Towards a National Learning Health System - Aziz Sheikh

Towards a National Learning Health System

Aziz Sheikh OBE

BSc, MBBS, MD, MSc, FRCGP, FRCP, FRCPE, FFPH, FRSE, FMedSci, FACMI

Professor of Primary Care Research & Development and

Co-Director, Centre for Medical Informatics, The University of Edinburgh

Director, Asthma UK Centre for Applied Research

Visiting Professor of Medicine, Brigham and Women’s Hospital/Harvard Medical School

Birmingham, 9th June 2016

@DrAzizSheikh

Disclosures • Research funding from:

– Agency for Healthcare Research and Quality – Asthma UK – British Lung Foundation – Chief Scientist's Office of the Scottish Government – Department of Health – Digital Health Institute – Horizon 2020 – Medical Research Council – National Institute of Health Research Applied Programme Grants – NHS Connecting for Health Evaluation Programme – The Commonwealth Fund – Wellcome Trust – World Health Organization

• Submissions with decisions pending to: – Engineering and Physical Sciences Research Council – National Environmental Research Council – Robert Wood Johnson Foundation

Overview

• Current models of healthcare are fatally flawed • Health systems need to be digitised and the potential of

digital data needs to be unlocked • Examples of using these digitised datasets to:

– Undertake epidemiological investigations – Support evaluation of policy interventions – Increase the clinical efficiency of trials

• The need to move from the current ad-hoc bespoke researcher-led efforts to a ‘Learning Healthcare System’…

• Building a prototype national asthma learning healthcare system

• Looking ahead: From a Learning Healthcare System to a Learning Health System

THE BURNING PLATFORM…

Major challenges facing health systems internationally

• Changing demographics: ageing populations

• Increasing numbers of people living with long-term conditions

• Ongoing concerns about the safety and quality of healthcare

• Spiralling healthcare costs

Increasing UK life expectancy

The demographic time bomb: Forecasts for dependency ratios

The exponential rise of multi-morbidity

NHS Connecting for Health Evaluation Programme

HEALTH IT IS NOT A ‘SILVER BULLET’

How best to respond?

• Increase taxation: direct and indirect

• Increase retirement age

• Modify pension plans: average salary schemes

• Encourage immigration

• Cut expenditure on public services

• All are however deeply unpopular and make politicians very wary…

Possible solutions

The UK’s National Programme for IT

• Considerable policy interest in Health IT as being the answer

• 1998: “If I live in Bradford and fall ill in Birmingham then I want the doctor treating me to have access to the information he needs to treat me.” (Rt. Hon. Tony Blair, NHS Conference, London, July 2, 1998)

• 2002: £12billion ‘vision’ for the National Programme for IT approved by Tony Blair at an un-minuted 10-minute briefing in Downing Street with Bill Gates

18 |

Systems Optimisation: Turning data into information

Antibiotic - % Missed Doses

Date Intervention

A 15 April 2009 Pause function for doctors

B 04 August 2009 Missed Doses go live on clinical dashboard

C 15 December

2009

Introduction of coloured indicators to show due /

overdue drugs

D

* 24 February

2010

NPSA Rapid Response Alert

D

* 30 March 2010 Chief Executive Missed Dose Root Cause Analysis

meetings

Step change in % missed doses when information shared with clinicians / managers

Further highly significant change when CEO started RCA meetings

Coleman et al. Missed medication doses in hospitalised patients: a descriptive account of quality improvement measures and time series analysis. Int J Qual Health Care. 2013 Oct;25(5):564-72.

UNLOCKING THE POTENTIAL OF EHR-DERIVED DATA

Digital infrastructure

EPIDEMIOLOGICAL STUDIES

Centre for Medical Informatics, The University of Edinburgh

Observational studies using hospital data:

Hospital Episodes Statistics

Trends in hospital discharges for anaphylaxis, 1991-1995

Alves B, et al.

BMJ 2000; 320; 1441


Observational studies using hospital data:

Hospital Episodes Statistics

Trends in discharge rates for systemic allergic disorders, 1990-2001

Gupta R, et al.

BMJ 2003; 327: 1142-

43

GP and nurse consultation rates by sex in those with and without eczema

Simpson CR, et al. JRSM.

2009; 102:108-17

Observational studies using GP data:

QResearch


Constructing birth cohorts: investigating

“the allergic march” in the

General Practice Research Database

+ asthma (15.4% also had rhinitis)

+ rhinitis (10.0% also had asthma) n=7608

+ rhinitis (10.0% also had eczema)

+ eczema(11.9% also had rhinitis) n=3567

+ eczema (12.2% also had asthma)

+ asthma(10.9% also had eczema) n=1316

Describing numerous variants of “the

allergic march” in GPRD birth cohort

24,112

patients

Punekar Y, et al.

Clin Exp Allergy

2009;39:1889-95

Increase in lifetime prevalence rate of COPD in England, 2001- 2005

>5%

4-5%

3-4%

<3%

% increase

Mapping changing COPD prevalence

Simpson C, et al. BJGP

2010;60: 277-84

Investigating the relationship between asthma and exam performance

Sturdy P, et al. PLOS One 2012; 7:e43977

www.qrisk.org/

Bhopal R, et al. Eur J

Pub Hlth 2015

Sheikh A, et al. BMC Medicine (invited resubmission)

QUASI-EXPERIMENTAL STUDIES


Investigating the impact of the Low Emission Zone on asthma

Griffiths C, et al.

(submitted)

9,536,003

patient-years

GP data

(1997-2012) July 2007

April 2007

April 2007

Maarch 2006

Impact of the smoking ban on hospitalisations for respiratory tract infections in children

Been J, et al.

ERJ 2015


Vaccine effectiveness

in pandemic influenza

Preparing for future pandemics…

Core funding in place and

release of additional funds at

first signs of pandemic

influenza

This has enabled:

o Creation of data structures

to permit real-time

evaluations

o Permissions and approvals

for data linkage and

analysis

o Development of detailed

analysis and reporting plans

SUPPORTING CLINICAL TRIALS

t+asthma Abingdon,Oxford

CYMPLA trial

Password protected website

Pinnock et al. BMJ 2012

Supporting recruitment

32 practices

(311,926 patients)

Computer searches: 13,101 potentially eligible

1,020 excluded by practice

12,081 postal

invitations

393 eligible and

first visit booked

Expressions of interest: 1,016

623 excluded at pre-screening telephone call

• 470 too well controlled (ACQ<1.5)

• 124 phone/network incompatible

• 29 ‘other’

Attended baseline assessment : 346

58 excluded at baseline assessment

• 37 too well controlled (ACQ<1.5)

• 11 declined

• 10 ‘other’

288 randomised

Improving prescribing safety

Base-line

Results +

Evidence +

Consent

letters

Initial meeting

During this meeting I would like to feed back the results of the

searches…..

6 & 12 months

Action plan

Actions recorded

GP practice My computer

Simple feedback Pharmacist intervention (2 days per week for 12 weeks)

+

“Exit” meeting

Data-base

FTP

FTP

Centre for Medical Infrmatics, The University of Edinburgh

Share_no_titles.mp4

In summary…

SCALING-UP EFFORTSD: THE FARR INSTITUTE

• Develop UK Health Informatics Research Network Strategy.

• Provide a blueprint for the Network activities which are designed to harness expertise and engage stakeholders for the coming five years and beyond.

The Farr Institute vision

“To harness health data for patient and

public benefit by setting the

international standard for the safe and

secure use of electronic patient

records and other population-based

datasets for research purposes”

Our 10 key activities

1. Collaborative leadership. 6. Enabling datasets

2. Cutting-edge research 7. Harmonized e-infrastructure

3. Public engagement. 8. Industrial partnerships

4. Governance (‘safe havens’) 9. Training and capacity building

5. Methods development 10. Communications

To deliver impact nationally and internationally

eHealth Research Group, The University of Edinburgh

Looking ahead: Integration of EHR data

with biomedical data to support

personalised medicine…

• Genetics

• Omics

• Imaging

• Phenotypes

Enabling administrative

and social data

Phase 1: Administrative data

Phase 2: Business data

Phase 3: Voluntary sector and

social media data

WE NEED TO MOVE FROM THE CURRENT AD-HOC ARRANGEMENTS TO A LEARNING HEALTH CARE SYSTEM…

The idea of the LHS builds on two era-defining publications…

What is a Learning Healthcare System?

The Institute of Medicine has defined this as a healthcare system:

• ‘that is designed to generate and apply the best evidence for the collaborative healthcare choices of each patient and provider;

• to drive the process of discovery as a natural outgrowth of patient care;

• and to ensure innovation, quality, safety, and value in health care.’

Engineering new models of health care

BUILDING A PROTOTYPE NATIONAL ASTHMA LEARNING HEALTH CARE SYSTEM

6

7

Our Vision

To create a world-class centre and

associated UK network that will:

Reduce

asthma

hospital

admissions

Improve

asthma

control

Reduce asthma deaths

UK Asthma Observatory Platform Framework Goals

• To identify and utilise relevant data on asthma across the UK in order to create a UK-wide repertoire:

– For interactive monitoring of

real-time estimates of the burden of asthma

– As a hub for the various AUKCAR research and policy outputs

– Repository for AUKCAR research data

– Other asthma activities in the UK

7

3

Number of UK inpatient episodes with

asthma as the primary diagnosis

7

4 0

20,000

40,000

60,000

80,000

100,000

120,000

2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12

Number of UK deaths with asthma as the underlying condition

7

5

0

200

400

600

800

1000

1200

1400

1600

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Improving

asthma

outcomes

Engaging academics, physicians and patients

AUKCAR/Farr Frontiers Asthma Meeting, Edinburgh, 27th October 2015

Asthma patients, clinicians, policy makers, asthma charities, pharma industry, university researchers…

“You cannot

know too

much about

asthma…”

Michael Bang,

Patient Representative

AUKCAR Advisory Group

.

Potential Low-hanging Targets

Some other immediate priorities…

Each can increase the risk of asthma exacerbations

and current practice is ***highly variable***

Pharma

81

Patients Payers

Researchers

Government /Public Health

Local Healthcare Delivery Systems (Patient journey)

Research Institutes Biomedical Data

Tech Industry Devices

Network requirement: Coherent analysis by heterogeneous source data transformation

COHERENT RESPONSE

UK LHS DATA MODEL

RESEARCH QUESTION/QUERY

TRUSTWORTHY RESEARCH

ENVIRONMENTS

3rd Sector

LOCAL DATA REQUIRES

TRANSFORMATION TO A COMMON

MODEL

& DISSEMINATION AT SCALE

LOCAL STAKEHOLDER

DATA IS HETEROGENEOUS

Network requirement: Real-time bidirectional flow

using common interface to/from stakeholder systems

CHI (Community Health Index)

EXAMPLE PATIENT JOURNEY DATA SOURCES

HEALTH

OUTCOME

S

(including

patient

reported)

Compute Infrastructure

Analysis

LEARNING

CYCLE

UK LHS Data Model

FARR

SCOTLAND

SAFE

HAVEN

Linkage

TRUSTWORTHY RESEARCH

ENVIRONMENTS

COMMON

INTERFACE

Working with the Farr @Scotland Safe Haven

FUNCTIONS

• Data linkage

• Anonymisation (or at source)

• Pseudonymisation (or at source)

• Study management for

medical/clinical

social research

• Analytical services

• Specialist compute services

Data from Scotland’s

Source Systems via Contract reporting (ESCRO,

Structured, Flat-file, Imaging)

Standard interfaces

with other Farr TRE’s

Common data model

to develop interoperability

and cross-site search

Hardware/software evaluation to

fit use cases or research questions

Trustworthy Research Environment TRUSTWORTHY RESEARCH

ENVIRONMENT

Analysis

Data Enclave with Credentialed Access

The scalability challenge

2-7m Patient Population Networks

Agreed patient benefit use cases

can drive and synchronize cross-site

work The power of scale - cannot achieve

clinical goals in single or few centres

Cross-site interoperability can be

achieved through Farr Infrastructure

group coordination

Its non-trivial, no complacency, but…

LOOKING AHEAD…

From ‘Learning Healthcare System’ to ‘Learning Health System’

The ‘triple aim’

Macro-level

• Thinking about health cross-sectorally

• Giving policymakers the tools and information they need to support decision making

– Burden of disease estimates

– Considering options and modelling their impact

– Prioritisation exercises for candidate interventions

– Programmatic evaluations of the impact of policy interventions

Meso-level

Members of the National Advisory Group on Health Information Technology in England

• Robert Wachter (Chair)

• Julia Adler-Milstein

• David Brailer

• Sir David Dalton

• Dave deBronkart “e-Patient Dave”

• Mary Dixon-Woods

• Rollin (Terry) Fairbanks

• John Halamka

• Crispin Hebron

• Tim Kelsey

• Richard Lilford

• Christian Nohr

• Aziz Sheikh

• Christine Sinsky

• Ann Slee

• Lynda Thomas

• Wai Keong Wong

• Harpreet Sood

Micro-level

Conclusions

• Healthcare needs to be reengineered

• New models will need to be more patient-centred, focused on preventive and ambulatory care, and aligned to the needs of the very large sections of our population now living with long-term non-communicable disorders

• The concepts of a ‘Learning Healthcare System’ and in particular a ‘Learning Health System’ provides a framework to begin to conceptualise future health systems

• Building such systems rank amongst the greatest challenges of the 21st Century…

Towards a National Learning Health System

Aziz Sheikh OBE

BSc, MBBS, MD, MSc, FRCGP, FRCP, FRCPE, FFPH, FRSE, FMedSci, FACMI

Professor of Primary Care Research & Development and

Co-Director, Centre for Medical Informatics, The University of Edinburgh

Director, Asthma UK Centre for Applied Research

Visiting Professor of Medicine, Brigham and Women’s Hospital/Harvard Medical School

Birmingham, 9th June 2016

@DrAzizSheikh