introduction to decision analytic modelling · 2019-04-05 · modeling approaches •technique...
TRANSCRIPT
1
CAPHRI School for Public Health and Primary Care
Introduction to decision analytic modelling
Prof. Dr. Manuela Joore Dept. Clinical Epidemiology & MTA, Maastricht UMC+, [email protected]
EUHEA pre-conference workshop July 11th 2018
2
CAPHRI School for Public Health and Primary Care
Outline • Decision analytic modelling for health
economic evaluation: – What is it & why use it?
– Types/Approaches
– Components
– Analyses
3
CAPHRI School for Public Health and Primary Care
What is a decision analytic model? (for health economic evaluation)
4
CAPHRI School for Public Health and Primary Care
What is a definition of ‘model’?
5
CAPHRI School for Public Health and Primary Care
What is a definition of ‘model’?
A model is a simplified reflection
of reality
6
CAPHRI School for Public Health and Primary Care
Can you think of examples of models?
7
CAPHRI School for Public Health and Primary Care
8
CAPHRI School for Public Health and Primary Care Δ
€
Δ QALYs
Health economic evaluation: two approaches
9
CAPHRI School for Public Health and Primary Care Δ
€
Δ QALYs
Health economic evaluation: two approaches
R R
Intervention A
Intervention B
€, QALYs
€, QALYs
Mo
de
l Tri
al
Mild Severe
Death
10
CAPHRI School for Public Health and Primary Care
Why model?
11
CAPHRI School for Public Health and Primary Care
Policy decisions
Scientific knowledge
Health economic decision modelling
RCTs, cohort studies / registries, surveys, hospital information, price information, expert opinion...
Is A cost-effective compared to B (C, D, …)
12
CAPHRI School for Public Health and Primary Care
Model based economic evaluation: Why?
1. Extending results from a trial = extrapolation
2. Combining / synthesizing available evidence
3. Generalising trials into practice (location, population, …)
4. Modelling prior to trials and studies
5. Modelling to address uncertainty
population S
?
13
CAPHRI School for Public Health and Primary Care
Components in modelling
14
CAPHRI School for Public Health and Primary Care
Components in modelling (see ISPOR/SMDM taskforce papers, Value in Health 2012 vol 15: p 796 etc.)
1. Conceptualising a model
2. Specific modeling techniques
3. Parametrizing a model
4. Handling uncertainty
5. Model validation & transparency
15
CAPHRI School for Public Health and Primary Care
1. Conceptualising a model (ISPOR/SMDMD taskforce 2)
16
CAPHRI School for Public Health and Primary Care
1. Conceptualising a model
1. Model should match the decision, not vice versa
2. Specify the decision problem
– population, intervention, comparators, outcomes, time horizon, perspective, audience)
3. Conceptual structure of the model driven by decision problem not by data availability
4. Conceptual model should be able to indentify key uncertainties
17
CAPHRI School for Public Health and Primary Care
Components in modelling (see ISPOR/SMDM taskforce papers, Value in Health 2012 vol 15: p 796 etc.)
1. Conceptualising a model
2. Specific modeling techniques
3. Parametrizing a model
4. Handling uncertainty
5. Model validation & transparency
18
CAPHRI School for Public Health and Primary Care
Modeling approaches
19
CAPHRI School for Public Health and Primary Care
Modeling approaches
• Technique should match conceptual model
– Decision tree • No explicit modeling of time, easy/messy
– State transition model • Widely used
– Discrete event simulation • Constrained resources / interaction between individuals / time-
to-event data / need to record individual experience
– Dynamic transmission model • Infectious disease
20
CAPHRI School for Public Health and Primary Care
Decision tree
Option A
Option B
Healthy
Disease
Death
Healthy
Disease
Death
Healthy
Disease
Death
IDEM
0.6
0.1
0.3
Etc.
Etc.
Etc.
Etc.
21
CAPHRI School for Public Health and Primary Care
State transition model (ISPOR/SMDM taskforce 3)
• Conceptualizes a decision problem in terms of – mutually exclusive and collectively exhaustive health states
– transitions among those states
22
CAPHRI School for Public Health and Primary Care
State transition model: mutually exclusive & collectively exhaustive health states & transitions between those states
• Health states
• Transitions
• Utilities & Costs
– Health states
– Transitions
Alive
Death
23
CAPHRI School for Public Health and Primary Care
State transition model: mutually exclusive & exhaustive health states
• Population with disease X, mild stage
• Patients may progress to severe and have increased mortality risk, but may also improve
• New treatment may decrease probability of progression and mortality
• Can you think of a model structure?
24
CAPHRI School for Public Health and Primary Care
State transition model: mutually exclusive & exhaustive health states
Healthy Disease
Death
25
CAPHRI School for Public Health and Primary Care
State transition model: comparing health care technologies
Healthy Disease
Death
Healthy Disease
Death
26
CAPHRI School for Public Health and Primary Care
State transition model: Timing - discrete
Healthy Disease
Death
Healthy Disease
Death
Healthy Disease
Death
Cycle 0
Cycle 1
Cycle 2
Etcetera Time horizon
27
CAPHRI School for Public Health and Primary Care
State transition model (ISPOR/SMDM taskforce 3)
• Conceptualizes a decision problem in terms of – mutually exclusive and collectively exhaustive health states
– transitions among those states
• Takes timing into account
• Suitable to model the dynamics of disease and intervention (recurrent events)
• Markovian property (hence “Markov model”) – transition probabilities do not depend on history neither on past
states nor on the time spent in the current state
– very limiting!
28
CAPHRI School for Public Health and Primary Care
Training 1 – conceptualising a model
• 20 minutes!
29
CAPHRI School for Public Health and Primary Care
Manuela Joore
Professor HTA & Decision Making Deputy Head Department Clinical Epidemiology & Medical Technology Assessment
Maastricht University Hospital PO Box 5800
6202 AZ The Netherlands [email protected]
30
CAPHRI School for Public Health and Primary Care
State transition model: mutually exclusive & exhaustive health states
Healthy Disease
Death
31
CAPHRI School for Public Health and Primary Care
Training 1
• Time horizon:
• Modelling approach:
• Structure:
• What types of information do you need?