Work Package 4: Data linkage of existing administrative data sources to inform the prototype MDS.

WP4 is led by Arne Wolters and Adam Steventon at the Health Foundation

The NHS collects a huge amount of data, including the electronic medical record in general practice, administrative hospital data, and operational data sets from the emergency services, urgent care and community health. Local authorities (LA) also collect data on packages of social care funded publicly, and needs assessments. This work package will identify, document (create metadata) and link these existing data sources together to establish a basis for the MDS. This will demonstrate what a MDS could look like without the need of additional data collection by care homes.

Led by The Health Foundation (THF) who have experience bringing data together for research and evaluation within its secure data environment this WP focuses on two Integrated Care Systems (ICS) sites who have agreed to participate (Surrey Heartlands & Nottinghamshire) to:

  1. Identify and document sources of data on all care home residents in the two ICS held by health and social care.
  2. Create two linked administrative datasets on care home residents, one for each ICS.

iii. Demonstrate the benefit of routinely linking these data together by providing descriptive analysis of the linked datasets, and share the results with the ICS and care home managers involved.

(i) Identify and document sources of data on care home residents held by health and social care: To test the feasibility of using data routinely held by NHS and Social Care to populate a MDS for improving quality of care as well as for research. The team will:

  • Map out the data sets available in two Integrated Care Service sites, including existing information about the contents and scope of the data (Meta data), existing information governance agreements, and privacy impact assessments.
  • Explore using address information stored in GP records. Address information can be used to identify care home residents that are self-funded, and not included in LA datasets.

Informed by the national findings from the expert consultation group, by input from the care home based PPIE groups and the PPIE panel and WP 1-3 the data sets will be prioritised based on which are likely to produce most value to the minimum data set. It will include hospital administrative data (from the Secondary Uses Service), GP records, local authority social care data, and data from urgent and emergency care and community health.

(ii) Create two linked administrative datasets on care home residents, one for each ICS. Based on the priorities identified above, relevant data sets will be transferred, linked and processed in the secure data environment at THF. This is a complex process that involves agreeing with each of the relevant stakeholders the purposes of the data collection, the way in which patients and the public will be kept informed of the data processing, and the safeguards that will be put in place. All data will be processed in the ISO 27001 accredited secure data environment at the Health Foundation. It will also be ‘pseudonymised’ before the data are transferred. However, three linkage keys will be created: two of these will relate to the individual (one formed by encrypting the NHS number, and another by encrypting a combination of the person’s initials, gender and date of birth, to be used to validate the linkage by NHS number or when the NHS number is not present). The other key will relate to the address of the person involved, formed by encrypting the unique property reference number. We will support the data providers to assign the unique property reference number to their data sets where they are not already present.

The data linkage process will operate in two stages, the first will identify the information required to create the linkage keys required to link the data. Once created, we will extract the relevant information on care home residents from the various administrative data sources. Information on clinical needs will be collected from NHS data (for example GP prescriptions and hospital data), and other information such as mobility assessments from social care datasets.

(iii) Analysis of the linked datasets, and presentation of results with the ICS and care home managers. Using the administrative datasets, we will demonstrate the benefit of routinely linking these data by providing descriptive analysis of these data to stakeholder in the ICS sites and care home managers involved in WP 5. Previous engagement with stake holders demonstrated that descriptive statistics (e.g. a breakdown of the reason of admission to hospital for potentially avoidable admissions at care home level) are useful for care home managers. The specification of the descriptive analysis will be agreed with local stakeholders to ensure relevance and make explicit the data from routine data that could be linked with a care home generated MDS.

Research Partners