Using System Dynamics in practice: a case study from emergency health services - PowerPoint PPT Presentation

Loading...

PPT – Using System Dynamics in practice: a case study from emergency health services PowerPoint presentation | free to download - id: 4b4bc7-ZmNiY



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Using System Dynamics in practice: a case study from emergency health services

Description:

Using System Dynamics in practice: a case study from emergency health services Sally Brailsford1, Valerie Lattimer2, PanayiotisTarnaras1 and Joanne Turnbull2 – PowerPoint PPT presentation

Number of Views:225
Avg rating:3.0/5.0
Slides: 51
Provided by: SCB54
Learn more at: http://mitacs.chcm.ubc.ca
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Using System Dynamics in practice: a case study from emergency health services


1
Using System Dynamics in practice a case study
from emergency health services
  • Sally Brailsford1, Valerie Lattimer2,
  • PanayiotisTarnaras1 and Joanne Turnbull2

1School of Management 2School of Nursing and
Midwifery University of Southampton, UK UBC
Centre for Health Care Management, 8 Dec 2006
2
Outline of talk
  • Brief background to the Nottingham Emergency Care
    / On Demand project
  • Using system dynamics qualitative and
    quantitative approaches
  • Our practical experiences
  • Patient preference study
  • Key results, implementation of findings, and
    conclusions

3
The city of Nottingham
  • Robin Hoods home town
  • City with population just under 650,000 in east
    Midlands of England
  • Mainly urban population with some areas of social
    deprivation

4
Health services in Nottingham
  • Two large NHS Trusts (i.e. hospitals)
  • Queens Medical Centre University teaching
    hospital, 1100 beds
  • Nottingham City Hospital 850 beds
  • One Accident Emergency (AE - the ER)
    department at QMC
  • 5 Primary Care Trusts, 350 GPs

5
Nottingham Health Authority
6
Queens Medical Centre, Nottingham
7
Background to the project
  • Increasing emergency hospital admissions in
    Nottingham (gt4 year on year increase since 1999)
  • Busiest (?) Accident Emergency Department in
    the country gt122,000 patients in 2000/01
  • Winter beds crises red alerts and ward
    closures
  • Pressure on staff stress, recruitment and
    retention problems
  • Steering Group set up in 2001 to develop Local
    Services Framework for unscheduled care
  • University of Southampton commissioned to provide
    research support to project

8
Membership of steering group
  • Clinicians and managers from hospitals (plus AE)
  • In-hours and out-of-hours GP services
  • Ambulance Service
  • Social Services
  • Mental Health Services
  • NHS Direct (integrated with out-of-hours GP
    service)
  • NHS Walk-in Centre
  • Patient representative groups
  • Community Health Council representatives

9
The Southampton research team
  • Val Lattimer, MRC Research Fellow, School of
    Nursing and Midwifery
  • Helen Smith, Reader in Primary Medical Care,
    Health Care Research Unit
  • Karen Gerard, health economist, HCRU
  • Steve George, Reader in Public Health Medicine,
    HCRU
  • Mike Clancy, AE Consultant, Southampton
    University Hospitals Trust
  • Me
  • Panayiotis Tarnaras and Jo Turnbull, RAs

10
Strands of the research
  • Literature review and comparison with other
    Health Authorities
  • Stakeholder interviews
  • Activity data collection
  • System dynamics modelling
  • Descriptive study of patient pathways
  • Patient survey and preference study

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
System Dynamics
  • Based on Jay Forresters Industrial Dynamics
    (1969)
  • Aim to analyse complex interacting systems
  • Principle structure determines behaviour
  • Qualitative aspect causal loop (influence)
    diagrams, to gain understanding of system
    behaviour
  • Quantitative aspect stock - flow models

17
Qualitative models influence diagrams

Student numbers
Staff stress levels

Research papers published
  • Link system constructs (real or abstract)
  • Identify feedback loops
  • Balancing loops have odd number of signs
  • Reinforcing loops or vicious circles have even
    number of signs

18
Feedback loop

Student numbers
Staff stress levels


Research papers published
Student recruitment


Reputation of university
19
A balancing loop

Student numbers
Staff stress levels


Research papers published

Student recruitment


Reputation of university
20
Behaviour over time
21
A balancing loop

Waiting lists
Hospital beds available


GP referral rate
22
A vicious circle

Extra Govt money


Patient demand

23
Pros cons of qualitative models
  • Can explore unanticipated side-effects, and
    identify performance indicators to flag up when
    these side-effects begin to be felt
  • Cannot tell which loops will dominate without
    quantifying effects can be difficult and
    subjective

24
Quantitative models
  • Need to quantify model parameters to tell which
    loops dominate, and when
  • Can suggest useful performance indicators even if
    numerical data is not available (e.g. staff
    stress levels)
  • Software Vensim, Stella (ithink)

25
Quantitative models stocks and flows
26
The underlying maths
  • Stock-flow equations ordinary differential
    equations, discretised as difference equations
    with finite timestep dt
  • Various solution methods used, in different
    software packages
  • Deterministic - simulation is not stochastic

27
Stella software
28
Why System Dynamics?
  • Huge, diverse, complex system
  • Many stakeholders with opposing viewpoints
  • Long timescale (5 years)
  • Hundreds of thousands of entities
  • Waiting times less important than process flows
  • Lack of accurate data in sufficient detail from
    some providers
  • Gaining insights more important than numerical
    predictions

29
Modelling phases
  • Qualitative stakeholder interviews and
    development of patient flow map influence
    diagramming used to focus discussion about
    specific subsystems
  • Quantitative Stella model, populated with 2000
    01 data, used to investigate (24) different
    scenarios, some suggested by Steering Group and
    others by us

30
Stakeholder interviews
  • Outline draft of patient pathways map derived in
    orientation visit (August 2001)
  • 30 interviews during Sept - Oct 2001
  • Respondents were asked
  • About own work area and areas of influence
  • To identify where they thought bottlenecks arose
  • To discuss factors which had shaped the system,
    and barriers to future development (local
    politics!)
  • To scribble on and amend the map where they
    thought we had got it wrong

31
Patient flow map
32
(No Transcript)
33
Data for the Stella model
  • Many problems obtaining data (!!!) especially,
    but not exclusively, in primary care
  • Used 2000-01 activity data for arrivals
  • Length of stay, and patient pathways within the
    hospitals, obtained from Dept of Health Hospital
    Episode Statistics data, patient surveys and from
    interviews with hospital staff
  • Internal validation by checking flow balances

34
Model validation baseline run
35
Using the Stella model
  • Regular trips to Nottingham to demonstrate the
    model as it evolved
  • Different people at each meeting!
  • No problems accepting continuous patient flows
    happy with SD technicalities once explained
  • Panel found the computer model fascinating and
    were keen to suggest scenarios to test

36
Experimental scenarios
  • Reconfigurations of services, e.g.
  • Longer opening hours for Walk-in Centre
  • Minor cases sent to WiC instead of AE
  • More step-down beds to reduce LoS
  • New services, e.g.
  • (Diagnostic and) Treatment Centre
  • Services targeted at specific patient groups

37
Scenario Areas
1 Increased admissions a) 4 growth in emergency admissions b) 3 growth in elective admissions
2 Changing front door demand
3 Reducing emergency admissions for specific groups of patients
4 Early discharge
5 Beds crisis ward closures (MRSA)
6 Streaming in AE (the ER)
38
Trust me, Im a computer
  • Wide spectrum of computer literacy and
    quantitative skills in the Steering Group panel
  • Stella model looked impressive because it was
    complicated
  • Clients warned not to over-interpret the numbers
  • Balance provided by couple of computer sceptics
    in the Steering Group

39
Main results from Stella model
  • Current rate of growth is not sustainable without
    extra resources up to 400 cancelled elective
    admissions per month after 5 years
  • High impact of relatively small changes
  • Alternatives to admission more effective than
    discharge management in reducing occupancy
  • Some benefits of moving less severe patients away
    from AE

40
Patient preference study
  • Discrete choice experiment (designed and led by
    health economist Karen Gerard)
  • Enable trade-offs between different aspects of
    service to be evaluated
  • Respondents - the users of emergency services (n
    378)
  • Patients also asked what factors influenced their
    choice of service on that particular day

41
Attributes to be compared
42
(No Transcript)
43
Main findings
  • Keep people informed!! Patients prepared to wait
    extra 86 minutes for better information
  • Younger patients (lt45) preferred doctor advice
    would trade for services located nearer home
    this was less important for older patients
  • Lack of interruptions important location less
    so
  • Potential need to tailor services for older
    patients, who are happier to accept treatment by
    specialist nurses and paramedics

44
Influence diagrams
  • Mainly used to focus panel discussion on specific
    issues arising from interviews and patient
    preference study, e.g.
  • Increased re-admission rates due to premature
    discharge
  • Effect of GPs sending patients to AE to
    queue-jump waiting lists for investigations
  • Patient behaviour due to long expected waits
  • Other behavioural effects stimulating demand by
    providing improved service?

45
Creating demand? - a feedback loop
46
Creating demand? - a feedback loop
47
Implementation
  • Results presented to Steering Group in May 2002
  • Stakeholder day at Nottingham Forest Football
    Club, June 2002
  • Local Services Framework
    developed and
    implemented by
    August 2002!

48
Pros and cons of SD
  • Excellent for studying interconnections between
    individual departments/providers and the wider
    health system
  • Very powerful tool giving global view of whole
    system
  • Loss of individual patient information and
    variability between individuals
  • Cannot produce highly detailed numerical results
  • Difficult to use for operational decision-making
    better for strategic policy-making

49
My personal view of using SD
  • Qualitative aspects were very useful (interviews,
    maps influence diagrams)
  • Stella model was compelling focus for stimulating
    discussion and ideas
  • Suspect that some people still fixated on the
    numbers despite all the health warnings
  • Some places where software was inadequate for
    modelling e.g. effects of variability, decision
    logic governing flows

50
References
  • S.C. Brailsford, V.A. Lattimer, P.Tarnaras and
    J.C. Turnbull, Emergency and On-Demand Health
    Care Modelling a Large Complex System, Journal
    of the Operational Research Society, 2004,
    5534-42.
  • V.A. Lattimer, S.C. Brailsford et al. Reviewing
    emergency care systems I insights from system
    dynamics modelling. Emerg Med J, 2004, 21685-691
  • K. Gerard, V.A. Lattimer, H. Smith, S.C.
    Brailsford et al. Reviewing emergency care
    systems II measuring patient preferences using a
    discrete choice experiment. Emerg Med J, 2004,
    21692697
About PowerShow.com