Pulmonary Arterial Hypertension (PAH) population epidemiological analysis platform formation

Hospital Episode Statistics (HES) data was supplied to IQVIA Solutions Ltd and IQVIA Technology Service by the Health and Social Care Information Centre (which has since become NHS Digital) for the purpose of a research study that aims to evaluate at the diagnostic and treatment pathways for patients suffering with pulmonary arterial hypertension.

IQVIA Ltd was added as a joint Data Controller to the previous agreement as IQVIA simplified the number of trading legal entities it had in the UK by transferring the business and assets from IQVIA Solutions UK Ltd into IQVIA Ltd.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study.

BACKGROUND:

Pulmonary Arterial Hypertension (PAH) is a disease primarily of small arteries in the lung which results in a progressive rise in lung blood pressure and heart failure. There are several types of PAH including Idiopathic PAH (iPAH) and Associated PAH related to a range of disease processes, including cirrhosis, connective tissue disease, congenital heart disease, HIV infection and sickle-cell disease.

The difficulties of early PAH diagnosis are well understood; signs and symptoms are subtle, there is no single approach for non-invasive, specialist diagnosis and misdiagnosis is common (Gibbs et al, 2015). Contemporary PAH literature discusses the challenges of PAH diagnosis and the urgent need for novel tools to detect patients earlier (Lau et al, 2014) (Forfia and Trow, 2013).

Late diagnosis of PAH is common and leads to significantly worse outcomes, however identifying patients with PAH earlier can allow targeted therapies to be started before the development of significant right heart failure and thus vastly improve patients overall survival and quality of life (Hoeper et al., 2013)

IQVIA Solutions UK Limited were previously commissioned by GlaxoSmithKline to carry out a retrospective analysis of UK iPAH patients in the English Hospital Episode Statistics (HES) data. The study focused on diagnosis pathways but also considered post-diagnosis treatment patterns of patients. This was initially commissioned to improve GSK’s understanding of PAH disease and patient care in England. For the purpose of the current amendment (please see Outputs section), GSK is no longer funding the research and has now been removed as a sponsor (please see Funding section).

The findings further confirmed there is a large unmet need for early diagnosis, with results showing that there is a high level of activity pre-diagnosis with the average patient having 25 events in 3 years prior to diagnosis. Of those, 12 are within the final year before Right Heart Catheterisation (the confirmatory diagnostic test for PAH).

IQVIA Ltd believe there are opportunities to identify iPAH patients earlier based on the pattern of patients' interaction with secondary care facilities, symptoms shown and demographics, therefore identifying predictive signals/ markers which could lead to an earlier diagnosis of iPAH patients.

The original HES analysis highlighted that when patients hit Sheffield Teaching Hospitals NHS Foundation Trust (STHFT) they appear to be diagnosed quicker than other centers, thus leading the applicant to hypothesize that the patient care pathway at the Sheffield Pulmonary Vascular Disease Unit (SPVDU) is optimised for quicker patient diagnosis and potentially leads to improved PAH patient outcomes. Therefore understanding the differences in patient pathways can lead to learning’s which could influence patient management at other centres.

These outputs, gave cause to believe that there is potentially high value in pursuing further analysis of this data when coupled with the enhanced diagnostic clinical data jointly held by STHFT and the University of Sheffield (UoS), leading to the IQVIA Solutions UK Limited (as was the correct legal entity at that time) approaching STHFT/University of Sheffield for partnership.

Research overview:

The goal of the research is to:

• Validate the original analysis using STHFT’s data to confirm patient diagnosis of the selected cohort

• Understand the patients diagnostic pathway and outcomes of going through different routes to diagnosis

• Understand how SPVDU has streamlined their diagnostic process to allow quicker diagnosis of PAH patients when they enter the specialist center

• Utilising linked clinical and biological data (available in Sheffield’s data) to define novel disease phenotypes

• Develop a predictive algorithm which would be able to flag patients with a high probability of having idiopathic PAH (iPAH) from their data “fingerprint”. This will support finding undiagnosed patients through developing a predictive algorithm

In order to achieve the objectives, the IQVIA Solutions UK Limited proposes to build a joint dataset in order to develop analysis to test these hypotheses. The database will be comprised of identifiable patient data derived from the STHFT “deep” clinical databases which collect data on all patients attending the SPVDU and national level hospital interactions from HES data.

Parties involved in the research:

Each party in the collaboration will have a different role during the research:

• STHFT will take responsibility for ethics approval for the study, provide expert clinical insight on the research findings, support on datasets de-identification, linkage and transformation in addition to supporting the publication of research findings

• IQVIA Limited will support STHFT ethical approvals activities, conduct the transformation and data processing of de-identified data into analysable format and perform the analysis described in this agreement. IQVIA Limited has significant experience with HES data, other retrospective databases, outcomes research expertise and advanced machine learning capabilities for predictive algorithm development.

• The University of Sheffield (UoS) will provide clinical interpretation of the results. UoS are not permitted to access record level HES data. UoS only ever have access to aggregated data with small number suppressed in line with the HES Analysis Guide.

STHFT as one of England’s leading PAH diagnostic and treatment centres benefits from the research by furthering their understanding of patient journeys outside of the Sheffield Pulmonary Vascular Disease Unit (SPDVU), in addition to the verification that their unique diagnostic process is beneficial to patients, allowing them to share their learnings with other centres.

The research focuses on the diagnostic pathway of patients, in a disease area where specialists and publications indicate there is a large degree of late diagnosis and this in turn impacts the efficacy of medicines and thus outcomes of the patients. However to ensure findings are published fairly and not suppressed there will be a clinical interpretation group in place. This is comprised of 2 representatives of each STHFT and UoS with IQVIA Health limited chairing the group.

The committee will perform the following functions:

1) Provide clinical interpretation of the results to support refinements of the analysis within the bounds of the protocol

2) Agree the dissemination / publication routes for research findings (e.g. conference posters vs peer review papers etc.) based on the nature and strength of findings. (Please see output section for further information)

No organisation on the clinical interpretation group will have the ability to suppress any of the findings or outputs of the analysis. The clinical interpretation group members do not have any access to record level data.

The studies chief investigator from STHFT, will oversee the research and offer clinical insight on the findings. The patient selection criteria has been based on patients who attended STHFT and those who share similar symptomology to PAH patients, this has been developed and chosen by IQVIA Solutions UK Limited in conjunction with the chief investigator from STHFT. The dissemination of findings have been pre-agreed and outlined in the outputs section.

Data retention times has been agreed in CAG, REC and in the data sharing agreement that will be in place with the NHS Digital upon approval of the application. If IQVIA Ltd requires more time for the analysis they will request an extension on the agreement with NHS Digital.

Why link data:

It is important to link HES data with the STHFT dataset in order to utilise the confirmed and sub-typed PAH patient diagnoses present in the STHFT dataset, where the patient PAH classification has been confirmed by world leading clinical experts. This will allow the IQVIA Solutions UK Limited to identify patients with confirmed PAH (and subtypes of PAH) within the HES data for investigation and analysis with high certainty. Current ICD-10 coding (the International classification system for coding of disease types, maintained by the World Health Organisation) does not have a specific code for PAH, with multiple different pulmonary diseases coded under the same ICD-10 code. In addition coding is not consistently applied across centres, meaning that PAH patients in HES are coded across many different ICD-10 codes and therefore confirmation of disease and subtype in HES alone is not possible with complete certainty.

In addition to providing clarity on the patients actual diagnosis, the STHFT data will provide insight on all the patients who have attended SPDVU, this is important as the applicant wishes to understand the diagnostic pathway and process at SPDVU, including those patients suspected of having a PAH diagnosis and subsequently being diagnosed with other conditions.

What data is requested:

The study design is a retrospective database analysis of data collect on patients who have attended the SPVDU at STHFT.

In order to facilitate this project the applicant is requesting 2 different cohorts of patients from NHS Digital:

1) Cohort A: Patients who have been managed at the SPDVU since 2000 – which will allow IQVIA Solutions UK Limited to confirm the patient diagnosis (and subtype) in HES data, verify the original cohort selection in the previous HES analysis and understand the diagnostic pathway in SPDVU and why it is quicker than other centres (as shown by previous HES analysis)

2) Cohort B: A comparison group of patients - This group will be used in the development of the predictive algorithm, which will allow the applicant to use statistical techniques to compare the differences in care pathways of confirmed PAH patients (from cohort 1) and those patients who do not have confirmed PAH (from cohort 2). This requires IQVIA Solutions UK Limited to look in detail at a group of patients similar to the confirmed cohort. IQVIA Solutions UK Limited have done this by selecting patients with confounding or differential diagnosis to the PAH diagnosis, and there is various scientific literature which shows the association of these conditions with PAH/ pulmonary hypertension (PH).

The second cohort selection criteria are as follows:

• Historical patient data for selected cohort from 2000

• No patients under the age of 18

• Full (including historical) records for patients with any of the following ICD-10 codes within any diagnosis position: Dilated cardiomyopathy (I42.0), Hypothyroidism (E03.9), Mitral Stenosis (I05.0, I34.2 OR Q23.2), Mixed Connective-Tissue Disease (M35.1), Obstructive Sleep Apnoea (G47.3), Systemic Lupus Erythematosus (M32), Portal Hypertension (K76.6), Pulmonic Stenosis (I37.0), Scleroderma (L94.0, L94.1 OR M43), Ischaemic heart diseases (I20-I25), Heart failure (I50), Pulmonary heart disease and diseases of pulmonary circulation (I26 – I28), Asthma (J45), COPD (J47 OR J40 - J44) and Interstitial lung disease (J84.9).

If a patient has any of the above ICD-10 codes the applicant would like to have the full longitudinal patient record. Due to the complicated disease area and goals of the research the patient pathway analysis requires a long period of data for the following reasons:

• Understanding impact of STHFT changes to service:

Previous work at STHFT has resulted in the improvement of the diagnostic process of pulmonary conditions. Firstly by streamlining the diagnostic process within STFHT to allow the majority of patient to be diagnosed within 2 consultations, secondly by continuing medical education outreach to satellite centres through talks and guideline publications. The historical length of data will allow the measurement of the impact of these improvements and support messaging to other specialist centres to allow them to adopt the learnings from these efforts, thus potentially improving diagnostic efforts and thus patient outcomes.

Furthermore the requested length of HES data aligns with the length of data held by STHFT allowing the applicant to utilise the full breadth of clinical data that STHFT hold.

• Having sufficient time to understand patient activity from onset of symptoms to diagnosis:

IQVIA Solutions UK Limited are requesting 2 ~15 year historical extract of data for the PAH project to cover both requested cohorts (patients who have attended SPDVU and cohort for development of the predictive algorithm).

The reason being that PAH patient populations (especially at subtype level) are very small and the diagnosis pathway is a long multi-year processes and often complex, the previous HES analysis showed that patients have a very high level of activity pre-diagnosis with >1/5th of patients experiencing hospitalisations, consultations or symptoms relating to IPAH disease >3 years before a positive diagnosis.

In addition the need to create a sophisticated algorithm that has the potential to perform well in the live clinical environment, a large sample of data is required. This is driven by the following reasons::

• Disease characteristics:

Cohort B was selected to try and ensure that the applicant adheres to data minimisation rules but also has enough data for meaningful analysis. The comparison group (cohort B) needs to be similar enough to the confirmed PAH cohort (cohort A), so the algorithm development process can start to identify the differences between patients who are often confused for PAH patients and those with a confirmed PAH diagnosis. PAH signs and symptoms are subtle and often confused with a range of different conditions. This means that the comparison group (cohort B) needs to be created from a sample of patients who share symptomology which is similar to PHA or occurs in conjunction with PAH disease. Minimising this data will lead to the development of a biased algorithm (For further information see the 180119_PAH Predictive algorithm overview- HES application Vf.dox).

• Refining the cohort based on clinical characteristics:

In order to select the most appropriate cohort of patients to act as a comparison group to confirmed PAH patients (cohort A), IQVIA Solutions UK Limited require to undergo analysis of the patient data, this is a data driven approach coupled with insights from the clinical specialists. As noted previously PAH patients are often misdiagnosed as other conditions due to the rarity of the disease and huge range of clinical manifestations they can present with. The aim is to identify a cohort of patients which do not have a confirmed diagnosis but share very similar clinical features, have contaminant diagnosis, visit the same specialists etc. This allows development of the algorithm on a comparison group as close to the real cases physicians experience in clinical practice as possible and thus stretch the algorithm as much as possible.

For example, in previous work, IQVIA created an algorithm to identify a rare disease population (Idiopathic Pulmonary Fibrosis), which manifests as a lung condition commonly misdiagnosed as asthma or COPD. To focus the algorithm on the clinical challenge IQVIA developed the algorithm to distinguish between IPF patients (8,574 patients) and those with COPD/Asthma (7.5m patients). In order to find the most appropriate comparison group to our confirmed PAH patients (cohort A) it requires a deep dive into the data to align the patient cohorts

• Refining the cohort based on availability of appropriate length of historic data:

The size of a cohort is limited not only by the number of patients with given diagnosis, but also the need to have available a sufficient time period both prior to the diagnosis (to observe baseline characteristics) as well as after the event (to observe relevant outcomes) for analysis. For example a recent project in Fabry Disease, one focus of analysis was to understand the diagnostic pathway, in order to identify any predictive signals/ markers which would allow earlier diagnosis of Fabry disease and thus slowing progression of the disease by allowing earlier treatment. The study by IQVIA identified 665 patients with suspected Fabry disease, of those patients only 90 patients had 3+ years of historical data available to allow analysis of the lead up to patient diagnosis (which was much shorter than desirable given the often 20 year symptom onset in this condition). This patient cohort size prevented IQVIA Solutions UK Limited from having sufficient numbers to conduct robust predictive analytics on the data to find signals/ markers of disease.

A recent study of idiopathic PAH (IPAH) patients found that a significant delay of 3.9 years from symptom onset to a diagnosis of IPAH (Strange et al. 2013). Indicating that a long time window is required and limiting that number of patients that will have the time window available for analysis.

• Bringing the algorithm to clinical practice:

If the algorithm were to be implemented in real clinical practice setting the algorithm can only run on patients who fit inclusion and exclusion criteria used to pull HES data. Therefore the narrower the patient sample requested means that the more limited real world sample that can be assessed for risk of disease. For example if IQVIA only requested a sample of HES data made up of male patients who are over 40 years old. This would mean that IQVIA could not expect the model to produce robust predictions for any female patients or patients under the age of 40.

Due to these reasons the applicant requires HES data for a longer period than the usual 5 year period routinely offered by NHS Digital in order to capture sufficient patients for the analysis.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

As a result of initial analysis of the diagnostic treatment pathway for different subtypes of PAH, the following abstracts were accepted and presented at conferences:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” - accepted by ERS (European Respiratory Society) International Congress 2018. This abstract explained the methodology of predictive algorithm development and quantified the resulting algorithm’s performance.

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” – accepted by ATS (American Thoracic Society) International Conference 2018. This abstract described the process and rationale of linking HES data with specialist centre data and results from an exploration of the resulting dataset to conclude that sufficient patient numbers and depth of clinical information was available to support predictive algorithm development.

Building on the acceptance of these abstracts to publish findings in more detail:

• The following manuscript was published in the Journal of Pulmonary Circulation: “High levels of healthcare utilization prior to diagnosis in idiopathic pulmonary arterial hypertension support the feasibility of an early diagnosis algorithm: the SPHInX project”.

• The following manuscript has been submitted to both The Lancet and to Nature: “Utilising artificial intelligence to determine patients at risk of a rare disease: idiopathic pulmonary arterial hypertension”.

The following manuscripts are also under development:

• “Descriptive overview of the iPAH diagnostic process and pathway at SPVDU and impact of distance to specialist centre and social deprivation on diagnostic rates” (final title TBC) – targeted for the European Respiratory Journal. This manuscript will include:

o Overview of Sheffield’s approach to diagnosing iPAH patients after they enter the pulmonary unit

o Findings from an analysis of iPAH patient secondary care interventions as they approach diagnosis

• “Analysis of the diagnostic pathway for PH patients attending SPVDU” - final title and target journal TBC. This manuscript will include:

o A descriptive overview of the patient profiles of PH patients who have attended SPVDU (focussing on demographics and subtypes)

o Broader overview of SPVDU approach to diagnosing patients, focussing on other subtypes of PH patient who are seen at SPVDU

In addition to the current manuscript submissions and those in development, additional dissemination plans are as follows:

• Published results will be shared with the PHA UK patient advocacy group

• Abstracts and links to publications will be hosted on IQVIA Limited’s online bibliography which is publicly available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

An extension of the HES licence is required to both support any supplementary questions arising from publications and to support further publications in this area.

Included below are the “outputs expected” submitted with the original data access application, for context and reference. This includes details of the nature of outputs from predictive algorithm development.

Previous Outputs

IQVIA Solutions UK Limited have conducted initial analysis of the diagnostic and treatment pathway for different subtypes, in addition to investigating predictive patient characteristics within the data to support the flagging of patients earlier within their diagnostic pathway via a predictive algorithm. The findings of both indicated that further refinement of results will be beneficial.

IQVIA Solutions UK Limited are in the process of publishing several findings, the following abstracts have been submitted:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

An extension of the HES licence is required to both support any supplementary questions arising from publications, in addition to supporting further publications in this area. The extension is also required for IQVIA and Sheffield to hold the data and all the work done to date while the PH National Audit data is still being onboarded to the DARS portal and not available to IQVIA and Sheffield.

In addition to the current abstract submissions, additional decimation plans are as follows:

• The applicant will submit the additional findings of the research to a peer review journal e.g. Thorax - BMJ Journals.

• The applicant will submit and present on findings at the 2018 ATS conference, in addition to other important pulmonary conferences, in order to further the knowledge of other specialist physicians

• Published results will be shared with the PHA UK patient advocacy group

• Furthermore the abstracts and links to publications will be hosted on IQVIA Solutions UK Limited’s online bibliography which is publically available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

The current output of the algorithm generation is currently uncertain. However any implementation would need to be conducted by or with NHS bodies, because IQVIA are working with pseudonymous data and will not seek to re-identify patients at any stage. The nature of any implementation would need to be driven by the predictive sensitivity and specificity of the algorithm. In other words, the false positive and false negative detection rate. Implementing an algorithm with a high false positive rate would lead to many people tested with very few identified, conversely if the algorithm has a high false negative rate, it will likely miss many patients who should be tested for the disease. The health economics of the algorithm and any associated intervention would need to be carefully assessed prior to any implementation. Prior to any algorithm playing a role in supporting clinical practice / being implemented it will require peer review publication and broad acceptance before any uptake could be successful.

The algorithm will be free of charge and openly available. Access methods will be dependent on the strength of the algorithm but may include presentation at seminars, publications on risk factors or a clinical support tool provided directly to physicians (subject to any relevant approvals).

For an algorithm with weaker predictive potential IQVIA envisions the generation of publications in peer reviewed journals and generation of medical educational materials to use with clinical specialities who are potentially exposed to PAH patients. The literature will document the methodology used and the risk factors which would help to identify PAH patients earlier. These will potentially be presented at symposiums or other forums, depending on the findings.

If an algorithm with high predictive potential is generated, it could be used to create a clinical support tool for physicians to help diagnose patients, allowing the summarisation of large quantities of data in a more manageable format. This tool could support physicians by providing a risk score which they can interpret themselves to support clinician decisions.

If the applicant does not find any information of merit they will submit the methodology utilised in the research to a peer reviewed journal, this will allow other researchers to benefit from their research efforts. In addition the methodology will be shared via IQVIA Solutions UK Limited’s online bibliography and which is publically available.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

Several abstracts have been submitted and approved, with several more underway:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

Results of analysis will be presented at the European Respiratory Society this year.

The initial predictive algorithm is indicating a large benefit to disease detection efforts and it is anticipated that the predictive algorithm, upon further refinement can be used to support screening programs. The algorithm is developed to flag the high risk patients who share characteristics in common with iPAH patients. iPAH disease prevalence is 5 in 1 million meaning that >20 million patients at random would have to be tested to find 100 patients with iPAH disease. Utilising the developed predictive algorithm on the HES data it was found that if the top 1000 patients flagged as “high risk” were analysed based on our algorithm patients would be detected with the disease.

Hospital Episode Statistics (HES) data was supplied to IQVIA Solutions UK ltd and IQVIA Technology Service by the Health and Social Care Information Centre (which has since become NHS Digital) for the purpose of a research study that aims to evaluate at the diagnostic and treatment pathways for patients suffering with pulmonary arterial hypertension.

IQVIA Ltd has been added as a joint Data Controller as IQVIA are simplifying the number of trading legal entities it has in the UK by transferring the business and assets from IQVIA World Publications Ltd and IQVIA Solutions UK Limited into IQVIA Ltd.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

BACKGROUND:

Pulmonary Arterial Hypertension (PAH) is a disease primarily of small arteries in the lung which results in a progressive rise in lung blood pressure and heart failure. There are several types of PAH including Idiopathic PAH (iPAH) and Associated PAH related to a range of disease processes, including cirrhosis, connective tissue disease, congenital heart disease, HIV infection and sickle-cell disease.

The difficulties of early PAH diagnosis are well understood; signs and symptoms are subtle, there is no single approach for non-invasive, specialist diagnosis and misdiagnosis is common (Gibbs et al, 2015). Contemporary PAH literature discusses the challenges of PAH diagnosis and the urgent need for novel tools to detect patients earlier (Lau et al, 2014) (Forfia and Trow, 2013).

Late diagnosis of PAH is common and leads to significantly worse outcomes, however identifying patients with PAH earlier can allow targeted therapies to be started before the development of significant right heart failure and thus vastly improve patients overall survival and quality of life (Hoeper et al., 2013)

IQVIA Solutions UK Limited have previously been commissioned by GlaxoSmithKline to carry out a retrospective analysis of UK iPAH patients in the English Hospital Episode Statistics (HES) data. The study focused on diagnosis pathways but also considered post-diagnosis treatment patterns of patients. This was initially commissioned to improve GSK’s understanding of PAH disease and patient care in England. For the purpose of the current amendment (please see Outputs section), GSK is no longer funding the research and has now been removed as a sponsor (please see Funding section).

The findings further confirmed there is a large unmet need for early diagnosis, with results showing that there is a high level of activity pre-diagnosis with the average patient having 25 events in 3 years prior to diagnosis. Of those, 12 are within the final year before Right Heart Catheterisation (the confirmatory diagnostic test for PAH).

The IQVIA Solutions UK Limited believes that there are opportunities to identify iPAH patients earlier based on the pattern of patients' interaction with secondary care facilities, symptoms shown and demographics, therefore identifying predictive signals/ markers which could lead to an earlier diagnosis of iPAH patients.

Secondly, the original HES analysis highlighted that when patients hit Sheffield Teaching Hospitals NHS Foundation Trust (STHFT) they appear to be diagnosed quicker than other centers, thus leading the applicant to hypothesize that the patient care pathway at the Sheffield Pulmonary Vascular Disease Unit (SPVDU) is optimised for quicker patient diagnosis and potentially leads to improved PAH patient outcomes. Therefore understanding the differences in patient pathways can lead to learning’s which could influence patient management at other centres.

These outputs, gave cause to believe that there is potentially high value in pursuing further analysis of this data when coupled with the enhanced diagnostic clinical data jointly held by STHFT and the University of Sheffield (UoS), leading to the IQVIA Solutions UK Limited approaching STHFT/University of Sheffield for partnership.

Research overview:

The goal of the research is to:

• Validate the original analysis using STHFT’s data to confirm patient diagnosis of the selected cohort

• Understand the patients diagnostic pathway and outcomes of going through different routes to diagnosis

• Understand how SPVDU has streamlined their diagnostic process to allow quicker diagnosis of PAH patients when they enter the specialist center

• Utilising linked clinical and biological data (available in Sheffield’s data) to define novel disease phenotypes

• Develop a predictive algorithm which would be able to flag patients with a high probability of having idiopathic PAH (iPAH) from their data “fingerprint”. This will support finding undiagnosed patients through developing a predictive algorithm

In order to achieve the objectives, the IQVIA Solutions UK Limited proposes to build a joint dataset in order to develop analysis to test these hypotheses. The database will be comprised of identifiable patient data derived from the STHFT “deep” clinical databases which collect data on all patients attending the SPVDU and national level hospital interactions from HES data.

Parties involved in the research:

Each party in the collaboration will have a different role during the research:

• STHFT will take responsibility for ethics approval for the study, provide expert clinical insight on the research findings, support on datasets de-identification, linkage and transformation in addition to supporting the publication of research findings

• IQVIA Solutions UK Limited will support STHFT ethical approvals activities, conduct the transformation and data processing of de-identified data into analysable format and perform the analysis described in this agreement. IQVIA Solutions UK Limited has significant experience with HES data, other retrospective databases, outcomes research expertise and advanced machine learning capabilities for predictive algorithm development.

• The University of Sheffield (UoS) will provide clinical interpretation of the results. UoS are not permitted to access record level HES data. UoS only ever have access to aggregated data with small number suppressed in line with the HES Analysis Guide.

STHFT as one of England’s leading PAH diagnostic and treatment centres benefits from the research by furthering their understanding of patient journeys outside of the Sheffield Pulmonary Vascular Disease Unit (SPDVU), in addition to the verification that their unique diagnostic process is beneficial to patients, allowing them to share their learnings with other centres.

The research focuses on the diagnostic pathway of patients, in a disease area where specialists and publications indicate there is a large degree of late diagnosis and this in turn impacts the efficacy of medicines and thus outcomes of the patients. However to ensure findings are published fairly and not suppressed there will be a clinical interpretation group in place. This is comprised of 2 representatives of each STHFT and UoS with IQVIA Health limited chairing the group.

The committee will perform the following functions:

1) Provide clinical interpretation of the results to support refinements of the analysis within the bounds of the protocol

2) Agree the dissemination / publication routes for research findings (e.g. conference posters vs peer review papers etc.) based on the nature and strength of findings. (Please see output section for further information)

No organisation on the clinical interpretation group will have the ability to suppress any of the findings or outputs of the analysis. The clinical interpretation group members do not have any access to record level data.

The studies chief investigator is Professor David Kiely from STHFT, who will oversee the research and offer clinical insight on the findings. The patient selection criteria has been based on patients who attended STHFT and those who share similar symptomology to PAH patients, this has been developed and chosen by IQVIA Solutions UK Limited in conjunction with Professor David Kiely from STHFT. The dissemination of findings have been pre-agreed and outlined in the outputs section.

Data retention times has been agreed in CAG, REC and in the data sharing agreement that will be in place with the NHS Digital upon approval of the application. If IQVIA Solutions UK Limited requires more time for the analysis they will request an extension on the agreement with NHS Digital.

Why link data:

It is important to link HES data with the STHFT dataset in order to utilise the confirmed and sub-typed PAH patient diagnoses present in the STHFT dataset, where the patient PAH classification has been confirmed by world leading clinical experts. This will allow the IQVIA Solutions UK Limited to identify patients with confirmed PAH (and subtypes of PAH) within the HES data for investigation and analysis with high certainty. Current ICD-10 coding (the International classification system for coding of disease types, maintained by the World Health Organisation) does not have a specific code for PAH, with multiple different pulmonary diseases coded under the same ICD-10 code. In addition coding is not consistently applied across centres, meaning that PAH patients in HES are coded across many different ICD-10 codes and therefore confirmation of disease and subtype in HES alone is not possible with complete certainty.

In addition to providing clarity on the patients actual diagnosis, the STHFT data will provide insight on all the patients who have attended SPDVU, this is important as the applicant wishes to understand the diagnostic pathway and process at SPDVU, including those patients suspected of having a PAH diagnosis and subsequently being diagnosed with other conditions.

What data is requested:

The study design is a retrospective database analysis of data collect on patients who have attended the SPVDU at STHFT.

In order to facilitate this project the applicant is requesting 2 different cohorts of patients from NHS Digital:

1) Cohort A: Patients who have been managed at the SPDVU since 2000 – which will allow IQVIA Solutions UK Limited to confirm the patient diagnosis (and subtype) in HES data, verify the original cohort selection in the previous HES analysis and understand the diagnostic pathway in SPDVU and why it is quicker than other centres (as shown by previous HES analysis)

2) Cohort B: A comparison group of patients - This group will be used in the development of the predictive algorithm, which will allow the applicant to use statistical techniques to compare the differences in care pathways of confirmed PAH patients (from cohort 1) and those patients who do not have confirmed PAH (from cohort 2). This requires IQVIA Solutions UK Limited to look in detail at a group of patients similar to the confirmed cohort. IQVIA Solutions UK Limited have done this by selecting patients with confounding or differential diagnosis to the PAH diagnosis, and there is various scientific literature which shows the association of these conditions with PAH/ pulmonary hypertension (PH).

The second cohort selection criteria are as follows:

• Historical patient data for selected cohort from 2000

• No patients under the age of 18

• Full (including historical) records for patients with any of the following ICD-10 codes within any diagnosis position: Dilated cardiomyopathy (I42.0), Hypothyroidism (E03.9), Mitral Stenosis (I05.0, I34.2 OR Q23.2), Mixed Connective-Tissue Disease (M35.1), Obstructive Sleep Apnoea (G47.3), Systemic Lupus Erythematosus (M32), Portal Hypertension (K76.6), Pulmonic Stenosis (I37.0), Scleroderma (L94.0, L94.1 OR M43), Ischaemic heart diseases (I20-I25), Heart failure (I50), Pulmonary heart disease and diseases of pulmonary circulation (I26 – I28), Asthma (J45), COPD (J47 OR J40 - J44) and Interstitial lung disease (J84.9).

If a patient has any of the above ICD-10 codes the applicant would like to have the full longitudinal patient record. Due to the complicated disease area and goals of the research the patient pathway analysis requires a long period of data for the following reasons:

• Understanding impact of STHFT changes to service:

Previous work at STHFT has resulted in the improvement of the diagnostic process of pulmonary conditions. Firstly by streamlining the diagnostic process within STFHT to allow the majority of patient to be diagnosed within 2 consultations, secondly by continuing medical education outreach to satellite centres through talks and guideline publications. The historical length of data will allow the measurement of the impact of these improvements and support messaging to other specialist centres to allow them to adopt the learnings from these efforts, thus potentially improving diagnostic efforts and thus patient outcomes.

Furthermore the requested length of HES data aligns with the length of data held by STHFT allowing the applicant to utilise the full breadth of clinical data that STHFT hold.

• Having sufficient time to understand patient activity from onset of symptoms to diagnosis:

IQVIA Solutions UK Limited are requesting 2 ~15 year historical extract of data for the PAH project to cover both requested cohorts (patients who have attended SPDVU and cohort for development of the predictive algorithm).

The reason being that PAH patient populations (especially at subtype level) are very small and the diagnosis pathway is a long multi-year processes and often complex, the previous HES analysis showed that patients have a very high level of activity pre-diagnosis with >1/5th of patients experiencing hospitalisations, consultations or symptoms relating to IPAH disease >3 years before a positive diagnosis.

In addition the need to create a sophisticated algorithm that has the potential to perform well in the live clinical environment, a large sample of data is required. This is driven by the following reasons::

• Disease characteristics:

Cohort B was selected to try and ensure that the applicant adheres to data minimisation rules but also has enough data for meaningful analysis. The comparison group (cohort B) needs to be similar enough to the confirmed PAH cohort (cohort A), so the algorithm development process can start to identify the differences between patients who are often confused for PAH patients and those with a confirmed PAH diagnosis. PAH signs and symptoms are subtle and often confused with a range of different conditions. This means that the comparison group (cohort B) needs to be created from a sample of patients who share symptomology which is similar to PHA or occurs in conjunction with PAH disease. Minimising this data will lead to the development of a biased algorithm (For further information see the 180119_PAH Predictive algorithm overview- HES application Vf.dox).

• Refining the cohort based on clinical characteristics:

In order to select the most appropriate cohort of patients to act as a comparison group to confirmed PAH patients (cohort A), IQVIA Solutions UK Limited require to undergo analysis of the patient data, this is a data driven approach coupled with insights from the clinical specialists. As noted previously PAH patients are often misdiagnosed as other conditions due to the rarity of the disease and huge range of clinical manifestations they can present with. The aim is to identify a cohort of patients which do not have a confirmed diagnosis but share very similar clinical features, have contaminant diagnosis, visit the same specialists etc. This allows development of the algorithm on a comparison group as close to the real cases physicians experience in clinical practice as possible and thus stretch the algorithm as much as possible.

For example, in previous work, IQVIA created an algorithm to identify a rare disease population (Idiopathic Pulmonary Fibrosis), which manifests as a lung condition commonly misdiagnosed as asthma or COPD. To focus the algorithm on the clinical challenge IQVIA developed the algorithm to distinguish between IPF patients (8,574 patients) and those with COPD/Asthma (7.5m patients). In order to find the most appropriate comparison group to our confirmed PAH patients (cohort A) it requires a deep dive into the data to align the patient cohorts

• Refining the cohort based on availability of appropriate length of historic data:

The size of a cohort is limited not only by the number of patients with given diagnosis, but also the need to have available a sufficient time period both prior to the diagnosis (to observe baseline characteristics) as well as after the event (to observe relevant outcomes) for analysis. For example a recent project in Fabry Disease, one focus of analysis was to understand the diagnostic pathway, in order to identify any predictive signals/ markers which would allow earlier diagnosis of Fabry disease and thus slowing progression of the disease by allowing earlier treatment. The study by IQVIA identified 665 patients with suspected Fabry disease, of those patients only 90 patients had 3+ years of historical data available to allow analysis of the lead up to patient diagnosis (which was much shorter than desirable given the often 20 year symptom onset in this condition). This patient cohort size prevented IQVIA Solutions UK Limited from having sufficient numbers to conduct robust predictive analytics on the data to find signals/ markers of disease.

A recent study of idiopathic PAH (IPAH) patients found that a significant delay of 3.9 years from symptom onset to a diagnosis of IPAH (Strange et al. 2013). Indicating that a long time window is required and limiting that number of patients that will have the time window available for analysis.

• Bringing the algorithm to clinical practice:

If the algorithm were to be implemented in real clinical practice setting the algorithm can only run on patients who fit inclusion and exclusion criteria used to pull HES data. Therefore the narrower the patient sample requested means that the more limited real world sample that can be assessed for risk of disease. For example if IQVIA only requested a sample of HES data made up of male patients who are over 40 years old. This would mean that IQVIA could not expect the model to produce robust predictions for any female patients or patients under the age of 40.

Due to these reasons the applicant requires HES data for a longer period than the usual 5 year period routinely offered by NHS Digital in order to capture sufficient patients for the analysis.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

As a result of initial analysis of the diagnostic treatment pathway for different subtypes of PAH, the following abstracts were accepted and presented at conferences:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” - accepted by ERS (European Respiratory Society) International Congress 2018. This abstract explained the methodology of predictive algorithm development and quantified the resulting algorithm’s performance.

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” – accepted by ATS (American Thoracic Society) International Conference 2018. This abstract described the process and rationale of linking HES data with specialist centre data and results from an exploration of the resulting dataset to conclude that sufficient patient numbers and depth of clinical information was available to support predictive algorithm development.

Building on the acceptance of these abstracts to publish findings in more detail:

• The following manuscript was published in the Journal of Pulmonary Circulation: “High levels of healthcare utilization prior to diagnosis in idiopathic pulmonary arterial hypertension support the feasibility of an early diagnosis algorithm: the SPHInX project”.

• The following manuscript has been submitted to both The Lancet and to Nature: “Utilising artificial intelligence to determine patients at risk of a rare disease: idiopathic pulmonary arterial hypertension”.

The following manuscripts are also under development:

• “Descriptive overview of the iPAH diagnostic process and pathway at SPVDU and impact of distance to specialist centre and social deprivation on diagnostic rates” (final title TBC) – targeted for the European Respiratory Journal. This manuscript will include:

o Overview of Sheffield’s approach to diagnosing iPAH patients after they enter the pulmonary unit

o Findings from an analysis of iPAH patient secondary care interventions as they approach diagnosis

• “Analysis of the diagnostic pathway for PH patients attending SPVDU” - final title and target journal TBC. This manuscript will include:

o A descriptive overview of the patient profiles of PH patients who have attended SPVDU (focussing on demographics and subtypes)

o Broader overview of SPVDU approach to diagnosing patients, focussing on other subtypes of PH patient who are seen at SPVDU

In addition to the current manuscript submissions and those in development, additional dissemination plans are as follows:

• Published results will be shared with the PHA UK patient advocacy group

• Abstracts and links to publications will be hosted on IQVIA Solutions UK Limited’s online bibliography which is publicly available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

An extension of the HES licence is required to both support any supplementary questions arising from publications and to support further publications in this area.

We note that in the current DSA, a data retention period of approximately 10 years (until 01/01/2027) was approved, to comply with NHS HRA and the MRC guidelines for data storage. It is recognised in the agreement that should this period exceed the term of the DSA, a new one will need to be put in place.

We have included below the “outputs expected” submitted with the original data access application, for context and reference. This includes details of the nature of outputs from predictive algorithm development.

*******************Below is the Outputs section from original application*****************

IQVIA Solutions UK Limited have conducted initial analysis of the diagnostic and treatment pathway for different subtypes, in addition to investigating predictive patient characteristics within the data to support the flagging of patients earlier within their diagnostic pathway via a predictive algorithm. The findings of both indicated that further refinement of results will be beneficial.

IQVIA Solutions UK Limited are in the process of publishing several findings, the following abstracts have been submitted:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

An extension of the HES licence is required to both support any supplementary questions arising from publications, in addition to supporting further publications in this area. The extension is also required for IQVIA and Sheffield to hold the data and all the work done to date while the PH National Audit data is still being onboarded to the DARS portal and not available to IQVIA and Sheffield.

In addition to the current abstract submissions, additional decimation plans are as follows:

• The applicant will submit the additional findings of the research to a peer review journal e.g. Thorax - BMJ Journals.

• The applicant will submit and present on findings at the 2018 ATS conference, in addition to other important pulmonary conferences, in order to further the knowledge of other specialist physicians

• Published results will be shared with the PHA UK patient advocacy group

• Furthermore the abstracts and links to publications will be hosted on IQVIA Solutions UK Limited’s online bibliography which is publically available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

The current output of the algorithm generation is currently uncertain. However any implementation would need to be conducted by or with NHS bodies, because IQVIA are working with pseudonymous data and will not seek to re-identify patients at any stage. The nature of any implementation would need to be driven by the predictive sensitivity and specificity of the algorithm. In other words, the false positive and false negative detection rate. Implementing an algorithm with a high false positive rate would lead to many people tested with very few identified, conversely if the algorithm has a high false negative rate, it will likely miss many patients who should be tested for the disease. The health economics of the algorithm and any associated intervention would need to be carefully assessed prior to any implementation. Prior to any algorithm playing a role in supporting clinical practice / being implemented it will require peer review publication and broad acceptance before any uptake could be successful.

The algorithm will be free of charge and openly available. Access methods will be dependent on the strength of the algorithm but may include presentation at seminars, publications on risk factors or a clinical support tool provided directly to physicians (subject to any relevant approvals).

For an algorithm with weaker predictive potential IQVIA envisions the generation of publications in peer reviewed journals and generation of medical educational materials to use with clinical specialities who are potentially exposed to PAH patients. The literature will document the methodology used and the risk factors which would help to identify PAH patients earlier. These will potentially be presented at symposiums or other forums, depending on the findings.

If an algorithm with high predictive potential is generated, it could be used to create a clinical support tool for physicians to help diagnose patients, allowing the summarisation of large quantities of data in a more manageable format. This tool could support physicians by providing a risk score which they can interpret themselves to support clinician decisions.

If the applicant does not find any information of merit they will submit the methodology utilised in the research to a peer reviewed journal, this will allow other researchers to benefit from their research efforts. In addition the methodology will be shared via IQVIA Solutions UK Limited’s online bibliography and which is publically available.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

Several abstracts have been submitted and approved, with several more underway:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

Results of analysis will be presented at the European Respiratory Society this year.

The initial predictive algorithm is indicating a large benefit to disease detection efforts and it is anticipated that the predictive algorithm, upon further refinement can be used to support screening programs. The algorithm is developed to flag the high risk patients who share characteristics in common with iPAH patients. iPAH disease prevalence is 5 in 1 million meaning that >20 million patients at random would have to be tested to find 100 patients with iPAH disease. Utilising the developed predictive algorithm on the HES data it was found that if the top 1000 patients flagged as “high risk” were analysed based on our algorithm patients would be detected with the disease.

Hospital Episode Statistics (HES) data was supplied to IQVIA Solutions UK ltd and IQVIA Technology Service by the Health and Social Care Information Centre (which has since become NHS Digital) for the purpose of a research study that aims to evaluate at the diagnostic and treatment pathways for patients suffering with pulmonary arterial hypertension.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

BACKGROUND:

Pulmonary Arterial Hypertension (PAH) is a disease primarily of small arteries in the lung which results in a progressive rise in lung blood pressure and heart failure. There are several types of PAH including Idiopathic PAH (iPAH) and Associated PAH related to a range of disease processes, including cirrhosis, connective tissue disease, congenital heart disease, HIV infection and sickle-cell disease.

The difficulties of early PAH diagnosis are well understood; signs and symptoms are subtle, there is no single approach for non-invasive, specialist diagnosis and misdiagnosis is common (Gibbs et al, 2015). Contemporary PAH literature discusses the challenges of PAH diagnosis and the urgent need for novel tools to detect patients earlier (Lau et al, 2014) (Forfia and Trow, 2013).

Late diagnosis of PAH is common and leads to significantly worse outcomes, however identifying patients with PAH earlier can allow targeted therapies to be started before the development of significant right heart failure and thus vastly improve patients overall survival and quality of life (Hoeper et al., 2013)

IQVIA Solutions UK Limited have previously been commissioned by GlaxoSmithKline to carry out a retrospective analysis of UK iPAH patients in the English Hospital Episode Statistics (HES) data. The study focused on diagnosis pathways but also considered post-diagnosis treatment patterns of patients. This was initially commissioned to improve GSK’s understanding of PAH disease and patient care in England. For the purpose of the current amendment (please see Outputs section), GSK is no longer funding the research and has now been removed as a sponsor (please see Funding section).

The findings further confirmed there is a large unmet need for early diagnosis, with results showing that there is a high level of activity pre-diagnosis with the average patient having 25 events in 3 years prior to diagnosis. Of those, 12 are within the final year before Right Heart Catheterisation (the confirmatory diagnostic test for PAH).

The IQVIA Solutions UK Limited believes that there are opportunities to identify iPAH patients earlier based on the pattern of patients' interaction with secondary care facilities, symptoms shown and demographics, therefore identifying predictive signals/ markers which could lead to an earlier diagnosis of iPAH patients.

Secondly, the original HES analysis highlighted that when patients hit Sheffield Teaching Hospitals NHS Foundation Trust (STHFT) they appear to be diagnosed quicker than other centers, thus leading the applicant to hypothesize that the patient care pathway at the Sheffield Pulmonary Vascular Disease Unit (SPVDU) is optimised for quicker patient diagnosis and potentially leads to improved PAH patient outcomes. Therefore understanding the differences in patient pathways can lead to learning’s which could influence patient management at other centres.

These outputs, gave cause to believe that there is potentially high value in pursuing further analysis of this data when coupled with the enhanced diagnostic clinical data jointly held by STHFT and the University of Sheffield (UoS), leading to the IQVIA Solutions UK Limited approaching STHFT/University of Sheffield for partnership.

Research overview:

The goal of the research is to:

• Validate the original analysis using STHFT’s data to confirm patient diagnosis of the selected cohort

• Understand the patients diagnostic pathway and outcomes of going through different routes to diagnosis

• Understand how SPVDU has streamlined their diagnostic process to allow quicker diagnosis of PAH patients when they enter the specialist center

• Utilising linked clinical and biological data (available in Sheffield’s data) to define novel disease phenotypes

• Develop a predictive algorithm which would be able to flag patients with a high probability of having idiopathic PAH (iPAH) from their data “fingerprint”. This will support finding undiagnosed patients through developing a predictive algorithm

In order to achieve the objectives, the IQVIA Solutions UK Limited proposes to build a joint dataset in order to develop analysis to test these hypotheses. The database will be comprised of identifiable patient data derived from the STHFT “deep” clinical databases which collect data on all patients attending the SPVDU and national level hospital interactions from HES data.

Parties involved in the research:

Each party in the collaboration will have a different role during the research:

• STHFT will take responsibility for ethics approval for the study, provide expert clinical insight on the research findings, support on datasets de-identification, linkage and transformation in addition to supporting the publication of research findings

• IQVIA Solutions UK Limited will support STHFT ethical approvals activities, conduct the transformation and data processing of de-identified data into analysable format and perform the analysis described in this agreement. IQVIA Solutions UK Limited has significant experience with HES data, other retrospective databases, outcomes research expertise and advanced machine learning capabilities for predictive algorithm development.

• The University of Sheffield (UoS) will provide clinical interpretation of the results. UoS are not permitted to access record level HES data. UoS only ever have access to aggregated data with small number suppressed in line with the HES Analysis Guide.

STHFT as one of England’s leading PAH diagnostic and treatment centres benefits from the research by furthering their understanding of patient journeys outside of the Sheffield Pulmonary Vascular Disease Unit (SPDVU), in addition to the verification that their unique diagnostic process is beneficial to patients, allowing them to share their learnings with other centres.

The research focuses on the diagnostic pathway of patients, in a disease area where specialists and publications indicate there is a large degree of late diagnosis and this in turn impacts the efficacy of medicines and thus outcomes of the patients. However to ensure findings are published fairly and not suppressed there will be a clinical interpretation group in place. This is comprised of 2 representatives of each STHFT and UoS with IQVIA Health limited chairing the group.

The committee will perform the following functions:

1) Provide clinical interpretation of the results to support refinements of the analysis within the bounds of the protocol

2) Agree the dissemination / publication routes for research findings (e.g. conference posters vs peer review papers etc.) based on the nature and strength of findings. (Please see output section for further information)

No organisation on the clinical interpretation group will have the ability to suppress any of the findings or outputs of the analysis. The clinical interpretation group members do not have any access to record level data.

The studies chief investigator is Professor David Kiely from STHFT, who will oversee the research and offer clinical insight on the findings. The patient selection criteria has been based on patients who attended STHFT and those who share similar symptomology to PAH patients, this has been developed and chosen by IQVIA Solutions UK Limited in conjunction with Professor David Kiely from STHFT. The dissemination of findings have been pre-agreed and outlined in the outputs section.

Data retention times has been agreed in CAG, REC and in the data sharing agreement that will be in place with the NHS Digital upon approval of the application. If IQVIA Solutions UK Limited requires more time for the analysis they will request an extension on the agreement with NHS Digital.

Why link data:

It is important to link HES data with the STHFT dataset in order to utilise the confirmed and sub-typed PAH patient diagnoses present in the STHFT dataset, where the patient PAH classification has been confirmed by world leading clinical experts. This will allow the IQVIA Solutions UK Limited to identify patients with confirmed PAH (and subtypes of PAH) within the HES data for investigation and analysis with high certainty. Current ICD-10 coding (the International classification system for coding of disease types, maintained by the World Health Organisation) does not have a specific code for PAH, with multiple different pulmonary diseases coded under the same ICD-10 code. In addition coding is not consistently applied across centres, meaning that PAH patients in HES are coded across many different ICD-10 codes and therefore confirmation of disease and subtype in HES alone is not possible with complete certainty.

In addition to providing clarity on the patients actual diagnosis, the STHFT data will provide insight on all the patients who have attended SPDVU, this is important as the applicant wishes to understand the diagnostic pathway and process at SPDVU, including those patients suspected of having a PAH diagnosis and subsequently being diagnosed with other conditions.

What data is requested:

The study design is a retrospective database analysis of data collect on patients who have attended the SPVDU at STHFT.

In order to facilitate this project the applicant is requesting 2 different cohorts of patients from NHS Digital:

1) Cohort A: Patients who have been managed at the SPDVU since 2000 – which will allow IQVIA Solutions UK Limited to confirm the patient diagnosis (and subtype) in HES data, verify the original cohort selection in the previous HES analysis and understand the diagnostic pathway in SPDVU and why it is quicker than other centres (as shown by previous HES analysis)

2) Cohort B: A comparison group of patients - This group will be used in the development of the predictive algorithm, which will allow the applicant to use statistical techniques to compare the differences in care pathways of confirmed PAH patients (from cohort 1) and those patients who do not have confirmed PAH (from cohort 2). This requires IQVIA Solutions UK Limited to look in detail at a group of patients similar to the confirmed cohort. IQVIA Solutions UK Limited have done this by selecting patients with confounding or differential diagnosis to the PAH diagnosis, and there is various scientific literature which shows the association of these conditions with PAH/ pulmonary hypertension (PH).

The second cohort selection criteria are as follows:

• Historical patient data for selected cohort from 2000

• No patients under the age of 18

• Full (including historical) records for patients with any of the following ICD-10 codes within any diagnosis position: Dilated cardiomyopathy (I42.0), Hypothyroidism (E03.9), Mitral Stenosis (I05.0, I34.2 OR Q23.2), Mixed Connective-Tissue Disease (M35.1), Obstructive Sleep Apnoea (G47.3), Systemic Lupus Erythematosus (M32), Portal Hypertension (K76.6), Pulmonic Stenosis (I37.0), Scleroderma (L94.0, L94.1 OR M43), Ischaemic heart diseases (I20-I25), Heart failure (I50), Pulmonary heart disease and diseases of pulmonary circulation (I26 – I28), Asthma (J45), COPD (J47 OR J40 - J44) and Interstitial lung disease (J84.9).

If a patient has any of the above ICD-10 codes the applicant would like to have the full longitudinal patient record. Due to the complicated disease area and goals of the research the patient pathway analysis requires a long period of data for the following reasons:

• Understanding impact of STHFT changes to service:

Previous work at STHFT has resulted in the improvement of the diagnostic process of pulmonary conditions. Firstly by streamlining the diagnostic process within STFHT to allow the majority of patient to be diagnosed within 2 consultations, secondly by continuing medical education outreach to satellite centres through talks and guideline publications. The historical length of data will allow the measurement of the impact of these improvements and support messaging to other specialist centres to allow them to adopt the learnings from these efforts, thus potentially improving diagnostic efforts and thus patient outcomes.

Furthermore the requested length of HES data aligns with the length of data held by STHFT allowing the applicant to utilise the full breadth of clinical data that STHFT hold.

• Having sufficient time to understand patient activity from onset of symptoms to diagnosis:

IQVIA Solutions UK Limited are requesting 2 ~15 year historical extract of data for the PAH project to cover both requested cohorts (patients who have attended SPDVU and cohort for development of the predictive algorithm).

The reason being that PAH patient populations (especially at subtype level) are very small and the diagnosis pathway is a long multi-year processes and often complex, the previous HES analysis showed that patients have a very high level of activity pre-diagnosis with >1/5th of patients experiencing hospitalisations, consultations or symptoms relating to IPAH disease >3 years before a positive diagnosis.

In addition the need to create a sophisticated algorithm that has the potential to perform well in the live clinical environment, a large sample of data is required. This is driven by the following reasons::

• Disease characteristics:

Cohort B was selected to try and ensure that the applicant adheres to data minimisation rules but also has enough data for meaningful analysis. The comparison group (cohort B) needs to be similar enough to the confirmed PAH cohort (cohort A), so the algorithm development process can start to identify the differences between patients who are often confused for PAH patients and those with a confirmed PAH diagnosis. PAH signs and symptoms are subtle and often confused with a range of different conditions. This means that the comparison group (cohort B) needs to be created from a sample of patients who share symptomology which is similar to PHA or occurs in conjunction with PAH disease. Minimising this data will lead to the development of a biased algorithm (For further information see the 180119_PAH Predictive algorithm overview- HES application Vf.dox).

• Refining the cohort based on clinical characteristics:

In order to select the most appropriate cohort of patients to act as a comparison group to confirmed PAH patients (cohort A), IQVIA Solutions UK Limited require to undergo analysis of the patient data, this is a data driven approach coupled with insights from the clinical specialists. As noted previously PAH patients are often misdiagnosed as other conditions due to the rarity of the disease and huge range of clinical manifestations they can present with. The aim is to identify a cohort of patients which do not have a confirmed diagnosis but share very similar clinical features, have contaminant diagnosis, visit the same specialists etc. This allows development of the algorithm on a comparison group as close to the real cases physicians experience in clinical practice as possible and thus stretch the algorithm as much as possible.

For example, in previous work, IQVIA created an algorithm to identify a rare disease population (Idiopathic Pulmonary Fibrosis), which manifests as a lung condition commonly misdiagnosed as asthma or COPD. To focus the algorithm on the clinical challenge IQVIA developed the algorithm to distinguish between IPF patients (8,574 patients) and those with COPD/Asthma (7.5m patients). In order to find the most appropriate comparison group to our confirmed PAH patients (cohort A) it requires a deep dive into the data to align the patient cohorts

• Refining the cohort based on availability of appropriate length of historic data:

The size of a cohort is limited not only by the number of patients with given diagnosis, but also the need to have available a sufficient time period both prior to the diagnosis (to observe baseline characteristics) as well as after the event (to observe relevant outcomes) for analysis. For example a recent project in Fabry Disease, one focus of analysis was to understand the diagnostic pathway, in order to identify any predictive signals/ markers which would allow earlier diagnosis of Fabry disease and thus slowing progression of the disease by allowing earlier treatment. The study by IQVIA identified 665 patients with suspected Fabry disease, of those patients only 90 patients had 3+ years of historical data available to allow analysis of the lead up to patient diagnosis (which was much shorter than desirable given the often 20 year symptom onset in this condition). This patient cohort size prevented IQVIA Solutions UK Limited from having sufficient numbers to conduct robust predictive analytics on the data to find signals/ markers of disease.

A recent study of idiopathic PAH (IPAH) patients found that a significant delay of 3.9 years from symptom onset to a diagnosis of IPAH (Strange et al. 2013). Indicating that a long time window is required and limiting that number of patients that will have the time window available for analysis.

• Bringing the algorithm to clinical practice:

If the algorithm were to be implemented in real clinical practice setting the algorithm can only run on patients who fit inclusion and exclusion criteria used to pull HES data. Therefore the narrower the patient sample requested means that the more limited real world sample that can be assessed for risk of disease. For example if IQVIA only requested a sample of HES data made up of male patients who are over 40 years old. This would mean that IQVIA could not expect the model to produce robust predictions for any female patients or patients under the age of 40.

Due to these reasons the applicant requires HES data for a longer period than the usual 5 year period routinely offered by NHS Digital in order to capture sufficient patients for the analysis.

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

As a result of initial analysis of the diagnostic treatment pathway for different subtypes of PAH, the following abstracts were accepted and presented at conferences:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” - accepted by ERS (European Respiratory Society) International Congress 2018. This abstract explained the methodology of predictive algorithm development and quantified the resulting algorithm’s performance.

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” – accepted by ATS (American Thoracic Society) International Conference 2018. This abstract described the process and rationale of linking HES data with specialist centre data and results from an exploration of the resulting dataset to conclude that sufficient patient numbers and depth of clinical information was available to support predictive algorithm development.

Building on the acceptance of these abstracts to publish findings in more detail:

• The following manuscript was published in the Journal of Pulmonary Circulation: “High levels of healthcare utilization prior to diagnosis in idiopathic pulmonary arterial hypertension support the feasibility of an early diagnosis algorithm: the SPHInX project”.

• The following manuscript has been submitted to both The Lancet and to Nature: “Utilising artificial intelligence to determine patients at risk of a rare disease: idiopathic pulmonary arterial hypertension”.

The following manuscripts are also under development:

• “Descriptive overview of the iPAH diagnostic process and pathway at SPVDU and impact of distance to specialist centre and social deprivation on diagnostic rates” (final title TBC) – targeted for the European Respiratory Journal. This manuscript will include:

o Overview of Sheffield’s approach to diagnosing iPAH patients after they enter the pulmonary unit

o Findings from an analysis of iPAH patient secondary care interventions as they approach diagnosis

• “Analysis of the diagnostic pathway for PH patients attending SPVDU” - final title and target journal TBC. This manuscript will include:

o A descriptive overview of the patient profiles of PH patients who have attended SPVDU (focussing on demographics and subtypes)

o Broader overview of SPVDU approach to diagnosing patients, focussing on other subtypes of PH patient who are seen at SPVDU

In addition to the current manuscript submissions and those in development, additional dissemination plans are as follows:

• Published results will be shared with the PHA UK patient advocacy group

• Abstracts and links to publications will be hosted on IQVIA Solutions UK Limited’s online bibliography which is publicly available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

An extension of the HES licence is required to both support any supplementary questions arising from publications and to support further publications in this area.

We note that in the current DSA, a data retention period of approximately 10 years (until 01/01/2027) was approved, to comply with NHS HRA and the MRC guidelines for data storage. It is recognised in the agreement that should this period exceed the term of the DSA, a new one will need to be put in place.

We have included below the “outputs expected” submitted with the original data access application, for context and reference. This includes details of the nature of outputs from predictive algorithm development.

*******************Below is the Outputs section from original application*****************

IQVIA Solutions UK Limited have conducted initial analysis of the diagnostic and treatment pathway for different subtypes, in addition to investigating predictive patient characteristics within the data to support the flagging of patients earlier within their diagnostic pathway via a predictive algorithm. The findings of both indicated that further refinement of results will be beneficial.

IQVIA Solutions UK Limited are in the process of publishing several findings, the following abstracts have been submitted:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

An extension of the HES licence is required to both support any supplementary questions arising from publications, in addition to supporting further publications in this area. The extension is also required for IQVIA and Sheffield to hold the data and all the work done to date while the PH National Audit data is still being onboarded to the DARS portal and not available to IQVIA and Sheffield.

In addition to the current abstract submissions, additional decimation plans are as follows:

• The applicant will submit the additional findings of the research to a peer review journal e.g. Thorax - BMJ Journals.

• The applicant will submit and present on findings at the 2018 ATS conference, in addition to other important pulmonary conferences, in order to further the knowledge of other specialist physicians

• Published results will be shared with the PHA UK patient advocacy group

• Furthermore the abstracts and links to publications will be hosted on IQVIA Solutions UK Limited’s online bibliography which is publically available

• Results will also be shared with other parties where appropriate e.g. Sharing results with other NHS trusts who also manage PAH patients or sharing with international centres which also diagnose and manage PAH patients

The current output of the algorithm generation is currently uncertain. However any implementation would need to be conducted by or with NHS bodies, because IQVIA are working with pseudonymous data and will not seek to re-identify patients at any stage. The nature of any implementation would need to be driven by the predictive sensitivity and specificity of the algorithm. In other words, the false positive and false negative detection rate. Implementing an algorithm with a high false positive rate would lead to many people tested with very few identified, conversely if the algorithm has a high false negative rate, it will likely miss many patients who should be tested for the disease. The health economics of the algorithm and any associated intervention would need to be carefully assessed prior to any implementation. Prior to any algorithm playing a role in supporting clinical practice / being implemented it will require peer review publication and broad acceptance before any uptake could be successful.

The algorithm will be free of charge and openly available. Access methods will be dependent on the strength of the algorithm but may include presentation at seminars, publications on risk factors or a clinical support tool provided directly to physicians (subject to any relevant approvals).

For an algorithm with weaker predictive potential IQVIA envisions the generation of publications in peer reviewed journals and generation of medical educational materials to use with clinical specialities who are potentially exposed to PAH patients. The literature will document the methodology used and the risk factors which would help to identify PAH patients earlier. These will potentially be presented at symposiums or other forums, depending on the findings.

If an algorithm with high predictive potential is generated, it could be used to create a clinical support tool for physicians to help diagnose patients, allowing the summarisation of large quantities of data in a more manageable format. This tool could support physicians by providing a risk score which they can interpret themselves to support clinician decisions.

If the applicant does not find any information of merit they will submit the methodology utilised in the research to a peer reviewed journal, this will allow other researchers to benefit from their research efforts. In addition the methodology will be shared via IQVIA Solutions UK Limited’s online bibliography and which is publically available.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide

This Data Sharing Agreement permits the retention of the data for an interim period.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated, and destruction of the data will be required.

The following information provides background information on the purpose of the original study. No new data will be released under this version of the agreement, and this agreement allows the applicant to hold data that has already been disseminated.

Several abstracts have been submitted and approved, with several more underway:

• “Real world data from hospital episode statistics can be used to determine patients at risk of idiopathic pulmonary arterial hypertension” submitted to ERS (European Respiratory Society) 2018

• “Development Of A Predictive Algorithm Based On Healthcare Behaviour To Support Earlier Diagnosis Of Idiopathic Pulmonary Arterial Hypertension: Results Of A Feasibility Study In The UK” submitted to ATS (American Thoracic Society) 2018

Results of analysis will be presented at the European Respiratory Society this year.

The initial predictive algorithm is indicating a large benefit to disease detection efforts and it is anticipated that the predictive algorithm, upon further refinement can be used to support screening programs. The algorithm is developed to flag the high risk patients who share characteristics in common with iPAH patients. iPAH disease prevalence is 5 in 1 million meaning that >20 million patients at random would have to be tested to find 100 patients with iPAH disease. Utilising the developed predictive algorithm on the HES data it was found that if the top 1000 patients flagged as “high risk” were analysed based on our algorithm patients would be detected with the disease.