Research Theme — MOD

Missed Opportunities Detector

Theme leads:

Matt Sperrin — Lecturer in Health Data Science at the University of Manchester
Peter Diggle — Professor of Statistics and Associate Dean at Lancaster University
Ben Brown — Clinical Academic Fellow in Primary Care at the University of Manchester
Ben Bridgewater — Hon. Professor and Cardiac Surgeon at University Hospital of South Manchester

For any patient presenting with a myocardial infarction, the local health system might reasonably ask “Was there an opportunity missed to prevent or delay that event?”

The opportunity might have been in smoking cessation services, early detection in primary care, medication in blood pressure and lipid control, revascularisation procedures etc. Clinical guidelines and care pathways synthesise the evidence for best care in most long term conditions. So it should be possible to use linked health e-records to compare ‘expected care’ from guidelines with ‘observed care’ from analysis of the integrated records. The difference between observed and expected care may then indicate missed opportunities for intervention, and aspects of real world healthcare where evidence from trials might not generalise, for example in patients with comorbidities. Most clinical audits, however, focus on settings rather than whole pathways of care across different settings. This creates islands of data and analysis, compounding the lack of generalisability of evidence by applying it too narrowly to detect “missed opportunities”.

MOD focuses on the gap between idealised care and what happens in reality. Using routinely collected data from electronic health records the tool will establish missed opportunities, determine their significance and provide intelligence on how to avoid them in future. It will initially focus on cardiovascular disease with plans to encompass a range of conditions. MOD represents a new approach to healthcare quality improvement. By providing information in a cost-effective, timely and automated way it will support better targeted decision-making by policymakers, providers and commissioners. Patients and clinicians will also have a foundation for highlighting and helping to avoid missed opportunities before they arise.

Objectives:

  • To develop a health informatics methodology for detecting missed opportunities in routinely collected healthcare data.
  • To discover computable clinical intelligence methods that can improve health outcomes.
  • To develop informatics interventions that can improve clinical governance with new intelligence methods.

Research Theme — CoOP

Coproducing Observations with Patients

Theme leads:

Will Dixon — MRC Clinical Scientist and Rheumatologist at the University of Manchester
Kate Pickett — Professor of Inequalities in Health at the University of York
Shôn Lewis — Director of the Institute of Brain Behaviour and Mental Health at the University of Manchester

Electronic healthcare data provide an enormous opportunity for advancing knowledge about health and healthcare provision through research, audit, quality improvement and other activities.

Analysis of anonymised UK primary care records has contributed significantly to the understanding of medicines’ safety and effectiveness. However, the types and quality of available data limit the value of current and future research. Importantly, many data sources lack information collected directly from patients.

Only patients themselves experience directly the symptoms of their disease, the benefits and harms of treatment, or know how they feel about their condition. This research theme will focus on developing and testing methods to collect observations directly from patients. It will extend beyond the Department of Health’s 2015 goal of providing online access to personal medical records, by allowing patients to contribute information to inform clinical care and research. Collecting observations from the patient at the relevant time has huge potential to allow more accurate phenotyping, stratification, exposure and outcome measurement.

Objectives:

  • To develop tools for capturing electronic patient reported outcomes (e-PROs) in clinical settings and remotely from patients’ homes via web based applications, mobile phones and tablets.
  • To understand how patients interact with such tools.
  • To align e-PROs for NHS and research purposes, enabling data collected once to have multiple uses.
  • To validate e-PROs against gold standards.
  • To support the development of new e-PROs instruments.

Research Theme — SEA-3

Scalable Endotypes in Asthma, Allergies & Andrology

Theme Leads:

Adnan Custovic — Professor of Allergy at the University of Manchester
Chris Taylor — Professor of Imaging Science and Associate Vice President at the University of Manchester
Mattia Prosperi — Lecturer in Biomedical Modelling at the University of Manchester

Epidemiology is reaching the limit of what can be achieved through conventional hypothesis-driven research. For example, there is mounting evidence that asthma is not a single disease, but a condition comprising multiple distinct disease entities (endotypes), each with characteristic pathophysiology and risk factors. HeRC is seeking novel endotypes of asthma, allergies and cardiovascular disease by applying a combination of machine learning and biostatistical methods to major UK birth cohort studies, NHS casecohorts and linkable data sources.

We will capitalise on the unique collection of well characterised birth cohorts with harmonised clinical outcomes within the MRC-funded network (STELAR consortium) and Born in Bradford cohort. Adult asthma and chronic obstructive pulmonary disease will be studied across genotyped case cohorts in South Manchester and Salford (incorporating the Salford Lung Study). The linked data in Salford will include clinical details of exacerbations from a paperless hospital. We are creating a secure web-based research environment (Asthma e-Lab) to support consistent recording, description and sharing of data and emerging findings across all partners, thus enabling collaborative epidemiology in near-real-time. We will create and maintain the annotated dependency graphs of the problem space around the organising principles underlying asthma, and use a machine learning approach interactively over the combined datasets via Asthma e-Lab to discover the unbiased endotypes of asthma. Andrology will be studied using similar methodologies with the European Male Ageing Study and the English Longitudinal Study of Ageing. The socioeconomic data will be linked to healthcare utilisation reflected in the Hospital Episode Statistics, and GP data through Clinical Practice Research Datalink.

Objectives:

  • To combine world-leading expertise in birth cohorts, epidemiologically oriented health informatics research and statistical machine learning.
  • To use innovative computational statistical methods to identify novel latent endotypes of asthma, allergy and andrology.
  • To generate findings that can underpin may underpin new trials of prevention and treatment, personalised for specific endotypes and may help identify novel targets for the discovery of endotype specific stratified treatments.

Research Theme — DOT

Diabetes and Obesity Outcomes Translator

The Diabetes and Obesity Outcomes Translator (DOT) is conducting investigations at the crossroads of type 2 diabetes and cancer.

The cross-section between the two traditionally separate disease areas of cancer and type 2 diabetes (T2D) was cast into the spotlight five years ago when a series of controversial articles linked a popular medication used in the treatment of T2D with an increased risk of cancer.

Since this time, HeRC’s investigators, in partnership with the international diabetes research community have been directing effort in seeking to understand more about the relationship between the two commonly diagnosed diseases.

The health data scientists working on DOT have been conducting data-led research using huge patient databases that contain vast amounts of information. To understand the true nature of any relationship between cancer and T2D the team must firstly unpick and account for a number of influencing factors that could impact upon their findings.

These factors can include information like the medications that are prescribed to treat patients with T2D, lifestyle factors like smoking, exercise and weight and more general patient information like a person’s age and sex. For the study team to understand the root cause of these complex diseases they need to apply statistical methods that account for the influence and impact of these factors on the available data. By doing this the team can be sure that their findings offer sound scientific evidence.

The linked patient data that is used by the researchers working on the DOT theme comes from a number of sources including:

All of the information used by the research team is anonymous with personal data like names and dates of birth removed.

By linking together these different sources of information, the investigators are able to create a more complete picture of any relationships that exist between the two diseases; the more real-world data that is available, the more confidently and accurately the team can state their findings.

The overall aim of the DOT research theme is to help people living with or at risk of developing T2D. Any findings could lead to the earlier screening and detection of certain cancer types. Findings might also lead to changes in the advice that doctors are able to give to patients regarding disease management and ultimately could help to reduce a patient’s risk of developing cancer.

Research projects taking place as part of the DOT theme feed into two bigger international research consortiums that help increase the range, scope and impact of their research:

  • Diabetes and Cancer Research Consortium (DCRC) – demystifying the range of conflicting evidence around T2D medications and their supposed risks and protection of certain cancer types
  • Canadian Institutes of Health Research (CIHR) – Understanding more about electronic patient records can be used to observe trends and outcomes in in the supply of medications

Examples of research conducted as part of the DOT research theme to date include:

  • Case study 1, A data-driven approach to understanding the relationship between diabetes, cancer and obesity
  • Case study 2, Diabetes and cancer – an update following research controversy

If you would like further information about the DOT theme or to contact a member of the team please click here.

Research Theme — FIN

Trials Feasibility Improvement Network

Theme leads:

Paula Williamson — Professor of Medical Statistics and Director of the MRC NW Hub for Trials Methodology Research at the University of Liverpool
Martin Gibson — Hon. Professor and Diabetologist/Endocrinologist, Director of Greater Manchester Comprehensive Local Research Network and CEO of North West e-Health

This research aims to harness advanced health informatics and electronic health record data to improve the design and conduct of clinical trials.

It will address the problem that most feasibility assessments for clinical trials are unreliable, with subsequent under-recruitment, poor retention and time over-runs. It will enhance recruitment feasibility assessments, optimise data collection and establish a coordinated approach to data repositories. This will also address the problem of outcomes being inconsistently measured across different trials or centres, particularly in open label real-world study designs. Strategic aims include the integration of MRC trials methodology and health informatics activities in this area, and the close working of methodologists with practitioners for maximum impact.

Objectives:

Enhance recruitment feasibility assessments:

  • We will investigate a variety of systems and available data for their ability to improve current approaches to recruitment feasibility assessments, including Clinical Practice Research Datalink, NIHR portfolio, and FARSITE (Feasability Assessment and Recruitment System for Improving Trial Efficiency).
  • We will develop a framework for evaluating the health informatics support of feasibility assessment and recruitment, and apply it to a number of trials including the Salford Lung Study.
  • FARSITE results will be compared across different localities to find systematic errors, such as miscoding, that could make feasibility assessments inconsistent.
  • We will search for potential instrumental variables in modelling recruitment: for example time since the introduction of an incentive payment for registering. We will also explore seeking patients’ consent for researchers to contact them as part of the consent process for online GP record access.

Optimise data collection:

  • We will evaluate the practicality of using routinely collected clinical data in trials, exploring the hypothesis that trials would be easier to run if the main outcomes are collected routinely.
  • We will develop a framework for evaluating the readiness of electronic health records for capturing core outcome measures and feed this into the MRC Core Outcome Measures in Effectiveness Trials (COMET) initiative.
  • The use of social media in patient engagement in clinical studies will be explored in terms of acceptability, feasibility and effectiveness.

Research Theme — CHIP-SET

Community Health Intelligence Partnership Semantic Epidemiology Toolkit

Theme Leads:

John Ainsworth – Senior Research Fellow in Health Informatics at the University of Manchester
Carole Goble – Professor of Computer Science at the University of Manchester
Goran Nenadic – Senior Lecturer in Computer Science at the University of Manchester

HeRC’s advanced methodology for harnessing linkable data, methods and expertise will be delivered through a new online toolkit, CHIP-SET.

CHIP-SET will extend the e-Lab concept and software to bring investigators together in a secure on-line environment. The software will codify methodology making it easier to use data through semantic technologies. By consolidating the best of research and existing data warehouses into a single coherent role and embedding methodology it will create distributed sense making of data available maximising its usability.

This will enable users to find, share and reuse the ingredients of real-world healthcare evidence to enable research needs and build information platforms.

Objectives:

  • To deliver the methodological advances through software.
  • To link bio-statistical and machine learning methodology in a unified digital laboratory (e-Lab).
  • To deploy a network of e-Labs across NHS North localities to cover a 5 million population within 5 years.