NHS Social Care Digital Programme - Black Country & West Birmingham STP

Commissioning

To use pseudonymised data to provide intelligence to support the commissioning of health services. The data (containing both clinical and financial information) is analysed so that health care provision can be planned to support the needs of the population within the Black Country & West Birmingham Sustainability and Transformation Partnership (STP) area.

The CCGs commission services from a range of providers covering a wide array of services. Each of the data flow categories requested supports the commissioned activity of one or more providers.

The following pseudonymised datasets are required to provide intelligence to support commissioning of health services:

- Secondary Uses Service (SUS+)

The pseudonymised data is required to ensure that analysis of health care provision can be completed to support the needs of the health profile of the population within the STP area based on the full analysis of multiple pseudonymised datasets.

Processing for commissioning will be conducted by Pi Limited (trading as Predict X).

Pseudonymised data will also be used to provide Health and Social Care tools that will support Clinical Commissioning Group and Local Authority in improving integrated working and the delivery of integrated health and social care in order to improve outcomes in ways such as those set out in the Better Care Fund (BCF).

Analyses of health and social care activity through population profiling will provide benefits that support care initiatives. It will support identification of areas of improvement, for example reablement, emergency admissions, reduction in length of stay and transfer of care delays. Analysis will assist to: improve integrated health and Social Care; improve outcomes (BCF related); profile the population to support care initiatives; and transfer care delays and reduce length of stay.

The analyses will benefit the local health economies by allowing them to baseline their current health and social care provision. They will provide an understanding of the interfaces between health and social care services and the areas that are most amenable to joint commissioning. Linked data can be used to predict the impact of any planned changes and monitor this once implemented. Understanding the baseline of health and care activities will enable the key partners to provide assurance that they have identified the correct areas and services of focus for integrated working and to evidence improvement as initiatives are implemented.

Health and Social Care Population Profiling

NHS Digital and the Local Government Association are working together to raise the importance of adult social care and support the delivery of person-centred care through digital technology both across Local Authorities and with social care providers.

To this end the Social Care Digital Innovation Programme is being run by NHS Digital in partnership with the Local Government Association and has been developed to provide funding for local authorities to support innovative uses of digital technology in the design and delivery of adult social care.

The work of the Social Care Programme focuses on improving digital maturity and supports the better understanding and use of digital technology across the social care sector.

It is intended to support the health and care sectors to share information securely between different systems and to simplify and standardise the information they collect and use.

There are many links between the health and care system, such as when someone is discharged from hospital into social care, but it's often difficult for health and care professionals to share information about patients and people accessing services.

A range of projects have recently been approved which aim to make transfers of care smoother and safer, improve people’s experience of care, support better care decisions and save care professionals’ time.

There is a need for the Local Authorities to use Pseudonymised outputs from Pi Ltd. This need is due to the reinvestigation of the pseudonymised record level outputs as outputs from Pi Ltd provide further opportunities to investigate the data than were originally outlined. The Data Controllers within this agreement have tasked PI Ltd to carry out specific analysis as outlined within this agreement and share these results with the Local Authorities and CCGs specified. The Local Authorities, having patient level pseudonymised outputs, will be able to investigate these results in order to ask and answer more questions as opposed to only being able to answer the question raised to PI Ltd. The Local Authorities being able to ask and answer questions will improve the reinvestigation into the pseudonymised outputs for future potential service improvement projects.

Depending on the success and the more involved the Local Authorities and CCGs work together, the CCGs may also require pseudonymised outputs from PI Ltd however this will be subject to a future application with NHS Digital. For removal of doubt, the CCGs involved in this agreement will only receive aggregated outputs with small numbers suppressed.

Purpose and approach

The grant funding award to the STP under the Social Care Digital Innovation Programme will be used to demonstrate how predictive analytics and digital information sharing can improve care and support for people needing social care services.

The STP project approach is based on a collaborative working between the Local Authorities Adult Social Care team, the CCGs and Pi Ltd, who have extensive experience in machine learning and predictive analysis.

The Machine learning approach uses the study of algorithms and mathematical models that computer systems use to progressively improve their performance on a specific task. Machine learning algorithms build a mathematical model of sample data, known as "training data", in order to make predictions or decisions without being explicitly programmed to perform the task

Predictive analytics is a form of advanced analytics that uses both new and historical data to forecast activity, behaviour and trends. It involves applying statistical analysis techniques, analytical queries and automated machine learning algorithms to data sets to create predictive models on the likelihood of a particular event happening in the future. Predictive analytics does not tell you what will happen in the future. It forecasts what might happen in the future with an acceptable level of reliability and includes what-if scenarios and risk assessment.

The Data Controllers under this Data Sharing Agreement are;

NHS Wolverhampton CCG

Wolverhampton City Council

NHS Dudley CCG

Dudley Metropolitan Borough Council

NHS Sandwell and West Birmingham CCG

NHS Walsall CCG

PI Ltd will carry out data processing with the local authorities receiving patient level pseudonymised outputs.

Data will be processed under GDPR Article 6(1)(e) and 9(2)(h).

Data must only be used for the purposes stipulated within this Data Sharing Agreement. Any additional disclosure / publication will require further approval from NHS Digital.

Data Processors must only act upon specific instructions from the Data Controller.

Data can only be stored at the addresses listed under storage addresses.

All access to data is managed under Role-Based Access Controls. Users can only access data authorised by their role and the tasks that they are required to undertake.

Patient level data will not be linked other than as specifically detailed within this Data Sharing Agreement. Data released will only be shared with those parties listed and will only be used for the purposes laid out in the application/agreement.

Onward Sharing

NHS Digital reminds all organisations party to this agreement of the need to comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract ie: employees, agents and contractors of the Data Recipient who may have access to that data)

Aggregated reports only with small number suppression can be shared externally as set out within NHS Digital guidance applicable to each data set.

Segregation

Where the Data Processor and/or the Data Controller hold both identifiable and pseudonymised data, the data will be held separately so data cannot be linked.

Where the Data Processor and/or the Data Controller hold identifiable data with opt outs applied and identifiable data with opt outs not applied, the data will be held separately so data cannot be linked.

All access to data is auditable by NHS Digital.

Data Minimisation

Data Minimisation in relation to the data sets listed within section 3 are listed below. This also includes the purpose on which they would be applied -

For the purpose of Commissioning:

• Patients who are normally registered and/or resident within the NHS Dudley CCG, Dudley Metropolitan Borough Council, NHS Sandwell and West Birmingham CCG, NHS Walsall CCG, NHS Wolverhampton CCG or Wolverhampton City Council region (including historical activity where the patient was previously registered or resident in another commissioner).

and/or

• Patients treated by a provider where NHS Dudley CCG, Dudley Metropolitan Borough Council, NHS Sandwell and West Birmingham CCG, NHS Walsall CCG, NHS Wolverhampton CCG or Wolverhampton City Council is the host/co-ordinating commissioner and/or has the primary responsibility for the provider services in the local health economy – this is only for commissioning and relates to both national and local flows.

and/or

• Activity identified by the provider and recorded as such within national systems (such as SUS+) as for the attention of NHS Dudley CCG, Dudley Metropolitan Borough Council, NHS Sandwell and West Birmingham CCG, NHS Walsall CCG, NHS Wolverhampton CCG or Wolverhampton City Council - this is only for commissioning and relates to both national and local flows.

Lima Networks Ltd supply IT infrastructure and are therefore listed as a data processor. They supply support to the system, but do not access data. Therefore, any access to the data held under this agreement would be considered a breach of the agreement. This includes granting of access to the database[s] containing the data.

Equinix do not access data held under this agreement as they only supply the building. Therefore, any access to the data held under this agreement would be considered a breach of the agreement. This includes granting of access to the database[s] containing the data.

Commissioning

The Data Services for Commissioners Regional Office (DSCRO) obtains the following data sets:

1) SUS

Data quality management of data is completed by the DSCRO. The SUS data is then pseudonymised using University of Nottingham open pseudonymiser tool - a standalone windows desktop application which creates a digest of one or more columns of a CSV file, using a shared key (SALT file) controlled by the Data Services for Commissioners Regional Office. The DSCRO then disseminated as follows:

1) Pseudonymised SUS, only is securely transferred from the DSCRO to Midlands and Lancashire CSU secure area.

2) Data quality management of social care data is completed by the Local Authority. The social care data is then pseudonymised using University of Nottingham open pseudonymiser tool. The pseudonymised Social Care Data is then sent to Midlands and Lancashire CSU secure area by secure FTP.

3 Midlands and Lancashire CSU then re tumble the data using the University of Nottingham open pseudonymiser tool with a different applied SALT key for the purpose of this project. The data is then transferred to PI Limited by secure FTP.

The pseudonymisation key cannot be used to re-identify data as the tool does not allow for this to happen, it only allows for one way pseudonymisation.

4) PI Limited then link the data using the re tumbled pseudo link, which is undertaken within a controlled environment by a named member of staff, who then produce online reports using CareTrak data analysis tool to provide the CCG and Local authority with a range of high level commissioning intelligence based on integrated pathways of care.

5) Predictive analytics will be applied which is a form of advanced analytics that uses both new and historical data to forecast activity, behaviour and trends. It involves applying statistical analysis techniques, analytical queries and automated machine learning algorithms to data sets to create predictive models on the likelihood of a particular event happening in the future. Predictive analytics does not tell you what will happen in the future. It forecasts what might happen in the future with an acceptable level of reliability and includes what-if scenarios and risk assessment.

6) PI Ltd send pseudonymised outputs back to Midlands and Lancashire CSU, which then warehouse within a dedicated database and is made available to the local authorities users via RBAC access, these outputs have a different pseudonymisation key applied to what is held by the Local Authority

7) Midlands and Lancashire CSU make available aggregate data reports with small number suppression to the CCGs.

8) Patient level data will not be shared outside of PI Limited, Midlands and Lancashire CSU or the Local Authorities and will only be shared within PI Limited, Midlands and Lancashire CSU and the Local Authority on a need to know basis, as per the purposes stipulated within the Data Sharing Agreement. External aggregated reports only with small number suppression can be shared.

For the avoidance of doubt, there will be no re-identification of individuals, although characteristics that define particular cohorts will be identified.

Health and Social Care Population Profiling

- Supporting identification of areas of improvement including but not limited to:

• Reablement

• Emergency admissions

• Reduction in length of stay

• Transfer of care delays

• Baseline of current health and social care provision for local health economies

• Understanding Interfaces between health and social care services

• Understanding the baseline of health and care activities will enable the key partners to provide assurance that they have identified the correct areas and services of focus for integrated working and to evidence improvement as initiatives are implemented.

• See patient journeys for pathway or service design, re-design and de-commissioning

• Undertake data quality and validation checks.

• Investigate the needs of the population

• Understand health needs of residents who are at risk

• Conduct Health needs Assessments

• The production of joint strategic needs assessments and joint health and well- being strategies.

• Planning and delivering effective health services, public health services and social care services.

At the core of the STP project is the use of pseudonymised health and social care data to develop predictive models which enable the early identification of adults with complex morbidities. This will help to inform service design and the improvement of intervention and prevention programmes.

This programme is designed to support the comments made by James Palmer, Head of the Social Care Programme at NHS Digital who said: “The successful projects span a wide range of areas and give a glimpse into the future of social care.’’

“There is great potential for these projects to be replicated easily to deliver benefits quickly for the system and pave the way for a truly integrated future.’’

‘’The work on predictive analytics is significant given its potential to support people at earlier stages which may help to reduce the need for long-term social care. Through the use of predictive models that forecast service need and target interventions, we have the chance to help people remain independent, in their own homes, for longer.”

Additional benefits include

Health and Social Care Population Profiling

- Supporting identification of areas of improvement including but not limited to:

• Reablement

• Emergency admissions

• Reduction in length of stay

• Transfer of care delays

• Supporting the objectives of the Local Authorities and CCGs collaboration plan.

• Analysis to support full business cases

• Develop Business models

• Learning from and predicting likely patient pathways for certain conditions, in order to influence early interventions and other support for patients.

• Analysis of outcome measures for different treatments, accounting for the patient pathway

• Monitoring of outcome indicators

• Monitoring financial and non-financial validation of activity

• Monitoring of successful delivery of integrated care within the health and care community within the STP.

• Monitoring frequent or multiple attendances to improve early interventions and avoid admissions.

• Support Care Service planning

• Support improved planning to better understand patient flows through the healthcare system, thus allow supporting organisations to design appropriate pathways to improve patient flow and provide commissioners to identify priorities and identify plans to address identified issues.

• Improved quality of services , by providing supportive information to introduce early intervention of appropriate care.

• Improved access to services by identifying which services may be in demand but have poor access, and from this identify areas where improvement is required.

• Better understanding of the health of and the variations in health outcomes within the population to help understand local population characteristics.

• Enables the identification of pressure points in the care and health system

• Provides a geographical understanding of service usage

• Understanding the baseline of health and care activities will enable the key partners to provide assurance that they have identified the correct areas and services of focus for integrated working and to evidence improvement as initiatives are implemented.

• Better understanding of contract requirements, contract execution, and required services for management of existing contracts, and to assist with identification and planning of future contracts

1. A better understanding of pressure points in the existing care system. - Dashboards combining health and social care data show key metrics such as A&E attendances, hospital admissions, hospital discharges, Delayed Transfers of Care (DTOCs) and capacity in care homes. This intelligence has given Wolverhampton City Council a better understanding of how the system can be improved.

2. A machine learning model predicting how many A&E patients end up being admitted to hospital. - City of Wolverhampton Council can use this data to provide best-fitting social care packages for each patient.

3. A new approach to population health involving the creation of care service user profiles to better determine service need in a geographical area. - The work has involved analysing the data of 3,000 users of domiciliary care and showing insight into:

- The services they use.

- Touchpoints they have with organisations in the system

- Socio-Economic data such as indices of deprivation.

This work has generated seven key profiles and unearthed an insight - amongst several others - that there are residents with long-term health conditions who do not access many services, whilst there are other residents with no conditions who access multiple services. The richness of data makes it possible to drill down into this further, investigate and re-organise services to address this.

The team are now looking to use the insight from the profiles and apply them to real-life situations to see whether they can inform the way that services can be delivered.

The commissioning team plan to use the data to better manage the health and care needs of the CCG’s communities to help people stay independent for longer and take pressure off more stretched services.