CPRD Data Linkage Scheme (ODR1819_CRPD2019)

Data has previously been shared with Clinical Practice Research Datalink (CPRD) when the data were controlled and managed by Public Health England (PHE). PHE facilitated data release via its Office of Data Release service (ODR). The ODR was responsible for providing a common governance framework for responding to requests to access PHE data for secondary purposes, including service improvement, surveillance and ethically approved research. All requests to access data were reviewed by the ODR and were subject to strict confidentiality provisions. The responsibility for the management of the National Disease Registration Service (NDRS) of which the National Cancer Registration and Analysis Service (NCRAS) is a part, transferred from PHE to NHS Digital on 1st October 2021.

CPRD is a not-for-profit research data service, jointly funded by the Medicines and Healthcare products Regulatory Agency (MHRA) and the National Institute for Health and Care Research (NIHR), to improve public health by providing observational and interventional research services.

CPRD’s research data services are based on patient-level primary care electronic health records (EHRs) extracted from GP practices that have consented to contribute data under the rigorous regulatory, information governance (IG) and security arrangements outlined below. All patient identifiers are removed at source by the GP practice system providers before releasing limited fields within EHRs to CPRD, and each record is assigned a data subject pseudonym before data flows to CPRD. EHR records sent from GP practice system providers to CPRD are from, from that point, considered de-identified pseudonymised records, which CPRD cannot identify (as only GPs themselves hold the key to reversing pseudonyms).

A central part of CPRD’s observational research services is to provide anonymised primary care data, linked to other health care data sets, for the purpose of health research leading to clinical benefits for the UK. The ability to link primary care data to other datasets is reliant on the presence of data subject pseudonyms being allocated by GP system providers, before collection by CPRD.

This Data Sharing Agreement (DSA) (1) supports the lawful disclosure of patient identifiers for the purpose of linkage with primary care data; and (2) the release of pseudonymised NDRS Data relating to patients represented in both data sources.

Releases of pseudonymised row level data will be for the purpose of conducting Observational Research for public health benefit. Access to the data by CPRD clients, and by internal CPRD researchers, will only be granted by CPRD following approval by the MHRA Independent Scientific Advisory Committee for MHRA Database Research (ISAC). Controls will be placed on the use of the data by the client through a licence agreement which mirrors the requirements of NHS Digital. The client must sign this agreement before any release of linked health data can occur.

Approval to Access Research Data

CPRD clients include academic institutions, life science companies, governmental bodies, and research charities. Linked NDRS Data are only ever made available by CPRD to such clients, for use within research protocols that are approved by ISAC. Following protocol approval by ISAC, the research organisation supporting the Chief Investigator enters into a legal contract with CPRD. The contract covers the terms, conditions and obligations related to access to and use of data for the approved research study.

All researchers are made fully aware they must not use data for any other purposes than the research specified in the approved protocol. An alteration to the agreed research protocol, an extension to the study or new use of data requires additional or new ISAC approval and a new legal arrangement to be put in place between CPRD

Researchers are encouraged to publish their findings and to share these with the regulators when appropriate.

As stated above, CPRD does not receive identifiable patient data (with the exception of gender) as all personal identifiers are removed at source by GP system providers before EHR record level data is provided to CPRD.

Scientific Governance

The MHRA Independent Scientific Advisory Committee for MHRA Database Research (ISAC) is a non-statutory expert advisory body that was established in 2006 by the Secretary of State for Health. Its role is to provide expert advice on Observational Research protocols that seek access to data available through CPRD. When new data linkages are established, CPRD works with the data source and Data Controller to establish suitable arrangements for representation on ISAC, or through an alternative streamlined review mechanism.

Applications may be shared with the Health Research Authority’s (HRA) Confidentiality Advisory Group (CAG) if they include disclosure-risk categories agreed in CPRD’s Section 251 approval (see below).

Contractual control

CPRD licence agreements control the use of data by individual researchers and their host organisations. Licence agreements (i.e. sub-licence and dataset agreement), which are contractually binding in nature, limit the use of data to medical and health research purposes and impose confidentiality controls to protect against any further risk of patient identification. CPRD ensures that the use of data by CPRD and their clients complies with Regulation 2 of The Health Service (Control of Patient Information) Regulations 2002. All research using linked NDRS data requires prior approval by ISAC, on a study-specific protocol basis, and NDRS releases study-specific data for linkage

Section 251 approval

HRA support is provided to enable NHS Digital, as CPRD’s Trusted Third Party (TTP), to receive and process a defined and minimum number of personal identifiers (NHS number, full date of birth, postcode, gender), without breaching the Common Law Duty of Confidentiality. These are securely and directly provided to NHS Digital by participating GP system providers, and by NDRS .

Identifiers are provided under HRA support to link CPRD primary care records to a wide range of secondary datasets relating to the provision of care and public health in England, as set out in CPRD’s Master Dataset List.

All observational studies operate under this HRA approval, termed ‘section 251 support’, as the appropriate legal gateway to enable the processing of identifiable data used for the purpose of linkage by the TTP without breaching patient confidentiality.

CPRD must obtain, and maintain, annually renewed approval from CAG to lawfully undertake linkages by the Trusted Third Party, and residual identifiers that have an agreed research purpose (e.g. Date of Death). CPRD is required to submit amendment requests to CAG for new linkages not covered by its Master Dataset List. A new dataset will only be linked once the CAG approval is obtained.

Ethics approval

The HRA’s East Midlands – Derby Research Ethics Committee (REC) has granted overarching approval to CPRD to collect and use pseudonymised primary care data for purposes of observational public health research studies. Any study that involves an intervention or interaction with patients requires a separate ethical review and approval before any data can be provided.

Agreed fields from primary care EHRs are collected from consenting GP practices and sent to CPRD via a Health and Social Care Network (HSCN) secure connection.

The data are verified for integrity and completeness before further processing.

CPRD uses NHS Digital as its Trusted Third Party for data linkage. It is NHS Digital that undertakes data linkage for CPRD using actual patient identifiers, as permitted under strict legal and ethical permissions and scrutiny provided by the HRA. Once NHS Digital has undertaken this linkage process, it provides pseudonymised patient level linked data back to CPRD, that (in confidentiality terms) is safe for research use. CPRD may then itself safely link pseudonymised datasets together, but only where NHS Digital has previously provided pseuodonymised linked data to CPRD as explained above. This physical and logical separation of the flow of de-identified EHR records from GP system providers to CPRD, from the flow of identifiable patient data to the TTP direct from GP system providers, is a fundamental tenet of CPRD’s governance, ethics and security model.

CPRD operates to high levels of security to ensure that when data is transmitted and or stored it is done so in a way that protects the data. All data in CPRD is stored in a "Tier 3" data centre that is compliant with Government standards to operate in a way that meets the full requirements for managing and storing such important data. The measures are always under review and are subject to audit.

Information Governance

CPRD mitigates the risk of inadvertent disclosure of patient identity through legal agreements preventing the use of the data in conjunction with other data sources, which when linked may potentially re-identify an individual.

CPRD also holds the right to audit data recipients to ensure they are adhering to the terms of data use (including the full terms of the data sharing agreement with the research user), security and confidentiality. On request from NHS Digital, CPRD will provide NHS Digital with the results of any such audit.

CPRD employees are appropriately trained in information governance and data security processes to ensure they have the necessary understanding of relevant laws and standards. Staff are aware that any misuse of data may result in disciplinary procedures and, in the case of a severe breach may lead to dismissal.

Training covering the use of data is mandatory for CPRD staff. All employees responsible for the interaction with contributing GP practices are precluded from access to data. Data are kept on restricted servers and drives accessible only to appropriately trained research staff.

De-sensitising Data

Additional processes are put in place to reduce the likelihood of deductive disclosure of an individual’s identity. Several variables included in the data are made less specific. For example, year of birth is provided rather than exact date of birth and in studies involving children, the month and year of birth are normally provided. Similarly, the geographical information provided by default is at a regional level, and geographical areas with fewer than one million residents are combined to prevent small cell counts.

CPRD has produced a Policy for Managing Anonymisation and the Risk of Identification in Observational Research, which sets outs such requirements which must be met before releasingof data to CPRD clients.

Pseudonymisation

Pseudonymisation is applicable in a number of contexts within CPRD to prevent the identity of the following from being revealed:

• a patient

• an individual recorded in their records (such as clinicians providing clinical

care)

• an organisation (such as hospital trusts or general medical practices)

• geospatial identifiers (such as postcode, or grid reference) which could lead to identification of an individual.

Primary Pseudonymisation

CPRD establishes data subject pseudonymisation through the assignment of a compound pseudonym key. The key comprises a practice identifier and a patient identifier (within that practice). This data subject pseudonym is not identifiable within the data held by the GP.

Multiple Pseudonym Layers

CPRD processes data and makes it available internally to CPRD researchers. In doing so, CPRD replaces the original data source pseudonym(s) with a second layer pseudonym or person ID. This creates multiple layers of separation, such that an adversary would need to translate the CPRD person ID back to the data source pseudonym ID and then gain access to the data source patient index, in order to directly identify a data subject. When linked NCRAS data are supplied to third parties, CPRD replaces the person ID again, by adding a third layer pseudonym ID that establishes a further layer of separation.

Encryption is used for data in transit between secure locations. This applies to both identifier data for linkage and clinical research data. Although the clinical data is pseudonymised, there remains the residual risk of re-identification or the risk of inclusion of disclosive content and data is only intended for processing by authorised recipients. Encryption mitigates this risk and provides assurance. The general default minimum standard for encryption is AES 256 using a complex pass-phrase consisting characters and a mix of upper case, lower case, numeric and special characters

CPRD clients using these linked data will be producing (on an on-going basis) research publications in peer-reviewed journals and presentations in scientific conferences

Studies using NHS Digital data linked to CPRD primary care database are expected and required to demonstrate likely benefits to patients in England which may be via informing better clinical care or public health policies. Some study results may also be used to support regulatory decision making both within and outside the UK, directly affecting the approval or removal of drugs or devices (and hence their availability) or guidance as to their use. Some examples of expected benefits included by researchers in recent RDG applications include:

• Estimates of prevalence, incidence and healthcare burden of specific types of psoriasis in England. This study will further the understanding of the burden of these rare diseases in the UK, particularly highlighting differences between the different presentations of psoriasis. (RDG reference: 21_000421, see: https://www.cprd.com/protocol/prevalence-incidence-and-healthcare-burden-generalised-pustular-psoriasis-palmoplantar

• An understanding of the risks associated with COVID-19 infection in patients with congenital heart disease to inform better clinical management of these patients (RDG reference: 20_000161, see: https://cprd.com/protocol/identifying-clinical-risks-associated-covid-19-patients-congenital-heart-disease-and-0)

• An understanding of potential risks associated with various medications prescribed during pregnancy to both mothers and their children, to inform prescribing decisions (RDG reference: 21_000362, see: https://cprd.com/protocol/maternal-prescriptive-drug-use-risks-and-benefits-mothers-and-neonates)

• An understanding of the different presentations of Long Covid and risk factors associated with developing Long Covid among non-hospitalised COVID patients with the aim of developing supportive interventions and treatments (RDG reference: 21_000423, see: https://cprd.com/protocol/long-covid-non-hospitalised-individuals-symptoms-risk-factors-and-syndromes-0)

• Examining the effects of COVID-19 on primary care management following self-harm in the UK, concluding that despite the challenges experienced by primary healthcare teams during the initial COVID-19 wave, prescribing and consultation patterns following self-harm were broadly similar to pre-pandemic levels (RDG reference 20_001, see: https://cprd.com/protocol/impact-covid-19-primary-care-contact-referrals-follow-care-and-patient-outcomes-after)

• Examples of how research using CPRD data benefits public health: https://cprd.com/examples-how-research-using-cprd-data-benefits-public-health

Further examples and other relevant publications resulting from linked data research which have informed clinical practice or public health policy are presented below. Some of the studies referenced are older (published up to 5 years before) as it can take 4-5 years to translate some research findings into clinical guidance or public health policy.

Pearson-Stuttard J, Cheng YJ, Bennett J et al. Trends in leading causes of hospitalisation of adults with diabetes in England from 2003 to 2018: an epidemiological analysis of linked primary care records. The Lancet Diabetes & Endocrinology, Volume 10, Issue 1, 2022. https://doi.org/10.1016/S2213-8587(21)00288-6 November 2021. The number of people with diabetes in the UK has increased substantially over past decades to almost 4 million with increasing costs to the health service.

This research will inform the provision of health services, management and prevention of diabetes and diabetes-related complications.

Masoli JAH, Delgado J, Pilling L et al. Blood pressure in frail older adults: associations with cardiovascular outcomes and all-cause mortality. Age and Ageing, Volume 49, Issue 5, September 2020, Pages 807–813. https://doi.org/10.1093/ageing/afaa028. In a study of 415,980 people, including those often excluded from studies, researchers reported that there was no increased mortality risk with hypertension in adults above 75 years with moderate to severe frailty and all above 85 years. Research supports the move to raise the blood pressure target for frail older people. https://evidence.nihr.ac.uk/alert/new-research-supports-the-move-to-raise-the-blood-pressure-target-for-frail-older-people/

Sheng-Chia Chung, Reecha Sofat, Dionisio Acosta-Mena, Julie A Taylor, Pier D Lambiase, Juan P Casas, Rui Providencia. Atrial fibrillation epidemiology, disparity and healthcare contacts: a population-wide study of 5.6 million individuals. The Lancet Regional Health - Europe, Volume 7, 2021. https://doi.org/10.1016/j.lanepe.2021.100157. From the paper: The study provides comprehensive evidence for the AF burden on population health and healthcare utilisation. We found approximately two in five AF patients had three or more comorbidities at the time of diagnosis

Data for this study has previously been share when the data were controlled and managed by Public Health England (PHE). As such there are some yielded benefits to be observed from the access to the data for the study prior to NHS Digital becoming data controller. These yielded benefits are noted below;

CPRD publish here https://cprd.com/approved-studies-using-cprd-data a register of all approved studies of which the detail on each study includes a lay and technical summary, the health outcomes to be measured and details on the organisations involved.

There are 3 case studies presented below highlighting how linked CPRD-NHSD data has supported public health research, especially in response to the COVID 19 effort in recent years. Research using linked CPRD data benefits patients in the UK indirectly by contributing to the evidence base for medicine and public health, which in turn informs public health policy, programmes and clinical guidelines.

Case study 1: Higher risks of flu and COVID-19 for cancer survivors (CPRD protocol 20_082).

Older individuals and people with certain health conditions are known to be at higher risk of severe illness if they contract viruses such as flu and COVID-19. This includes people who had certain cancers diagnosed recently and are receiving treatments like chemotherapy. In the UK there are more than two million cancer survivors. To investigate whether people who had cancer some time ago are also at higher risk from flu and COVID-19 a study was carried out using CPRD data. CPRD GOLD was linked to Hospital Episode Statistics Admitted Patient Care (HES APC) database, cancer registrations from the National Cancer Registration and Analysis Service (NCRAS), death registrations from the Office of National Statistics mortality database, and postcode-based index of Multiple Deprivation data. Researchers found that survivors from a wide range of cancers are more likely than people in the general population to be hospitalised or die from flu, even several years after their cancer diagnosis. The raised risks were most likely for blood cancer survivors. Because flu and COVID-19 are both respiratory viruses, this suggested that cancer survivors also have a higher risk of severe COVID-19. The study also showed that cancer survivors were more likely to have other diseases that are associated with increased risk of severe COVID-19, such as heart disease, diabetes, respiratory disease and kidney disease. The findings support the UK policy recommendation to include all blood cancer survivors as one of the priority groups to receive the COVID-19 vaccination. The study findings could also support any work by others to prioritise vaccinations and treatments for longer-term cancer survivors.

Reference 1: Carreira H, Strongman H, Peppa M, McDonald H, dos-Santos-Silva I, Stanway S, Smeeth L, Bhaskaran K. Prevalence of COVID-19-related risk factors and risk of severe influenza outcomes in cancer survivors: a matched cohort study using linked UK electronic health records data. EClinicalMedicine, Volume 29, 100656, December 2020. https://doi.org/10.1016/j.eclinm.2020.100656

https://cprd.com/protocol/covid-19-related-risks-cancer-survivors-matched-cohort-study-using-linked-uk-electronic

Case study 2: Vaccine uptake in pregnancy

Vaccination is an effective way to prevent infectious diseases. However, people with certain social characteristics – such as those living in more deprived areas – may be less likely to receive vaccination. Addressing inequalities in vaccine uptake to prevent infections is a key priority for public health. This cohort study examined the social factors that may be associated with lower uptake of flu vaccine and whooping cough vaccine by pregnant women. It considered a range of social determinants including maternal age, ethnicity, socioeconomic status, number of children in the household and region. The study used linkages between CPRD data, the CPRD GOLD Pregnancy Register, Hospital Episode Statistics (HES) and Office of National Statistics (ONS) small-area-level deprivation data.

The researchers concluded that more targeted campaigns have the potential to reduce vaccine-preventable disease among infants and pregnant women, and to reduce health inequalities. Identifying social factors that are associated with lower vaccination rates could help to design programmes to improve vaccine uptake for specific groups of individuals.

Reference 2: Walker et al. Social determinants of pertussis and influenza vaccine uptake in pregnancy: a national cohort study in England using electronic health records. BMJ Open 2021;11:e046545. https://doi.org/10.1136/bmjopen-2020-046545

https://cprd.com/protocol/social-determinants-uptake-maternal-influenza-and-pertussis-vaccine

Case study 3: Health of mothers of children with a life-limiting condition

More than 86,000 children and young people in England are now living with medical conditions that may ultimately shorten their life and cause death in childhood or young adulthood. Mothers of children with a severe health condition or whose child has died are more likely themselves to die earlier than other mothers.

The lack of studies quantifying the mental health of mothers of children with a life-limiting condition has been highlighted by the National Institute for Health and Care Excellence. In this first part of a larger research programme, researchers used CPRD data to investigate the types of physical and psychological health conditions diagnosed in mothers of children with a life-limiting condition. The CPRD GOLD Pregnancy Register was used to link mothers and their children's healthcare data. The study also used linkages to Hospital Episodes Statistics (HES), Mental Health Minimum Dataset (MHMDS) and Office for National Statistics (ONS) death certificate data.

The study concluded that mothers of children with life-limiting conditions have much higher rates of physical health problems, mental illness, and death. These findings were flagged as an ‘Alert’ for important research by the NIHR. Prior to this study, little research had explored the health of this group of women. Knowing more about the health problems that the women experience could help in the design of specific healthcare interventions.

Reference 3: Fraser et al. Health of mothers of children with a life-limiting condition: a comparative cohort study. Archives of Disease in Childhood 2021;106:987-993. http://dx.doi.org/10.1136/archdischild-2020-320655

https://cprd.com/protocol/life-limiting-conditions-health-children-and-their-mothers

NIHR Evidence - Mothers of children with life-limiting conditions are at risk of serious health problems - Informative and accessible health and care research https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fevidence.nihr.ac.uk%2Falert%2Fchildren-life-limiting-conditions-mothers-more-likely-to die%2F%3Futm_source%3DNIHR%2Bmailing%2Blist%26utm_campaign%3Dc836e7227e-NEWS_RESEARCH_26_8_2021_COPY_01%26utm_medium%3Demail%26utm_term%3D0_570d86f9cb-c836e7227e-33120976&data=04%7C01%7CRhian.Hortin%40mhra.gov.uk%7Ceeaa93de1cb4421aafec08d9b0242f33%7Ce527ea5c62584cd2a27f8bd237ec4c26%7C0%7C0%7C637734491713613980%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=TKPdVHwcavrPA3jVYfUKdqC0ds9O2bgjo6GzYGJAk34%3D&reserved=0