NHS Digital Data Release Register - reformatted

South London And Maudsley NHS Foundation Trust projects

341 data files in total were disseminated unsafely (information about files used safely is missing for TRE/"system access" projects).


🚩 South London And Maudsley NHS Foundation Trust was sent multiple files from the same dataset, in the same month, both with optouts respected and with optouts ignored. South London And Maudsley NHS Foundation Trust may not have compared the two files, but the identifiers are consistent between datasets, and outside of a good TRE NHS Digital can not know what recipients actually do.

Mental Disorder and Cancer Care: a Data Linkage Study in South London II (ODR1516_358) — DARS-NIC-659293-T1G7M

Type of data: information not disclosed for TRE projects

Opt outs honoured: Anonymised - ICO Code Compliant (Does not include the flow of confidential data)

Legal basis: Health and Social Care Act 2012 – s261(2)(a)

Purposes: No (NHS Trust)

Sensitive: Sensitive

When:DSA runs 2023-09-11 — 2026-09-10

Access method: One-Off

Data-controller type: SOUTH LONDON AND MAUDSLEY NHS FOUNDATION TRUST

Sublicensing allowed: No

Datasets:

  1. NDRS Cancer Registrations
  2. NDRS Linked DIDs
  3. NDRS Linked HES AE
  4. NDRS Linked HES APC
  5. NDRS Linked HES Outpatient
  6. NDRS Systemic Anti-Cancer Therapy Dataset (SACT)

Objectives:

South London and Maudsley NHS Foundation Trust (SLaM) requires access to NHS England data for the purpose of the following research project: “Mental Disorder and Cancer Care: A Data Linkage Study in South London II”.

The following is a summary of the aims of the research project provided by South London and Maudsley NHS Foundation Trust:

The original purpose of this project was to update the previous data linkage between mental healthcare data from SLaM and cancer registry data from the Thames Cancer Registry, thereby adding cases and increasing sample sizes for analysis. The updated linkage also incorporated additional linkage with hospital episode statistics for secondary physical healthcare and mortality data by Public Health England under reference number ODR1516_358. The linkage was completed in 2019 and the methodology and governance for the data linkage is described in detail in the accompanying agreement. This agreement is seeking to retain the data and no new data linkage is proposed.

The aims and objectives of the project are:

1. To compare receipt of screening uptake, referral to secondary healthcare, and timing of cancer treatments following cancer diagnosis between groups with and without mental disorder, taking into account type of cancer and spread at diagnosis.

2. To assemble cohorts of residents in South London (Lambeth, Southwark, Lewisham, and Croydon) receiving cancer diagnoses (i,e. ascertained from NCRS) with and without prior mental disorder diagnoses, supplemented by the linkages to HES, for the analyses described below:

a. To compare cancer incidence between groups.

b. To confirm previous findings from the initial linkage that people with mental disorders are not more likely to have cancer spread (beyond local) at diagnosis but do still have higher post-diagnosis mortality. These analyses will be extended to investigate whether the raised mortality is specific to particular causes of death.

3. To investigate the extent to which differences in intervention receipt for cancer account for differences between cohorts in post-diagnosis survival.

4. To assess the impact of cancer diagnosis/ treatment on mental health outcomes (i.e., hospitalisation duration and readmission, mental health service contact and adherence to mental health treatment) in people with pre-existing mental disorders. For people who have subsequently died after a cancer diagnosis: to compare place of death (home vs. hospital) in people with/without prior mental disorders, and to compare the amount of hospitalisation in the final 3-6 months of life.


The following NHS England data will be accessed:
• Cancer Registration – necessary to identify incidence of cancer diagnosis
• Hospital Episode Statistics Admitted Patient Care (HESAPC), Hospital Episode Statistics Accident & Emergency (HESAE), Hospital Episode Statistics Outpatients (HESOP) – necessary to compare treatment of cancer and or other morbidities between cases and controls
• Diagnostic Imaging Dataset (DID) – necessary to compare treatment of cancer, clinical features and other co-founding variables, e.g. spread of cancer, between cases and controls
• Systemic Anti-Cancer Therapy (SACT) - necessary to compare treatment of cancer between cases and controls


The level of the data will be pseudonymised.


The data will be minimised as follows:
• Limited to data from 01/01/2007;
• Limited to conditions relevant to the study identified by specific ICD or OPCS codes;
- ICD-10 codes: C00-D48
• Limited to the following geographic areas: Lambeth, Southwark, Croydon, and Lewisham, which are the boroughs served by the South London and Maudsley NHS Foundation Trust
• Controls will be limited to 5 patients for each linked case subject, control patients will be conterminous and resident in the areas covered by the SLAM population and same temporality and tumour site coding
The minimisation per use will be reviewed and approved by the CRIS Oversight Committee. The CRIS Oversight Committee carries representation from the SLaM Caldicott Guardian and is chaired by a service user. The CRIS Oversight Committee is responsible for ensuring all research agreements comply with ethical and legal guidelines. It closely reviews, monitors and audits agreements to access CRIS and the analyses subsequently carried out


South London and Maudsley NHS Foundation Trust is the controller as the organisation responsible for ensuring that the data will only be processed for the purpose described above.


The lawful basis for processing personal data under the UK GDPR is: Article 6(1)(e) - processing is necessary for the performance of a task carried out in the public interest

The lawful basis for processing special category data under the UK GDPR is: Article 9(2)(j) - processing is necessary for scientific research purposes
This processing is in the public interest because this study will enhance mental healthcare among cancer patients, whether or not they have existing mental disorder prior to cancer being diagnosed, and benefit the quality of end-life care for people with comorbid cancer and mental disorders.


The funding is provided by the NIHR Maudsley Biomedical Research Centre. Funding is in place until November 2027.
The funder(s) will have no ability to suppress or otherwise limit the publication of findings.


Microsoft Limited provides IT hosting services to South London and Maudsley NHS Foundation Trust and will store the data as contracted by South London and Maudsley NHS Foundation Trust.


No other organisation will access data.


Data will be accessed by researchers affiliated with South London and Maudsley NHS Foundation Trust and Masters students. Any student working with the data held under this Agreement must have completed relevant data protection and confidentiality training and are subject to South London and Maudsley NHS Foundation Trust‘s policies on data protection and confidentiality. Any students accessing the data will do so under the supervision of a substantive employee of South London and Maudsley NHS Foundation Trust. South London and Maudsley NHS Foundation Trust would be responsible and liable for any work carried out by students. These students would only work on the data for the purposes described in this Agreement.

The project has three individuals with honorary contracts, two are from Kings College London and one is from University College London. As the objectives are very broad, these are split into sub-objectives and all three honorary contract holders will be working on the data analysis of a sub-project which sits under the wider objectives of the study. The researchers accessing the data are subject matter experts and additionally there are no resources in SLaM to analyse the data so honorary contracts have been issued. The applicant confirms that they hold signed copies of all of their honorary contracts.


The data linkage was set up and has been developed in response to concerns raised by mental health service users locally and nationally over many years around the well-recognised but persisting physical health inequalities faced by people with mental health disorders. In a strategy prepared for the Maudsley Biomedical Research Centre by service users, physical health outcomes were highlighted as a high priority for research. This has resulted in a broader programme of development, of which the data linkage here is just one component, to assemble data resources in which inequalities in care and outcome might be more systematically investigated.

In line with the National data opt-out policy, opt-outs are not applied because the data is not Confidential Patient Information as defined in section 251(10) and (11) of the National Health Service Act 2006.

Where individuals have opted out of disease registration by the National Disease Registration Service (NDRS), their data has been permanently removed from the registry and therefore will not be disseminated under this Data Sharing Agreement (DSA).

Yielded Benefits:

This valuable and unique linkage between mental healthcare and cancer care data has allowed several studies to be carried out on cancer as experienced by people with mental disorders. One set of findings compares cancer incidence rates with those expected in the source population, following previous reports from other cohorts of lower than expected risk. Another study is investigating differences in stage of cancer at presentation, and several studies underway are comparing cancer care outcomes following the initial diagnosis in people with/without a diagnosed mental disorder. We therefore hope to put together a clearer picture of how mental health may impact on quality of care which we will endeavour to flag to policy makers.

Expected Benefits:

The hope is to put together a clearer picture of how mental health may impact on quality of care which the research team will endeavour to flag the journal papers to policy makers by working with SLaMs BRC communication colleagues who have strong established link to policy makers.

Benefits will be achieved through replicable points of inequality identified and effectively disseminated which in turn will feed into the design and implementation of improved healthcare pathways, likely to involve both cancer care and mental healthcare specialties.


The proposed outputs are relatively simple and easily communicated, namely whether people with mental health disorders have a higher-than-expected cancer incidence, whether they are more likely to present with cancer at a late stage, whether they receive equitable investigation and treatment following a cancer diagnosis, and the extent to which any of these inequalities account for worse outcomes. Some of these have more of a public health relevance (e.g., cancer incidence) and others are more relevant to primary or secondary care (e.g., inequalities in treatment/investigations), therefore communications will need to be appropriately targeted. As mentioned earlier, it is important to bear in mind that some of these questions are broader than just the cancer example – i.e., inequalities in investigation and treatment may well have commonalities across many physical health conditions (e.g., investigations underway into quality of diabetes care). In addition, there is over-arching clarification needed on the sub-populations of mental health service users who are most vulnerable to health inequalities (e.g., by diagnostic group, by level/nature of service contact, by cross-diagnostic symptom profiles).

These proposed outputs represent a step towards realising public benefits – i.e., by identifying and defining points of inequality so that these can be acted on. The strategies required to address inequalities clearly require further research and evaluation. However, at the moment, the paradigm focuses strongly on cardiovascular risk and prevention, whereas there may well need to be wider attention paid to receipt of equitable healthcare following a diagnosis. Preliminary research using an earlier cancer registry linkage, for example, showed that people with severe mental disorders did not have higher cancer incidence or later stage of cancer at presentation; however, their mortality after cancer diagnosis remained significantly higher, suggesting inequitable treatment as an important concern (and hence the continued proposed use of this valuable linked data resource).


The collaboration between NHS (South London and Maudsley) and university (King’s College London) sectors in the NIHR Maudsley Biomedical Research Centre will help with effective dissemination, as there are already extensive and longstanding links to policy makers. Examples include the influence of our early findings of reduced life expectancy in severe mental illness (still the only/primary UK findings for this outcome) on government policy around improving physical health in mental healthcare, as well as the influence of our work on early COVID-19 era mortality in achieving vaccine prioritisation for people with severe mental illness. SLaM also benefit from strong links with cancer specialists and primary care through King’s Health Partners, which have helped with previous dissemination. As well as policy links, personal connections maintained by SLaM/KCL colleagues with major charities and mental health service user groups can also be drawn on as findings emerge. This of course goes above and beyond standard academic routes of dissemination via publication and conference presentation.

Outputs:

The expected outputs of the processing will be submissions to peer reviewed journals throughout the lifetime of the project and through presentations.

The outputs will not contain NHS England data and will only contain aggregated information with small numbers suppressed as appropriate in line with the relevant disclosure rules for the dataset(s) from which the information was derived.

Key findings from this research will be disseminated through peer reviewed journals (published in open-source format) and via presentations. SLaM will work with their Biomedial Research Centre (BRC) who already have extensive and longstanding links to policy makers. Working with BRC communications colleagues will help to maximise these outputs to policy and public dissemination, including publishing CRIS blogs which are short lay overviews of important projects that are taking place using CRIS and highlight specific work taking place available to view on the website. CRIS is a computer system that allows researchers at the BRC to carry out research using information from SLaM which feature on the Maudsley BRC website (https://www.maudsleybrc.nihr.ac.uk/blog/?category=CRIS%20blog).

Initial manuscripts for peer reviewed journals will be prepared and submitted from June 2023. The planned target for initial outputs run into 2024 however outputs (i.e. publications in journals, presentations at academic conferences, etc.) will be ongoing throughout the life-span of the project until 2026.

Processing:

South London and Maudsley NHS Foundation Trust provided a pseudonymisation key to Public Health England 2019. The key was used by the analysts to pseudonymise the data at source. No personal identifiable data was ever supplied to Public Health England.

Public Health England provided the relevant records from the Cancer Registration, all HES datasets, Diagnostic Imaging Dataset and the Systemic Anti-Cancer Therapy datasets to South London and Maudsley NHS Foundation Trust. The data contained no direct identifying data items contained a pseudonymised NHS Number which was used to link the data with other record level data (Clinical Record Interactive Search [CRIS] ) already held by the recipient.

The data will not be transferred to any other location.

The data will be stored on servers at Microsoft Limited.

South London and Maudsley NHS Foundation Trust stores data on the Azure Cloud provided by Microsoft Limited.

The data will be accessed by authorised personnel both remotely and on site at the South London and Maudsley NHS Foundation Trust . The data will remain on the servers at Microsoft Limited at all times.

Personnel are prohibited from downloading or copying data to local devices.

The data will not leave England at any time.

Access is restricted to individuals with a substantive or honorary contract with the South London and Maudsley NHS Foundation Trust who have authorisation from the CRIS Oversight Committee.

Microsoft Limited is not permitted to access the data.

All personnel accessing the data have been appropriately trained in data protection and confidentiality and are required to provide evidence of Information Governance training on an annual basis.

The data will be linked at person record level with the (deidentified) Clinical Record Interactive Search (CRIS) and Thames Cancer Registry (TCR) datasets. CRIS is SLaMs mental health database. A data linkage was set up in 2011 between CRIS and Thames Cancer Registry (TCR) as a pilot attempt for feasibility. This contained TCR data on all cancer cases diagnosed in 1999-2008 from four boroughs of south London (with follow-up till June 2010), linked to CRIS data on those who had received mental healthcare from SLAM starting in 2007.

The pseudonymisation key will be stored in a separate database to the linked dataset used for analysis. All analyses will use the pseudonymised dataset. There will be no requirement and no attempt to reidentify individuals when using the pseudonymised dataset.

Researchers from the South London and Maudsley NHS Foundation Trust will analyse the data for the purposes described above.


MR808 - SLaM IG Clinical Dataset Linking Service — DARS-NIC-292279-Z2S5T

Type of data: information not disclosed for TRE projects

Opt outs honoured: Y, N, Yes - patient objections upheld, No - data flow is not identifiable, Anonymised - ICO Code Compliant, Identifiable, Yes, No (Section 251, Section 251 NHS Act 2006)

Legal basis: Section 251 approval is in place for the flow of identifiable data, Approved researcher accreditation under section 39(4)(i) and 39(5) of the Statistical Registration Service Act 2007 , National Health Service Act 2006 - s251 - 'Control of patient information'. , Health and Social Care Act 2012 – s261(7), Health and Social Care Act 2012 – s261(7); National Health Service Act 2006 - s251 - 'Control of patient information'., Health and Social Care Act 2012 – s261(7); Other-Section 251, Health and Social Care Act 2012 – s261(7); Other-Section 251 HRA approval, Health and Social Care Act 2012 – s261(7)

Purposes: No (NHS Trust)

Sensitive: Sensitive, and Non Sensitive, and Non-Sensitive

When:DSA runs 2018-11-01 — 2021-09-30 2017.09 — 2023.10.

Access method: Ongoing, One-Off

Data-controller type: SOUTH LONDON AND MAUDSLEY NHS FOUNDATION TRUST

Sublicensing allowed: No

Datasets:

  1. MRIS - Cause of Death Report
  2. MRIS - Flagging Current Status Report
  3. Hospital Episode Statistics Admitted Patient Care
  4. Hospital Episode Statistics Outpatients
  5. Hospital Episode Statistics Accident and Emergency
  6. MRIS - Cohort Event Notification Report
  7. Hospital Episode Statistics Critical Care
  8. Demographics
  9. Civil Registration - Deaths
  10. HES-ID to MPS-ID HES Admitted Patient Care
  11. HES-ID to MPS-ID HES Outpatients
  12. Emergency Care Data Set (ECDS)
  13. Hospital Episode Statistics Accident and Emergency (HES A and E)
  14. Hospital Episode Statistics Admitted Patient Care (HES APC)
  15. Hospital Episode Statistics Critical Care (HES Critical Care)
  16. Hospital Episode Statistics Outpatients (HES OP)
  17. Civil Registrations of Death

Objectives:

The South London and Maudsley NHS Foundation (SLaM) Trust provides the widest range of NHS mental health services in the UK. It also includes the National Institute for Health Research (NIHR) Biomedical Research Centre and Dementia Unit which works in partnership with the Institute of Psychiatry, Psychology & Neuroscience at King’s College London. Together they aim to develop more individualised treatments and support advances in the prevention, diagnosis, treatment and care of mental ill health and dementia. To do this, they bring together researchers, clinicians, allied health professionals and service users from across the University/Trust partnership to work together better in order to meet the challenges of finding better treatments and improved care for patients.

They form part of the NIHR infrastructure and one of 11 NIHR Biomedical Research Centres in England. However, they are the only Biomedical Research Centre specialising in the full spectrum of mental health research across the age range from infancy through to older adults. They are one of four centres in the country specifically tasked with meeting the Government's challenge to develop new tests and treatments to improve outcomes for people with dementia.

The Mental Health Biomedical Research Centre and Dementia Unit (BRC/U) has developed the Clinical Record Interactive Search (CRIS) system to provide authorised researchers with regulated access to de-identified information extracted from the South London and Maudsley NHS Foundation Trust (SLaM) electronic clinical records system. CRIS helps researchers study real life situations in large quantities, looking for patterns and trends - e.g. what treatments work for some but do not work for others.

Since 2011, SLaM has periodically linked the cohort to HES and ONS data via the HSCIC. In addition, the HSCIC has supplied HES data for residents within SLaM’s geographic catchment (Lambeth, Southwark, Lewisham and Croydon boroughs) with an indicator variable for those residents with data also present on the SLaM Case Register (i.e. who have received SLaM services), and a pseudonymised identifier allowing linkage of HES and SLaM Case Register data. This data is used to enhance CRIS data so that it can be used for the purpose of identifying health inequalities for patients with mental disorders. The HES and ONS data are not added to CRIS. The data are stored separately in the SLaM Clinical Data Linkage Service (CDLS) and are only accessible to a restricted number of approved technical support staff.

The objective of the data collection is to create a research resource to be used for research projects aiming to investigate physical health outcomes (including mortality) and receipt of health care in people with mental disorders attending secondary mental health care services provided by SLaM.

Projects are subject to individual approval by the CRIS Oversight Committee. The CRIS Oversight Committee carries representation from the SLaM Caldicott Guardian and is chaired by a service user. The CRIS Oversight Committee is responsible for ensuring all research applications comply with ethical and legal guidelines. It closely reviews, monitors and audits applications to access CRIS and the analyses subsequently carried out. Eligible applicants must hold a contractual obligation with SLaM such as an honorary contract or research passport and, if access to ONS data is involved, must have Approved Researcher accreditation. For an application to gain approval it must:

1. Satisfactorily demonstrate underlying value and potential benefits to patient care;
2. Have appropriate supervision and governance - e.g. research governance for research projects; formal clinical governance approval for audits; SLaM director sign-off for service evaluation;;
3. Have an appropriate design to minimise inadvertent risk of inappropriate re-identification of patients - e.g. the likelihood of particularly small cohort / cell sizes (< 10 cases); appearance of high profile publically known/published information, etc. In these cases additional measures may be put in place to safeguard confidentiality;
4. Be within the scope of investigating the associations between specific mental disorders in secondary mental health care (schizophrenia, schizoaffective disorder, bipolar disorder and dementia) and physical illness;

All research projects are carried out within the SLaM NIHR Biomedical Research Centre for Mental Health and the linked data remain within the SLaM NHS firewall at all times (as is the security model requirement for all analyses of SLaM data, regardless of data linkage). Researchers using the data are (and will be) required to have a SLaM substantive or honorary contract, or a research passport.

The honorary contract and research passport encompass a HR agreement between prospective researchers and the South London and Maudsley NHS foundation trust enabling individuals to have access to the linked data for research purposes. Both the honorary contract and research passport ensure that users with access to linked data are contractually obliged to adhere to relevant SLaM Trust policies regarding confidentiality and data protection. Whether a researcher requires an honorary contract or a research passport is dependent on the work that the individual will be doing within SLaM, the attached ‘HR Good Practice Resource Pack’ and ‘Honorary Research Contracts Principles and Legal Requirements’ documents from the NIHR set out the guidelines and requirements for honorary contracts.

Approval is only sought for use of HES data incorporating the SLaM linkage (i.e. not for analysis of HES data alone). Broadly, the studies using the linkage have adopted the following designs:

1. Investigations carried out on HES data from the SLaM catchment, identifying a HES-derived outcome and comparing its occurrence between people with/without a given mental disorder in order to derive standardised morbidity ratios (for example, some current research investigating respiratory disease admissions in people with learning disability compared to the local population).

2. Investigations restricted to people with a given HES-derived outcome and comparing subsequent events between people with/without a given mental disorder (for example, further analyses of people with/without a learning disability who have a respiratory disease admission, comparing duration of hospitalisation and risk of readmission between the two groups).

3. Investigations restricted to people with a given mental disorder investigating one or more HES-derived outcomes in relation to SLaM-derived information (for example, investigating the relationship between mental health symptom profiles and physical health events in people with severe mental illness).

4. Investigations primarily carried out using SLaM data, where HES-derived information is used to provide supplementary information (for example, the ability to adjust for serious physical illness in a number of analyses). This includes the use of mental healthcare data contained on HES for residents in the SLaM catchment to capture mental health service use by providers other than SLaM (e.g. out-of-catchment hospitalisations).

5. Investigations primarily carried out using SLaM data where a HES outcome is used to define the sample (for example, a series of analyses investigating medication and health outcomes before and after childbirth in women with pre-existing severe mental illness).

The approvals also cover ONS mortality data linked to SLaM, which had been first agreed in 2005 with the Office for National Statistics. The use of these data is for analyses of mortality outcomes in people with mental disorder, either within-group (comparing different characteristics as predictors in a survival analysis) or using national data for standardisation.

These studies have focused on describing the higher mortality experienced by people with mental disorders and are now moving into the investigation of specific causes of death, as well as other relevant outcomes (e.g. place of death).

The data will only be used for purposes relating to the provision of healthcare or the promotion of health in line with the requirements of the Health and Social Care Act 2012 as amended by the Care Act 2014.

Yielded Benefits:

Linkages of the South London and Maudsley Case Register to mortality and healthcare data were originally set up because of the recognised substantial disparities in health experienced by people with mental disorders. Although there have been calls for a shift from observational to interventional research in this area, it has also been pointed out that there are still many aspects of the link between mental and physical health which are unclear (Stewart, 2015). To begin with, the scale of the challenge needed delineating. Data derived from our Case Register, accessed via the Clinical Record Interactive Search (CRIS) platform (Stewart et al., 2009; Perera et al., 2016a), were linked to mortality data in order to provide estimates of life years lost in different mental disorder groups: data which remain the only UK source of information on this topic (Chang et al., 2011). Having demonstrated this, a key task has been to identify the disorders responsible and sub-populations most at risk – essential for targeting interventions appropriately, but still relatively under-investigated. Recently published CRIS data linked to ONS mortality data were used to estimate the contributions of different causes of death to life expectancy loss (and hence the life expectancy gains which could potentially be accrued if these were equalised to general population norms); key findings were that a wide range of different causes of death were responsible for the life-expectancy loss, indicating that public health interventions need to focus on factors with multiple health benefits rather than single disorders (Jayatilleke et al., 2017). However, we have used the linkages to investigate and highlight specific pathways, such as suicide (Lopez-Morinigo et al., 2014; 2016), the high mortality experienced by people with substance use disorders who experience transfers of care (Bogdanowicz et al., 2015; 2016), and unexpected deaths in people receiving antipsychotic medication (Mace et al., 2015). A particular advantage of CRIS is the wealth of data provided which can be used to ascertain clinical subgroups with higher or lower risk of mortality (both overall and by cause of death). Investigations to date have included the following predictors: ethnicity (Das-Munshi et al., 2017), clinical risk assessments (Wu et al., 2012), and global clinical/functional profiles (Hayes et al., 2012a; 2012b). Weekend admissions have been evaluated in view of concerns about higher mortality in other specialties, but were not found to predict mortality in mental healthcare (Patel et al., 2016). Antipsychotic polypharmacy exposure has been recently investigated, but was not found to be a significant risk factor in most analyses (PhD thesis – G Kadra), while clozapine use was found to be associated with a markedly reduced risk of mortality (Hayes et al., 2015). Finally, mortality has been evaluated in specific clinical diagnostic groups including personality disorder (Fok et al., 2012; 2014) and chronic fatigue syndrome (Roberts et al., 2016). The CRIS linkage to Hospital Episode Statistics (HES) has been used to investigate health inequalities at the level of hospitalised disorders, including a recently published demonstration of the high risk of respiratory disease admissions in patients with learning difficulties, as well as longer durations of hospitalisation and higher risk of readmission (Chang et al., 2017). Current ongoing investigations have described the most common reasons for hospitalisation in people with severe mental illness (PhD thesis – N Jayatilleke), as well as associations with antipsychotic polypharmacy (PhD thesis – G Kadra) and with symptom profiles (PhD thesis – N Jayatilleke). Finally, HES data have been used to define childbirths, and therefore pregnancy episodes in women with severe mental illness in order to investigate health and use of medication in pregnancy (Taylor et al., 2015; 2016), extending more recently into analyses of obstetric procedures and outcomes at the time of childbirth (PhD thesis – C Taylor). Linkages with hospitalisation and mortality data have been used more specifically in dementia research, given the nature of the disorder as a condition of late-life and thus associated causally or coincidentally with a number of other age-associated comorbidities. This was a recent focus of an All Party Parliamentary Group and we used combined CRIS and HES data to quantify levels of hospitalisation prior to and after clinical diagnoses of dementia – a topic which is being developed further (PhD thesis – U Gungabissoon). Other investigations using HES data have included a study of end-of-life hospitalisation burden in dementia (Sleeman et al., 2017), and a model of costs associated with different severity levels of dementia (Knapp et al., 2016), substantially improving on current data used by NICE for dementia treatment evaluation. Investigations using linked mortality data have included analyses of the accuracy of recording of dementia on death certificates (Perera et al., 2016b), predictors of mortality in delirium (Ward et al., 2015), and analyses of cognitive function (Su et al., 2014) and antipsychotic use (Sultana et al., 2014) as specific predictors. Translation of these findings into health and social care actions is an ongoing process. Part of this has simply involved ensuring that investigations are prioritised with the health inequalities agenda in mind and then maximally disseminated. As a result our demonstration of physical health inequalities faced by people with mental disorders has been influential in shaping government mental health policy with its increasing focus on health improvement. At a local level, the South London and Maudsley Trust was the first mental health service in the UK to adopt a smoke-free policy, largely driven by the investigations using linkages with mortality and hospitalisation data. Specific issues, such as high mortality in patients misusing opiates, have also been important in shaping and targeting clinical practice towards high risk groups. We have maintained high levels of patient involvement throughout, including more recently a group set up to consider and advise on data linkages and their output specifically.

Expected Benefits:

It has been established that people with most mental disorders, including neurodegenerative conditions, have substantially worse physical health outcomes (for example, 10-15 years lower life expectancy in analyses of SLaM data, with much of this life expectancy loss accounted for by mortality associated with physical disorders). However, relatively little is known about the health conditions underlying these inequalities, although this knowledge is clearly important in order to develop interventions to improve the situation.

The over-arching objective of this research programme is to provide information that will assist in narrowing the mortality and physical morbidity disadvantage experienced by people with mental disorders. Improvement in the physical health of people with mental disorders is highlighted regularly in Government policy (e.g. ‘Closing the gap: priorities for essential change in mental health’, 2014) and the monitoring of physical health outcomes is increasingly becoming a metric for mental health Trusts, as well as for national structures such as the PHE Mental Health Intelligence Network. The SLaM-KCL collaboration using CRIS and associated linked data has led the field in informing and influencing such policy, for example generating what are to date the only UK data on life expectancy in mental disorder (PLoS One 2011;6:e19590) and providing the Department of Health with data on premature mortality rates in mental disorders (J Campion – personal communication).

At a local level, these findings have also been more specifically influential in driving the implementation of a smoke-free policy across SLaM estates. As the largest centre for mental health research in Europe, SLaM are well-placed to ensure that findings from this project are effectively disseminated and will be nationally influential.

The research SLaM enables will provide novel and important information to inform these policy initiatives. Physical health disadvantages are likely to cross multiple disorders and multiple levels of morbidity: from mortality to non-fatal conditions, and from the individual impact of serious health conditions to the wider economic impacts of increased secondary care use, longer hospitalisations, and increased risk of readmission. There is therefore a need for a coordinated series of analyses to inform on specific areas of inequality in order to target interventions to improve health.

In order to improve morbidity and mortality through health and social care interventions, it is important both to have information on the adverse outcomes potentially underlying disadvantages and to be able to characterise groups most at risk of these outcomes.

The HES and mortality outcomes will address the first information need, through the proposed analyses investigating the most common reasons for acute hospitalisation in people with mental disorders, and their relative risk in relation to the local population – not only for the hospitalisation itself, but also for adverse outcomes following hospitalisation such as longer duration of hospitalisation, lower intervention receipt (where intervention coding is available – e.g. for surgical procedures), and higher risk of readmissions with the physical condition in question or a recognised complication, as well as mortality for different causes of death.

CRIS (mental healthcare) data will in turn be used to address the second information need – i.e. allowing the definition of mental health characteristics of people most at risk of adverse physical health outcomes. For example, SLaM have already demonstrated that all-cause mortality is more strongly predicted by functional impairment than general symptom severity in severe mental illness (Journal of Psychosomatic Research 2012;72:114-9, PLoS One 2012;7:e44613) and more by clinician-appraised risk of self-neglect than by appraised risk of suicide or violence (Psychological Medicine 2012;42:1581-90). This is important because mental healthcare priorities (on symptom improvement and risk of suicide/violence) have therefore not been optimally focused for mortality prevention and there has consequently been a shift in emphasis towards wider health promotion.

In relation to clarifying populations at risk, rapid advances in text-mining and their implementation in CRIS now allow detailed information to be gathered for analyses not only on mental disorder diagnoses, demographic factors and service contacts (i.e. what might be available administrative data in any other Mental Health Trust), but also on pharmacotherapeutic and psychotherapeutic interventions, detailed symptom profiles, risk lifestyles (e.g. smoking, illicit drug use), and adverse drug events. For example, ‘negative’ symptoms of schizophrenia have been ascertained through text-mining, have been demonstrated to predict worse mental healthcare outcomes (BMJ Open 2015;5:e007619) and are currently being investigated along with 45 psychotic and 15 depressive symptoms as predictors of adverse physical health outcomes, including mortality. Symptom profiles are recognised to be important predictors of psychosis outcomes and are the primary focus for mental healthcare interventions; however, they are ‘invisible’ in routine healthcare data because they are recorded in text rather than structured fields. The proposed analyses against HES and mortality outcomes are thus only possible at the moment (internationally as well as in the UK) in the CRIS resource at SLaM – hence they are uniquely positioned to provide influential investigations of physical health outcomes in mental disorders at a level of detail which would not be possible through any other route.

The information on physical health outcomes from HES and mortality will allow policies and interventions to be developed and evaluated within acute and primary care services to serve better people with mental disorders, while information from the mental health record is particularly relevant to mental health services in clarifying patient groups who are most vulnerable. Dissemination targets are therefore broad and cut across primary, acute and mental health care sectors, as well as involving multiple levels from clinicians and Trusts delivering care to commissioners and NHS structures overseeing this, as well as bodies such as PHE with broader health improvement oversight. Furthermore, audiences differ by age-ranges and disorders of interest (e.g. those interested in physical disorders in young adults with severe mental illness may be different from those interested in the acute care impact of dementia in later life). Therefore, in addition to the proposed programme of academic publication and dissemination, SLaM will prepare a 3-yearly report summarising findings over the previous 3 years and their implications.

Outputs:

The primary output of the linkage is the production and maintenance of a research resource for the purpose of use in informative research analyses for publication in peer-reviewed journals and other standard routes of academic dissemination (e.g. conference presentations).

All secondary outputs (whether tables or visuals) will only include aggregated data and will not include any description of a cell size below 10.

In terms of target dates, SLaM expects that a minimum of five research papers would be published per year using the proposed data linkages. Examples of papers recently published or planned for publication include:

1. A study investigating respiratory disease admissions in people with learning disabilities receiving SLaM services. This has shown clear disadvantage in terms of risk of readmission and length of stay, and is important for service planning for learning difficulties populations. Currently submitted for publication.

2. A current BRC PhD studentship using CRIS to investigate medication and medication changes during pregnancy in women with severe mental disorders (e.g. schizophrenia and bipolar disorder), and mental health outcomes before and after childbirth. This is one of the world’s largest investigations of this question and highly relevant for services caring for women with severe mental disorders and the treatment decisions required around pregnancy. Two papers have been published (Taylor et al. BMC Psychiatry 2015; 15: 88. Taylor et al. Archives of Women’s Mental Health 2016 May 13. [Epub ahead of print]) and a further three have been submitted or are in preparation.

3. A study of care home and hospitalisation costs associated with levels of cognitive function in people with dementia, carried out in order to inform NICE decisions about cost-benefits of dementia treatments – the largest and most generalisable study of this issue to date. Currently submitted for publication.

4. A current BRC PhD studentship study of medication profiles in people with severe mental disorders and physical health outcomes associated with these. The focus so far has been on antipsychotic polypharmacy – an important issue in routine clinical care but one for which there has been little or no evidence base to date. Two papers have been published (Kadra et al. BMC Psychiatry 2015; 15: 166. Kadra et al. Schizophrenia Bulletin 2016 Apr 15. [Epub ahead of print]) and a further two papers are in preparation. In addition, recent findings of a protective association of clozapine with mortality, despite multiple adjustments and applying both to natural and external-cause deaths, has been influential and very highly cited (Hayes et al. Schizophrenia Bulletin 2015; 41: 644-655).

5. A current BRC PhD studentship investigating symptom profiles in people with severe mental disorders and their associations with cardiovascular disease admissions. Three papers are currently in submission or preparation, the aim being to investigate the profiles of people with mental disorders who are most at risk of adverse physical health outcomes.

6. A study describing acute sector hospitalisations in people with eating disorder diagnoses. Analyses have been completed, showing substantial increased risk of a range of adverse physical health outcomes and the paper is currently in preparation.

7. A study of describing stroke incidence and its predictors in people with dementia. The paper is currently in submission.

8. Several studies investigating predictors of suicide as a specific cause of death. The E-Host-IT study (funded by an Academy of Medical Sciences Fellowship) investigates fine-grain text predictors of suicide risk. The Pheme consortium (funded by EU FP7) seeks to investigate suicide risk temporally associated with phenomena on social media. In addition a Mental Health Research UK funded PhD studentship is investigating antidepressant profiles in relation to suicide and suicide-related behaviour.

9. The range of publications to date on mental disorders and mortality risk have been influential in shaping national mental health policy (our findings on reduced life expectancy remain the only UK data on this to date), promoting physical healthcare in these populations, as well as on local policy (e.g. SLaM has been the first mental healthcare provider to adopt a smoke-free policy in all its units).

10. Data on general hospital use before and after a dementia diagnosis, using the CRIS-HES linkage, has been cited in a recent All-Party Parliamentary Group report on comorbidity in dementia.

For a full list of CRIS publications including those that have used linked data please see the Maudsley BRC website at the following address: http://www.maudsleybrc.nihr.ac.uk/about-us/core-facilities/clinical-record-interactive-search-cris/cris-publications/.

The utility of the linkages with HES and mortality data is to allow investigations of physical health outcomes in people with mental disorders, as the linked databases primarily contribute this information. Most of the work using the linked information is best classified as ‘research’. However, where an investigation is primarily evaluating these outcomes in a single service with a view to evaluating its performance with a view to improving this, it is likely to be more appropriately categorised as clinical audit. Examples of previous clinical audit projects which have used such linked data include: i) surgical outcomes in people with severe mental illness; ii) deaths occurring in patients receiving care from Trust Addictions services; iii) physical healthcare of substance misusing patients in Lambeth. Linked data have also been used to supplement outcomes in studies best categorised as ‘service development/evaluation’, for example a monitoring and evaluation workstream of an adult mental health programme, and of a home treatment team intervention in Croydon. These studies do not differ in their nature from research studies using the linked data; it is simply that their focus is on specific service evaluation and the relevance of their findings is generally for the Trust rather than the research community. Linked data are primarily of use for baseline or single-stage audits, because SLaM do not have a ‘live’ data feed from these linkages for real-time evaluation of interventions or repeat audits.

All the potential uses of the linked data fall within the stated primary purpose of investigating physical health in people with mental disorders. The only additional request is to be permitted to use the linked data on occasions to ascertain mental healthcare received outside the SLaM catchment. This is particularly important in longitudinal studies of SLaM’s patient cohorts where admissions for relapses in mental disorders of interest may occur in different inpatient units (particularly within London) and is therefore an example of the use of linked data outside the physical health outcome category.

The data being requested will only be used for the purpose described. Any proposed changes will be submitted to the HSCIC for approval before implementation.

Publication targets will clearly depend on the nature of individual findings and the potential audience envisaged. Where possible, SLaM will target general medical and/or public health journals with a broad audience, because analyses are likely to cross disciplines; however, they will also consider specialist journals within the mental health field as well as the individual medical specialties implicated. Dissemination at national and international conferences will adopt a similar strategy of aiming for as broad as possible a reach. They will include mental health focused meetings such as the Royal College of Psychiatrists and European Psychiatric Association congresses, and psychiatric epidemiology meetings such as the International Federation of Psychiatric Epidemiology (IFPE) but SLaM will also seek presentations at medical specialty conferences where results have relevance to those audiences, as well as meetings where commissioners are likely to be represented.

For each application received, the CRIS Oversight Committee, considers the study design and advises on optimisation of benefits. The CRIS Oversight Committee also has a responsibility for publicity and dissemination of findings to relevant parties, media and patient groups.

For example, a major programme of work has involved describing the physical health needs and inequalities faced by people with severe mental disorders. As well as featuring in regular patient/public focused dissemination events on CRIS findings, periodic blogs and editorial pieces have been written summarising the CRIS-derived evidence as it emerges, so that an up-to-date picture is maintained. This work has also been influential at a national level in shaping government policy on physical health needs in mental healthcare and several policy-focused reports have been prepared.

In addition to these traditional routes of academic dissemination, SLaM will seek internal funding to support the production and dissemination of a 3-yearly report of all work on this project, summarising the key findings and their health/social care implications. The report will be made available online with appropriate supporting resources, and will have an executive summary of key points raised.

Processing:

Excluding patients that opted out of participation, SLaM extracts patient identifiers for all registered users of SLaM’s facilities, SLaM extracts patient identifiers from SLaM’s electronic patient record, as well as the CRIS pseudonym (known as the BRCID) from the CRIS system. The identifiers are securely transferred to the HSCIC and the HSCIC returns HES and ONS Mortality data linked to the CRIS BRCID with all patient identifiers removed other than Date of Death. Additionally the HSCIC provides a bespoke HES extract of all residents in SLaM’s geographic catchment (Lambeth, Southwark, Lewisham and Croydon boroughs). All supplied HES and Mortality data are held separately by the SLaM Clinical Data Linkage Service (CDLS) and are only accessible to a restricted number of approved technical support staff.

Researchers do not have access to the raw HES or ONS data.

The HES and ONS data will not be linked with patient identifiers from SLaM’s electronic patient record and no attempt will be made to identify individuals in the data under any circumstances.

When an application has been approved by the CRIS Oversight Committee, technical staff – all of whom are substantive employees of SLaM – assemble bespoke de-identified linked databases meeting the approved requirements of the research study. These are deposited in shared network drives within the SLaM network. Approved researchers can only access the data on location within the SLaM network. All research databases remain within the SLaM firewall at all times on the SLaM network. A dedicated office suite, the BRC Nucleus, has been set up at the SLaM Biomedical Research Centre in order to facilitate analyses using SLaM data. Removal of data from this environment is expressly forbidden other than in the form of aggregated summary data with small numbers suppressed in line with the HES Analysis Guide.

For each research database created a different encoded identifier variable (anonym) is assigned meaning there are no common identifiers or pseudo-IDs across different databases making it impossible for researchers to link their database with source SLaM or HES data. This uses a one-way encryption method following which anonyms cannot be reverse engineered.

At the completion of research projects, the databases used are removed from the shared network drive and archived for a period of 10 years and then permanently destroyed.