Evaluating protocols for identifying and managing patients with FH

The University of Nottingham is conducting a study to evaluate protocols for identifying and managing patients with Familial Hypercholesterolaemia (FH), an inherited condition that means their cholesterol levels are higher than normal from birth. Latest available pseudonymised mortality data for the Simon-Broome registry cohort is requested to support this study.

The Simon-Broome Registry was established in 1980 as a register of consented patients with familial hypercholesterolaemia (FH), in order to provide a resource for research with the aim of furthering more effective diagnosis and treatment and preventing early heart disease. Familial hypercholesterolaemia (FH) is an inherited condition that means individual's cholesterol levels are higher than normal from birth. Initially the Register was co-ordinated by the University of Oxford, but subsequently transferred to UCL, who are the data controller for the Register (but is only acting in an advisory role for the FH study). UCL will continue to hold the Simon Broome Registry (under instruction of the joint data controllers) to offer the opportunity for the Simon Broome Registry committee to compare aggregated results to previous analyses completed by the committee on the ONS linked dataset.

This data sharing agreement has been merged with another agreement (NIC-115405-P6X6Q-v0.11) where Hospital Episode Statistics (HES) Admitted Patient Care, Outpatients and Accident & Emergency datasets were linked to the same cohort and disseminated for the same purpose. Due to the merging agreements, UCL are permitted to retain the data disseminated under NIC-5405-P6X6Q until such time that NIC-300282-G9Q0Q expires. If the agreement is not extended or renewed, all data disseminated under NIC-115405-P6X6Q and NIC-300282-G9Q0Q will be required to be securely destroyed in line with NHS Digital data destruction processes.

The University of Nottingham and University of York are joint Data Controllers who also process data for this project. University College London is a data processor for this project.

The justification of processing this data under the principles of GDPR, is Article 6(1)(e): ‘Processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller’ as this is a task within the public interest and as the research involves health data, Article 9(2)(j) is also applicable, as this details that processing is necessary for scientific and research purposes, subject to appropriate safeguards.

Familial Hypercholesterolaemia (FH) is a common inherited cause of raised cholesterol, affecting up to 320,000 people in the UK. However, over 80% of individuals are not identified, leading to many avoidable heart attacks and early deaths. Use of cholesterol lowering medication can prevent over half of these premature heart attacks.

National (NICE) guidelines recommend general practitioners (GPs) identify people with possible FH by using cholesterol levels, family information and physical examination. If FH seems likely, the GP should refer patients to a specialist to confirm diagnosis, usually by genetic testing. If FH is confirmed, treatment should be given and their relatives contacted for genetic testing - known as “cascade” testing.

Implementation of NICE recommendations for cascade testing FH has been very limited to date. Existing cost-effectiveness analyses of FH identification protocols (UK and internationally) have focused on whether or not specific protocols for cascading are cost-effective. Commissioners and policy makers are uncertain about whether current designs for cascading programmes represent the best value for money. In particular, they have questioned whether tighter criteria for cascading could offer better value for money, and whether current service designs offer value for money in terms of maximising the number of relatives tested. This reflects recognition amongst commissioners and policymakers that previous cost-effectiveness results were subject to significant limitations and evidential uncertainties.

The Department of Health, through NICE, recognised this gap leading to a scoping review by the National Institute for Health Research (NIHR) Health Technology Assessment (HTA). As a result NIHR HTA commissioned this current NIHR project to address these limitations by comparing a wide range of protocols for cascade testing and addressing a wide range of evidential uncertainties and develop economic models to evaluate the cost-effectiveness of cascade testing in the NHS with the aim of informing the implementation of the most cost-effective and acceptable protocol of care for these patients. This project relies on using existing cohorts and registries, routine NHS data and secondary care service data to inform the parameterisation and structure of the models.

The study team propose in this programme of research to evaluate treatment patterns and short- and long-term cardiovascular outcomes and the NHS costs of patients with FH. The outputs of this linkage request will result in providing the most accurate and up-to-date outcome of FH patients to date. Data on cases managed in primary and secondary care will be linked to data on treatment patterns and cholesterol response. This will provide a rich and accurate source of information to parameterise and inform the structure of the short and long-term economic models, the main deliverable within the HTA programme of work.

In order to do this, the study required pseudonymised HES Accident & Emergency, Outpatients, Admitted Patient Care datasets and mortality data. The study requested multiple years of data to make sure it can model the full life-course of an FH patient, to include both their outcomes across follow-up and health care utilisation. Only then, can the study address all the current gaps in the evidence and capture all events in secondary care as well as conduct economic costing on a patient's health care utilisation over an extended period of time. This means this source of data is essential in developing the most robust economic model outcomes to inform commissioners and policy-makers on the impacts of various cascade screening strategies as accurately as possible. The team are requesting latest available pseudonymised mortality data to ensure all latest deaths are captured in the analysis.

Another area of uncertainty is the costs of the cardiovascular events, and more generally, the costs of FH patients to the NHS. To address this, the HES and mortality data is used by the study team to calculate the costs of the cardiovascular events. HES covers in hospital diagnoses and mortality but fails to capture out of hospital mortality rates. Without the mortality data, the models would significantly under-estimate the true costs and burden of the disease. The costs of the cardiovascular events will be used to inform the cost-effectiveness model that predicts the long-term costs of patients with FH and without treatment. The team are requesting latest available mortality data to ensure all latest deaths are captured in the analysis.

Below are all the stakeholders that are involved with the project, with detail as to their activities for the purpose of this project. There is no work in this programme of research taking place outside the UK. All processing activities are within bona-fide UK academic institutions which abide by the highest levels of Data Protection laws and Information Governance rules. No data or specific tools/analysis involving usage of data will be shared with any third parties. Only publication output on the aggregate results of the analysis will be widely disseminated in peer-reviewed journals, conferences, reports, and official HTA monograph.

University of Nottingham: This is the lead organisation carrying out the research. Their role is as a data controller who also processes data, given the remit for decision making and aims of the project and as they will receive the pseudonymised dataset from NHS Digital and will conduct the analysis from the linked data (Simon-Broome Registry to HES and Simon-Broome Registry to mortality).

University of York: This is a collaborating institution carrying out the research. Their role is as joint data controller who also process data, given their involvement in the aims and remit of the project and as they will receive the pseudonymised dataset from the University of Nottingham and will conduct the analysis from the linked data (Simon-Broome to HES and Simon-Broome Registry to mortality).

University College London (UCL): This is a collaborating institution. They hold the consented Simon-Broome registry dataset. They are a data processor as they continue to hold the mortality data disseminated under a previous iteration of this agreement (mortality data up to December 2016). UCL originally submitted the cohort patient identifiers to NHS Digital for linkage to mortality dataset and followed the same process for the previously disseminated HES datasets. They will not receive or have access to the requested mortality data under this agreement or the previously disseminated HES datasets.

Co-Investigators to the study from University of Southampton, University Hospital Southampton NHS Foundation Trust, University of Aberdeen, Cardiff University and Hampshire County Council are acting in an advisory role only, will not have access to NHS Digital data under this agreement and they play no role in making decisions about the means by which the personal data are being processed and have no influence on the outputs being produced.

HeartUK: This is a charity involved in promoting the Simon-Broome registry database and follow-up studies using the cohort. The steering group at HeartUK oversees and approves access to use the Simon-Broome registry for research in the public interest. They are not a Data Controller or Data Processor as they will not process, control, or store any of the data and not involved with the aims the research. It provides the forum (I.e. meeting space in London) for hosting the Simon Broome steering group (members are UK academics).

British Heart Foundation (BHF): The BHF support patient involvement and will support dissemination of the aggregated results and publications. They are not a Data Controller or Data Processor. They will only see aggregated data with small numbers suppressed (in line with the HES Analysis Guide) in the final report once published, and any publications and dissemination via conference presentations, posters, etc .

National Institute for Health Research – Health Technology Assessment Programme (NIHR-HTA): NIHR-HTA is UK government organisation under the Department of Health which funds research about clinical and cost-effectiveness of healthcare treatments and tests for those who plan, provide or receive care from the National Health Service or social care services. NIHR-HTA programme as specifically commissioned this study to be conducted. The funders have no role in the study rationale, processing decisions, study design, study results, planning, or dissemination of study results.

A Confidential Advisory Group (CAG) section 251 amendment has been approved to enable the same process for submission of the cohort patient identifiers by UCL to NHS Digital to provide the up to date mortality data. However, NHS Digital already hold the cohort patient identifiers so no patient identifiers will be flowing in to NHS Digital for this version of the agreement. NHS Digital will provide the data to the approved recipient at University of Nottingham via their secure portal and approved data transfer process who will then flow the disseminated data to University of York as detailed in this agreement.

The justification of data usage and analysis is that the purpose of the usage of NHS digital data (linked Simon-Broome to mortality) involves analysis to estimate the long-term rates of cardiovascular outcomes amongst patients with FH. One significant area of uncertainty is the frequency and timing of serious cardiovascular events in FH patients, and how these differ between undiagnosed and diagnosed individuals. Cost-effectiveness models to date have relied on general population data on cardiovascular events, crudely adjusted to reflect broad prognostic differences between the general population and FH patients. To address this the team therefore plan to use the Simon-Broome registry and HES data and mortality data in order to understand the clinical events experienced by diagnosed patients with FH.

Due to the longevity of the 3,553 confirmed FH patients in the Simon-Broome disease who were registered from the 1 Jan 1980 onwards, a current linkage to mortality data would allow for up to a potential 38 year follow-up period which uniquely positions this registry as the longest prospective FH study globally. Having such long-term outcomes will therefore allow for extremely robust extrapolation of cardiovascular mortality delivered within the economic models.

The Simon-Broome register is a register of consenting patients held at University College London within their Data Safehaven. The original Simon-Broome data set includes identifiable variables which includes forename, surname, date of birth, NHS number, sex, unique study identifier.

The data flow will be:

1. NHS Digital already hold the cohort patient identifiers so no patient identifiers will be flowing in to NHS Digital for this version of the agreement.

2. NHS Digital will link the cohort to the Civil registration - Deaths dataset (mortality). The pseudonymised mortality dataset along with the unique study identifier will be returned by NHS digital to University of Nottingham. No further HES data will disseminated under this agreement.

3. University of Nottingham will securely share the pseudonymised mortality data with University of York and data aggregated with small numbers suppressed in line with the HES analysis guide with UCL.

4. Analysis will take place by the HTA research team (substantively employed staff members of University of Nottingham and University of York under this agreement). The mortality data will be linked to the Simon-Broome disease registry which has already been linked to HES datasets via the study ID. All linked analysis datasets are pseudo-anonymised and no identifiable information on the patients will be retained.

Data from the linked dataset of diagnosed FH patients will be analysed using a form of survival analysis called multi-state modelling. Parametric survival models will be used to facilitate extrapolation of event rates over the full lifetime of the FH patients. Estimates of event rates will be conditioned upon patients baseline demographic and clinical characteristics using multivariate parametric survival regressions. Important prognostic characteristics will be included in the analysis including a covariate to capture whether patients were enrolled within the pre- or post-statin era (based upon the previously used cut-off for statin use of 1st January 1992).

Data on prognostic characteristics will be obtained from the Simon-Broome dataset. Date of death and cause of death by ICD-10 codes will allow for censoring of patients in survival analyses as well as ascertainment of cardiovascular mortality. Clinical outcomes will be selected for modelling based on their importance as predictors of worsened prognosis, quality of life and/or costs. These are likely to include: revascularisation, myocardial infarction, stroke and mortality. The final selection of outcomes will be approved by the study team and external advisors to the project.

The lawful basis under the Common Law Duty of Confidentiality for the data flows for this study is CAG support through an approved amendment to Section 251 (ref: 18/CAG/0007). Ethics and HRA approvals are also in place (refs: 17/EM/0209 and IRAS 214219).

All HTA research team members involved with accessing, processing, and analysis of the data are substantively employed staff members by their respective institutions (University of Nottingham and University of York).

There is no data access or processing activities occurring outside of the UK.

University of Nottingham and University of York are not permitted to re-identify individuals under this agreement.

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).

The main result (outputs) will be the parametric survival regressions for economic modelling and the cost of hospitalisations. Whilst individual patient level data is required to assist with populating the model, the data used in the modelling is not identifiable as the identifiers are removed by NHS Digital prior to delivery to the Universities of Nottingham, and subsequent transfer to University of York. The data will be used to generate predictions of the proportion of individuals experiencing relevant clinical events over time within the economic model and their costs. This will allow estimates of lifetime costs and QALYs (quality-adjusted life years) to be generated by the model. The models, along with a full representation of their uncertainty (derived using the variance-covariance matrix for each event) will be incorporated within the model.

As well as completing NIHR HTA monograph, standard dissemination strategies, of peer-reviewed open access primary care, public health, cardiovascular and genetic publications and conference presentations, will be further prioritised. No personal data will not be used in the write-up or publication of results. Descriptive statistics will only be at the aggregate level with small numbers suppressed in line with HES analysis guide.

The audience of the research outputs are patients, the public, charity groups, policymakers at the Department of Health, Public Health England, NICE. These outputs (publications, reports, monographs) will be made open access (freely available) to all audience members. No outputs will be used for any commercial purposes.

The project will ensure that findings are highlighted to the public, patients, and policymakers (NHS, Department of Health, Public Health England and NICE) in partnership with third sector voluntary organisations (Heart UK and British Heart Foundation - BHF). The project has bi-annual meetings with HEART UK and BHF to keep them updated on the study progress and any new potential findings, and having early conversations with Department of Health and Public Health England on a dissemination strategy. This will likely involve a published NIHR Signal in 2021. In the UK, the data controllers and processors are active current contributors to NICE guidelines for FH and will ensure the outputs of this cost-effectiveness analysis inform updated NICE guideline recommendations and quality standards.

The project will also work with the International FH foundation, US CDC Office of Public Health Genomics and European collaborators, and international advisers in Australia, USA and Europe to ensure the analysis informs international guidelines recognising potential differences between international settings. These discussions will occur in 2020 when the project has economic models developed.

All outputs and publications contain only aggregated data with small numbers suppressed in line with the HES Analysis Guide.

The project will support Heart UK to organise patient and health professional focus groups to help refine findings and produce digital dissemination strategies (e.g. involving website, email, Facebook, Twitter or similar). This will occur towards end of the study in 2021. The project is running some 2-3 consensus development groups with HEART UK/BHF in 2020 to interpret and analyse the findings from a patient and stakeholder prospective. Working closely with HEART UK and BHF, the project will develop a disseminate study from the study to the wider public by:

- Networking with existing and potential FH service commissioners across England, e.g. national NHS commissioners workshop

- Presenting study findings in a way that makes sense to managerial and commissioning audiences, e.g. publish articles in NHS professional and health service commissioner journals

- Sharing learning of early implementers of FH services

- Scoping the possibility of producing a FH toolkit incorporating advice on FH service design and a revised costing template for FH services that is generalizable to the English NHS, based on CCG geographies.

The project anticipates that the dissemination strategy will be fully implemented in 2021, with all the publications in the public domain.

The benefits to patients are, If patients with familial hypercholesterolaemia (FH) are not identified they can die at an early age from myocardial Infarction and other cardiovascular disease. It has been well-established that patients with FH who are not diagnosed and treated will have a 50% chance of coronary mortality by 50 years of age in men and 60 years of age in women. At a population level, the UK is incredibly poor at diagnosing FH with close to 80% under diagnosis. This accounts for anywhere from 120,000 to 180,000 individuals currently in the general population who do not know they are affected. Hence, improving the current low detection rate of FH is urgently needed.

One of the most cost-effective solutions is to develop protocols of efficient cascade testing to identify affected relatives, especially younger relatives, and to initiate early statin treatment to lower LDL cholesterol will prevent and reduce premature mortality, and long-term coronary morbidity. Available service data highlight the major extent of the problem. National audits show only around one affected relative is identified for each index case. To improve identification and health outcomes for these patients, evidence-based protocols need to be implemented. This data release of using one highest quality and well-established FH disease registries known internationally and linking outcomes to HES will provide the longest follow-up to date. This will allow for novel research to be generated on the long-term outcomes for patients affected by FH and the impact of treatment in real-world settings (as knowledge of impact of treatment for FH is limited to trials).

Many of our patient representatives who themselves have FH, from our partner charities, Heart UK and the British Heart Foundation, have expressed a dire need to understand the effectiveness of treatment

on long-term outcomes and how these outcomes can inform the development of more effective and efficient ways to cascade screen for relatives.

The benefits to health care practitioners are, many health care practitioners, in particular in primary care commissioners, are still unaware of the magnitude of underdiagnosis, how to identify, or manage FH in current clinical practice. The research outputs, once published, will improve awareness among practitioners and patients, leading to improved compliance anticipated by 2022.

Further, the economic model will improve the identification of FH through revised shared care protocols to be reviewed by the next NICE review of the FH guidelines 2024, leading to commissioning of evidence based FH cascade screening programmes and, ultimately, saving lives.

The benefits to the policy-makers and NHS commissioners are, cascade testing recommendations exist for the UK but policy-makers have acknowledged they do not reflect the current evidence base, which has evolved significantly since guidance was issued in 2008. Hence, this research project was commissioned by the Department of Health and HTA. There is significant uncertainty regarding whether cascade testing is cost-effective and how it should be implemented. As a result, cascade testing protocols are often piecemeal with variable implementation of services in the UK. To support commissioners deciding on whether to implement a new cascade testing service, or modify an existing service, there is an urgent need for new cost-effectiveness modelling that collates and synthesizes the best available current evidence, evaluates the overall impacts of cascade testing for all affected individuals and answers key questions about how the service should be designed to offer the best possible value for money.

The outputs, which are expected will be published in by 2021, will allow the team to identify the most cost-effective design for a FH cascade testing protocol in the NHS which will likely result in a change in the current ways cascade testing in implemented in the UK. The research outcomes will inform future NICE implementation for FH cascade testing based on our economic model, to support analysis of the cost, outcomes and savings associated with implementing FH cascade services at local level. Combined with findings from the proposed qualitative research, this work will provide valuable practical information to health care decision makers involved in commissioning or running FH services regarding the appropriate use of NHS resources.

Benchmarking: A very similar analysis has been conducted in the context of stable coronary artery disease as part of the University of York led CALIBER project using HES data. The team plan to benchmark the analysis and outcomes of cardiovascular outcomes against the well-established CALIBER programme findings. This will ensure a level quality control against cardiovascular outcome definitions and comparison of findings.

The study has allowed the team to produce long-term estimates of the risk of acute coronary syndrome, stroke/TIA and cardiovascular deaths in FH patients. This has supported the advocacy work of the respective charities (HeartUK and British Heart Foundation).

Further, the statistical models produced from this data is informing a cost-effectiveness model quantifying the costs and benefits of diagnosis and treatment for FH patients. This model will be used to assess the cost-effectiveness of different configurations of FH cascade testing services for the purposes of the ongoing NIHR HTA grant. The model will also provide the basis for future assessments of interventions to improve FH diagnosis and treatment.

Most recent analysis of the dataset has also highlighted to lipid specialists the persistent inequalities in management of the condition, with poorer morbidity and mortality in female patients and those from deprived backgrounds.