole sample or sub-sample descriptive) or inferential statistics (i.e. model coefficients with p values) will be contained in any outputs. No sub-groups with < 5 participants will be reported.

All workstreams will comprise a programme of epidemiological research that is geared towards researchers working on the topics of mental health and violence. The primary outputs of this work will be in the form of peer-reviewed academic articles submitted to recognised, impact factor journals. Where possible on the basis of publication agreements, these will be open access and/or accepted drafts will be made open access via the Queen Mary repository.

Some workstreams require a different approach: Workstreams 1,2, and 5 have a public health focus whereas 3 and 4 are more technical enquiries relating to nosological research in psychiatry.

Workstreams 1 ,2 and 5

The aim will be to publish the first paper from this study in a high impact journal. It may be attractive to a US psychiatric journal because the fall in violence, despite what many believe, and newspapers inform, extends to most countries, including the US and even developing countries with very high rates. There can be no precise answer to this question until the results are available. If there are findings that suggest that violence has fallen among those with mental disorders, or if the change is not due to this subgroup of the population, this is very important for psychiatry. If the findings suggest it is due to demographic changes and patterns of alcohol use, for example, then it will determine a different publication strategy.

Findings from these work-streams will also be disseminated through the research and clinical groupings around the Centre for Psychiatry, for example:

- The Queen Mary and East London NHS Foundation Trust Academic afternoons, occurring monthly and open to Service users.

- Where findings relate to specific patterns of violence or victimisation associated with particular ethnic or cultural groupings, these will be raised with the Synergi collaborative Centre (https://synergicollaborativecentre.co.uk/) for publicisation and debate.

Workstream 3:

This study will lead to papers to be published in journals specialising in general psychiatry, affective disorders and personality disorders. It is important to consider that this is a specialist paper which relates to diagnosis. BPD is categorised as a personality disorder. This study suggests it may be an affective disorder (based on previous analysis using 2000 and 2007 surveys). There must be a careful dissemination plan because, if the preliminary findings are supported, this goes against conventional thinking in psychiatry and psychology.

Workstream 4:

This is a highly technical statistical approach to investigating the distribution of psychotic-like experiences in the household population and any dissemination strategy must take an appropriate approach. Initial papers will be primarily for a specialist audience.

Allowing a year for data analysis and write-up, then journal submission the study team anticipate the outputs will appear between 18 and 24 months (amended to 36 months) from the date of receipt of the data (February 2020).

APMS low numbers and suppression

In order to protect patient confidentiality in publications resulting from analysis of APMS data users must:

• guarantee that any outputs made available to anyone other than those with whom this agreement is made, will meet required standards, including the guarantee, methods and standards contained in the Code of Practice for Official Statistics (http://www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html) and the ONS Statistical Disclosure Control (https://gss.civilservice.gov.uk/statistics/methodology-2/statistical-disclosure-control/) for tables produced from surveys;

• apply methods and standards specified in the Microdata Handling and Security Guide to Good Practice (http://www.data-archive.ac.uk/media/132701/UKDA171-SS-MicrodataHandling.pdf) for disclosure control for statistical outputs

The original benefits remain the same as version 0.4 of this Agreement. Many of the dissemination channels below (e.g. academic conferences) did not run in 2020 or 2021.

Original statement:

The study team are proposing a programme of epidemiological research rather than audit or clinical work so the benefits are not readily quantifiable as deliverables with a specified date. However the anticipated benefits will include:

1. Identification of differential pathways in the development of physical health problems in the presence of mental health problems that will benefit NHS integrated care services specifically, as well as secondary care services for mental disorders.

2. A better understanding of the relationship between violence, mental and physical health and their implications for policies and interventions. This will be of interest to commissioners as well as public health bodies and the police.

Within each workstream there are some specific deliverable benefits:

Workstream 1 (Fall in Violence)

This again depends on the nature of the outputs and whether they support the current hypotheses of the researchers. However, if they do the benefits could be considerable. This is a naturalistic study of a fall in violence. It may indicate new areas of investigation if subgroups can be identified who have shown a marked decline. It is possible it could show subgroups with increase, even though the overall trend is downwards. These subgroups may need targeted interventions.

Workstream 2:

Alcohol has now been robustly shown to be associated with harm to others (Benyon et al., 2019) and the study team have conducted preliminary work using earlier datasets that replicates this finding (Freestone, Igoumenou Coid & Bhui, 2019), and additionally suggests that the link between alcohol use and violence is not affected by personal income, challenging the accepted knowledge that pricing initiatives will reduce alcohol-related violence. However, the study team are seeking to replicate this finding and require a more recent dataset to translate these findings into policy recommendations for government bodies that can help reduce alcohol-related violence in the community.

Workstream 3

The notion that BPD is primary an affective disorder is not new but has been steadily resisted in the mainstream for many years. s.If it were convincingly demonstrated that BPD is primarily a disorder of mood then this would lead to new approaches to intervention for these patients. The key to advancing the message would be a dissemination strategy involving service users – the study team plan to do this by presenting the results collaboratively at conferences that encourage or prioritise service user involvement (e.g. The British and Irish Group for the study of Personality Disorder - https://bigspd.org.uk/ )

Workstream 4:

The findings will need to reach clinicians who see patients with psychotic experiences but who do not present with full picture of clinical psychosis, including early intervention services. The study team will do through targeted dissemination of the results through clinical societies and conferences where such clinicians (as well as patients) are regular attendees, e.g. the International Society for the Psychological and Social approaches to Psychosis, ISPS - http://www.isps.org/).

Workstream 5:

In addition to the current network of collaborators that researchers at the Centre of Psychiatry have ( Kidscape, Muslim Youth helpline, and MIND), other organisations such as ‘Standing together’ (domestic violence charity) will also be targeted. The summaries will assist with the dissemination of the findings from this project and will allow engagement and networking to optimise the impact of this work. This engagement will assist in reaching practitioners, policy makers and those who work with victims and perpetrators of violence.

Successful outputs to date:

1) Completed a data linkage of key variables between the 2000, 2007 and 2014 datasets to support Workstreams 1, 2 and 3. Fully linked datasets (including linkage scripts that can be made openly available to those with the data) allow for a number of powerful analyses and models to be built, including meta-analysis by year to assess population-level trends and the identification of pseudo-cohorts (e.g. a matched population aged 20 in 2000, 27 in 2007 and 34 in 2014 cf. Spiers et al., 2011) to model changes in disease and distress in specific populations over time. This enables more complex heuristic questions to be answered by Health and Social Care researchers.

2) Secured a collaboration with the British and Irish group for the Study of Personality Disorder for dissemination in 2023 of workstream 2 findings. The BiGSPD conference is the largest UK personality disorder conference with over 400 attendees and representation from the civil service, MPs, service users and clinicians as well as researchers. This provides us with a significant platform to disseminate new findings about population-level trends in personality disorder to a mixed audience where vital feedback can be obtained from some hard-to-reach groups, and impact can be maximised.

3) The Applicant has been invited to be an expert member of the Mayor of London's London Vanguard violence reduction project, providing a clear pathway to impact for the findings of the analysis. The Vanguard project is commissioning services from the new Integrated Commissioning Services (ICSs) to reduce violence by and against young people and improve mental health and criminal justice outcomes in the Greater London area. The applicant's presence on the expert advisory group means that findings relating to violence and mental disorder from the research based on the APMS dataset can be fed directly to commissioning groups and service leaders to increase impact.