Information - Yorkshire & Humber Academic Health Science

Report
The innovative use of information to achieve
the three AHSN goals
Dr Jeremy Wyatt DM FRCP
Leadership chair in eHealth Research (Health Informatics)
Yorkshire Centre for Health Informatics,
Leeds Institute of Health Sciences
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
@CLAHRC_LYB
www.yhahsn.org.uk
www.clahrc-sy.nihr.ac.uk
@YHANSNP
@CLAHRC_SY
#AHSN Event
@AHSN_YandH
@CLAHRC_SY
#AHSN
Event
ww.yhahsn.org.uk
eHealth / health informatics
“Use of information & communications technologies
to support & improve the delivery of health, social &
self care”
Focus is on:
– Information - both data and knowledge
– Decisions of clinicians, patients, public…
– Communication: appropriate messages,
channels and formats
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
www.yhahsn.org.uk
@AHSN_YandH
www.clahrc-sy.nihr.ac.uk
@CLAHRC_SY
#AHSN Event
What is information ?
Information: “organised data and knowledge used to
support decisions and actions”
Shortliffe EH: Textbook of Medical Informatics 1st edition, 1990
Data: the specifics of a case / patient - captured in
records
Knowledge: generic information that applies across
cases - captured in books / websites / guidelines...
Wyatt & Sullivan, BMJ 2005
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
www.yhahsn.org.uk
@AHSN_YandH
www.clahrc-sy.nihr.ac.uk
@CLAHRC_SY
#AHSN Event
Information cycles in healthcare
Assemble evidence
Apply knowledge
Knowledge
Evidence based health & behaviour change
Learn & apply lessons
Retrieve data
Clinical audit / CQI /
research cycle
Patient / self care
Insights, evidence
Clinical practice,
self care
Capture data
Analyse records
Records
www.clahrc-lyb.nihr.ac.uk
www.yhahsn.org.uk
www.clahrc-sy.nihr.ac.uk
AHSN Information theme
Aim: to ensure that robust, comprehensive information and
evidence are at the heart of decision making
Objectives:
1. Ensure accurate, timely information is delivered to every
point of need
2. Improve integration of health databases across sectors,
building on existing strengths
3. Bring latest developments in big data, cloud computing and
data modelling to healthcare frontline
4. Give health professionals access to analytical / reporting skills
ww.yhahsn.org.uk
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
www.yhahsn.org.uk
@AHSN_YandH
www.clahrc-sy.nihr.ac.uk
@CLAHRC_SY
#AHSN Event
Questions for table discussion
• What are the current information challenges
in transforming health care?
• How can research contribute to addressing
these challenges?
Health e-Research Centre
(HeRC) Update
Kate Pickett on behalf of HeRC Consortium
Leeds, 5th Mar 2013
HeRC Mission
Datasets
Link
Value
Science and Industry
(R&D)
Link
Ingredients
Insights
Data Quality
Improved Care for
Patients and Communities
(Service)
Methods
Experts
Delivering improved care for patients and communities
through large-scale sense-making methodology reusing health data
HeRC Research Themes
• CoOP
– “Coproducing observation with patients”
• MOD
– “Missed opportunities detector”
• SEA-3
– “Scalable endotypes of asthma, allergies and andrology”
• DOT
– Diabesity outcomes translator
• FIN
– Trials feasibility improvement network
HeRC Operations
Steering Group
Director + Management Team
PPI: (ethics) + (communities)
Manchester
Psychology
Lancaster
Statistics
Manchester
& Microsoft
Machine
Learning
Manchester
Biostatistics
Liverpool
MRC hub
Trials
Methods
Methods
 Health Informatics 
Manchester
Greater
Pharmaco Manchester
NHS trials
Epidemiol.
Real-world
Problems
and data
e-Research
FIN
SEA-3
Manchester
Clinical
Epidemiol.
York
Social Epidemiol.
DOT
Liverpool
Public
Health
MOD
Manchester
PROMS
& ARUK
Clinical
Epidemiol.
CoOP
Training
CHIP-SET software tools (and generic platform)
Using Data Linkage to assess the extent of health inequalities
and generate data informing a targeted intervention:
Maternal mental health example
•
•
Parental depression can have a profound impact on children’s health, wellbeing and
social development
Problem: Ethnic minority women have a higher rate of depression than white, are just as
likely to access care, but less likely to be diagnosed and therefore treated, with
consequences for the children
•
Understand characteristics of target sample
– What proportion of the variation is due to
•
•
•
•
•
•
•
•
Non-attendance
Variation in presentation of symptoms
Coding practices
Treatment uptake
Outcome variation
Comparison with cohort measures (demography, outcomes for mother & child)
Clustered by area (practice)? Area (geography?) Ethnicity?
Explore solutions
– Area-based
•
•
GP practice
Geographical barriers to care
– Treatment based
•
•
Acceptability of treatments
Outcome variation for different groups
– Target individual packages for those at the tail end of the distribution, or
– Population shifts in health seeking behaviour?
Maternal mental health example
Data linkage, primary care mental health:
Who comes?
Demographics
Spatial
For what?
Coding of
complaint
Dx. Coding
Treatment
F/U
Depression
Physical
Differences
in tx options?
Quantity
Coding
Outcome
Cohort measures: Mental health, children’s mental health
Extra data collection: e.g. GP practice characteristics, interviews?
Driving Research Evidence into
Practice
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
@CLAHRC_LYB
www.yhahsn.org.uk
www.clahrc-sy.nihr.ac.uk
@YHANSNP
@CLAHRC_SY
#AHSN Event
@AHSN_YandH
@CLAHRC_SY
#AHSN
Event
ww.yhahsn.org.uk
LYBRA - CLAHRC
Using patient data for service
improvement: an example from stroke
John Young
Professor of Elderly Care Medicine Head
Academic Unit of Elderly Care and Rehabilitation
Bradford Hospitals Trust and University of Leeds
on behalf of CIMSS team
LYBRA - CLAHRC
Clinical Information and Management
System for Stroke (CIMSS)
A novel IT supported approach to improving
stroke care through the collection of high
quality data as part of routine care
Calderdale; Leeds; Bradford; Airedale
and Harrogate
LYBRA - CLAHRC
CIMSS PUBLICATION TRAIL
A review of stroke outcome measures valid and
reliable for administration by postal survey
(Reviews in Clinical Gerontology 2010)
Frenchay Activities Index
Subjective Index of Physical &
Social Outcomes
Euro QoL
A systematic review of case-mix adjustment
models for stroke
(Clinical Rehabilitation 2012)
Six Simple Variables Model
(but predicts mortality)
Predicting patient reported outcomes: a
validation of the Six Simple Variable Model
(Cerebrovascular Diseases)
The SSV model can severity
adjust the SIPSO measure
Confirmation of the validity of a two-scale
structure for the SIPSO
(Archives of the Phys Med & Rehabilitation)
The two sub-scale structure
(physical & social outcomes)
is confirmed
LYBRA - CLAHRC
CIMSS PUBLICATION TRAIL
A cacophony of clinical datasets:
the example of stroke
(Geriatric Medicine)
Overlapping (but different) indicators
Data dictionary approach
A point of care electronic stroke data
collection system
(Health Technology 2013)
NHS IT climate is disjointed and
fragmented
“One size fits all” not appropriate
Linking existing systems useful
15 stage IT development plan developed
Agile development of an electronic data
collection system for stroke
(BMC Medicine)
How CIMSS was developed
Source codes
(The role of Diffusion Fellows in Service
Improvement)
CLAHRC Diffusion Fellow
LYBRA - CLAHRC
What have we learned?
Successful research service improvements based
on information innovation requires:
1. Valid PROM (or PREM) ± process measures
2. Severity adjustment approaches
3. Existing IT system > bespoke
4. Mechanisms for behaviour change
Driving Research Evidence into
Practice
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
@CLAHRC_LYB
www.yhahsn.org.uk
www.clahrc-sy.nihr.ac.uk
@YHANSNP
@CLAHRC_SY
#AHSN Event
@AHSN_YandH
@CLAHRC_SY
#AHSN
Event
ww.yhahsn.org.uk
information absent from
implementation/transformation
decision making: Part II - preferences
Carl Thompson
Professor
TRiP-LaB, University of York
Preferences: “Silent misdiagnosis”?
(Mulley et al. Kings Fund 2012)
Health care may be the only industry in which giving customers
what they really want would save money. Well-informed patients
consume less medicine… much less.
Wanless* estimated the potential annual savings at £30 billion,
or 16 per cent of the projected budget by 2022 (Wanless 2002).
* Based on maximum patient engagement
“Securing engagement”:
the defaults…
Aims
Elicit public
preferences
for
innovations
Elicit WTPs
for
innovations
Investigate
similarities and
differences among
the respondents
Methods
Discrete Choice Experiment with latent class modelling
Innovations viewed as a “bundle” of characteristics (cost, evidence
base, target groups, time to implementation….)
online and paper-pen surveys in West Yorkshire, UK in 2011
stratified random sampling of 3600 people using “Electoral Roll”
Register + Bradford NHS Foundation Trust membership list
Public Voice in Health Service Innovation Investment Decision: A Discrete Choice Experiment
The discrete choice
Results
 3 Latent Classes:
Class-1 (57%), Class-2 (25%), Class-3 (18%)
 Everyone prefers
• Implementing innovations to not.
• ‘scientifically’ proven, relatively cheap, innovations with clear health
benefits, and are quick to implement.
 people are unwilling to pay for innovations
• that are scientifically unproven, take ‘more than a year’ to implement,
and result in only ‘moderate’ health benefits.
• And those targeting ‘drug users’, ‘obese people’, and the ‘elderly’: the
“unpopular”
 The differences…
Results
 Class-1 (57%): Value ‘health gain’, less sensitive to costs, and like
innovations targeting people with cancer. Science and expert opinion
valued more than others; more likely to be satisfied with the quality of
their health care.
 Class-2 (25%): dislike spending on ‘unpopular’ groups. Willing to pay
twice for ‘best’ health (100%) than ‘good’ health (50%), and do not value
the speed of implementation. more likely to be male, full-time employed,
and less satisfied with the quality of health care services available to them.
 Class-3 (18%): accepting of ‘unpopular’ target groups. believe that
decisions on the prioritisation of innovation options should not be based
on the age and time-to-implementation.
Public &
Patient
Involvement
Involving the public: beyond the focus group
And for the AHSN?
• Chance to design services that reflect public
preferences
– realise some of Wanless’ £30 billion budget impact?
• improving the sampling, response rate, attribute
“bundle” size (applicability).
– MRC methods grant still in the game
• Prioritisation decision support
– more innovation than you can fund: how do you
choose?
• What are the current challenges
in using information?
• How can research contribute to
addressing these challenges?
www.clahrc-lyb.nihr.ac.uk
@CLAHRC_LYB
@CLAHRC_LYB
www.yhahsn.org.uk
www.clahrc-sy.nihr.ac.uk
@YHANSNP
@CLAHRC_SY
#AHSN Event
@AHSN_YandH
@CLAHRC_SY
#AHSN
Event
ww.yhahsn.org.uk

similar documents