MOPP, 2024
Measurement of Patient Preferences Study Guide
Course Learning Goals
❖ Explain important concepts of choice modeling
❖ Explain why choice modeling is useful in health
❖ Explain the steps of planning a preference study
❖ Generate relevant alternatives, attributes and levels for a health decision problem
❖ Explain and generate the 7 steps of data collection for a DCE study
❖ Explain and generate utility functions
❖ Explain and generate different types of experimental choice designs
❖ Explain and apply different types of choice models
❖ Interpret and explain choice model outputs, including model fit measures, parameter values,
and p-values
❖ Transform and explain choice model outputs that suit stakeholders in health
Week 1 - Introduction to Choice Modeling in Health
Learning Goals
❖ The student can explain why choice modeling is useful in health.
❖ The student can explain four key components of a choice model.
A - Why are we interested in patient preferences and choices in health?
❖ Improve informed decision-making
❖ Improve target product profiling
❖ Identify unmet medical needs
❖ Regulatory decisions
❖ Ethical to listen to the patient voice
❖ Increase patient/consumer satisfaction
Not aligning products, goods and services with patients/consumers’ preferences will end up in products,
goods, and services not being used
B - How to obtain insights into preferences and choices for health interventions → among others:
Discrete choice experiment (DCE)
DCE is a more complex, but also more complete method → if you understand this method, you will
understand the basis of most other choice models. It is popular because:
1. It has a mathematically rigorous framework
2. It can consider products and services that do not yet
exist in the market (i.e., it measures hypothetical
(stated) choices in an experiment instead of actual
(revealed) choices in real-life).
3. It gives the researcher complete control over its
characteristics
, MOPP, 2024
What is a DCE
A DCE typically consists of:
❖ Numerous respondents
❖ Who complete a number of choice sets in a survey
❖ In which they are asked to select one alternative
❖ DCE data is analyzed via choice modeling
This choice set includes three alternatives, and 5 attributes with
corresponding attribute levels
C - Why do we model choices in health
DCE data is analyzed via choice modeling. Software needs to know:
1. Choice set composition and choices made
2. Who made which choices
3. Distribution of preference parameters
4. Integration of parameter uncertainty
Two main reasons to model choices
1. Understanding choice behavior: not aligning products, goods, and services with
patient/consumers’ preferences will end up in products, goods, and services not being used
a. E.g., WTP, maximum acceptable risk, which alternative, willingness to wait
2. Forecasting demand
a. Avoid poor decision-making, trial-and-error implementation, demand-supply imbalance
b. Insights into elasticities → measures how demand responds to a change in a characteristic
D - How do we model choices
Step 1: we collect data
Step 2: we develop a mathematical model
Lancester’s theory of value (1996):
‘Subjective evaluations of a good (value/utility) are derived from the good’s characteristics (attributes)’
❖ Theoretical foundation for deconstructing objects in a number of attributes and levels
❖ Strategy: establish weights for each attribute and model the value of each object as the sum of its
parts (or a function of its parts)
McFadden’s Random Utility Theory (1974):
❖ McFadden combined Thurstone’s work with Lancester’s theory of value (1966), which postulates
that the utility of a good is derived from its characteristics
❖ Integrating statistics, he developed the Thurstone model into the tractable econometric Random
Utility Model (RUM), which relates choices to utility of attributes and alternatives available
❖ Practice now known as choice modeling
, MOPP, 2024
McFadden’s Random Utility Theory states that utility is a latent construct that can be partitioned into a
systematic (V) and a random, unexplainable component (ε):
‘Utility’ = V + ε
With people choosing the option that provides the highest utility
The utility function & preferences:
And by making assumptions on the unexplainable component (ε) it becomes possible to statistically
estimate preferences from the observed choices between different alternatives
Step 3: we estimate the model on the data to obtain results
Step 4: we transform the results into interpretable outcomes
Four key components of a choice model
1. Choice context: broader situation in which the choice information is applied and the decision is
made (e.g., choosing a treatment for illness)
2. Decision maker: individual or agent of an organization who is to make a choice (e.g., patient,
physician, policy-maker)
3. Choice set: set of possible alternatives in a choice task from which a decision maker is expected
to choose (e.g., surgery, radiotherapy, active surveillance)
4. Decision rule: behavioral process of choosing an alternative (i.e., mathematical approximation of
real behavior) (e.g., utility maximization, regret minimization)
Example of a choice study in health → flu vaccination (DCE)
❖ Flu vaccination is promoted by many health authorities, as the single option of flu prevention
❖ Recommended: annual flu vaccination to all individuals ≥ 60 yrs
❖ Many countries in Europe do not achieve high coverage in these groups
❖ In several countries, there is even a lowering trend of flu vaccination rate for elderly
❖ To satisfy vaccination coverage requirements in line with the EU and WHO goals, more effective
strategies have to be developed to increase flu vaccination coverage
❖ For developing effective flu vaccination strategies that are highly acceptable, a first important
step is knowing how vaccination and patient characteristics impact on the choice to opt for flu
vaccination, or not
Objective: quantify how vaccination and patient
characteristics impact on flu vaccination uptake
, MOPP, 2024
Methods
(I) define key components
(II) define attributes and levels via:
1. A literature study
2. Expert interviews
3. Focus groups
(III) generate a D-efficient choice design (i.e., how many
choice tasks, what attribute level combinations (week 3)
(IV) generate a survey including a DCE
❖ Brief introduction of the survey
❖ 4 background variable questions (age, education…)
❖ Explanation of attributes and levels of flu vaccination and stated choice task
❖ 1 warm-up question that carefully explained the stated choice task
❖ 8 stated choice tasks
❖ 8 influenza vaccination-related variable questions
❖ Questions related to decision style, health literacy, numeracy
❖ Questions about complexity and length of the survey
(V) data collection (week 2)
(VI) data analysis (week 5-6)
Conclusion
❖ Although vaccination characteristics proved to influence flu vaccination uptake, patient
characteristics had an even higher impact on flu vaccination uptake
❖ Policymakers and GPs can use these insights to improve their communication plans and
information regarding flu vaccination for individuals aged above 60 yrs
❖ Physicians should focus more on patients who experienced side effects due to vaccination in the
past
❖ Policymakers should tailor the standard information folder to patients who had been vaccinated
last year and to patients who had not
Measurement of Patient Preferences Study Guide
Course Learning Goals
❖ Explain important concepts of choice modeling
❖ Explain why choice modeling is useful in health
❖ Explain the steps of planning a preference study
❖ Generate relevant alternatives, attributes and levels for a health decision problem
❖ Explain and generate the 7 steps of data collection for a DCE study
❖ Explain and generate utility functions
❖ Explain and generate different types of experimental choice designs
❖ Explain and apply different types of choice models
❖ Interpret and explain choice model outputs, including model fit measures, parameter values,
and p-values
❖ Transform and explain choice model outputs that suit stakeholders in health
Week 1 - Introduction to Choice Modeling in Health
Learning Goals
❖ The student can explain why choice modeling is useful in health.
❖ The student can explain four key components of a choice model.
A - Why are we interested in patient preferences and choices in health?
❖ Improve informed decision-making
❖ Improve target product profiling
❖ Identify unmet medical needs
❖ Regulatory decisions
❖ Ethical to listen to the patient voice
❖ Increase patient/consumer satisfaction
Not aligning products, goods and services with patients/consumers’ preferences will end up in products,
goods, and services not being used
B - How to obtain insights into preferences and choices for health interventions → among others:
Discrete choice experiment (DCE)
DCE is a more complex, but also more complete method → if you understand this method, you will
understand the basis of most other choice models. It is popular because:
1. It has a mathematically rigorous framework
2. It can consider products and services that do not yet
exist in the market (i.e., it measures hypothetical
(stated) choices in an experiment instead of actual
(revealed) choices in real-life).
3. It gives the researcher complete control over its
characteristics
, MOPP, 2024
What is a DCE
A DCE typically consists of:
❖ Numerous respondents
❖ Who complete a number of choice sets in a survey
❖ In which they are asked to select one alternative
❖ DCE data is analyzed via choice modeling
This choice set includes three alternatives, and 5 attributes with
corresponding attribute levels
C - Why do we model choices in health
DCE data is analyzed via choice modeling. Software needs to know:
1. Choice set composition and choices made
2. Who made which choices
3. Distribution of preference parameters
4. Integration of parameter uncertainty
Two main reasons to model choices
1. Understanding choice behavior: not aligning products, goods, and services with
patient/consumers’ preferences will end up in products, goods, and services not being used
a. E.g., WTP, maximum acceptable risk, which alternative, willingness to wait
2. Forecasting demand
a. Avoid poor decision-making, trial-and-error implementation, demand-supply imbalance
b. Insights into elasticities → measures how demand responds to a change in a characteristic
D - How do we model choices
Step 1: we collect data
Step 2: we develop a mathematical model
Lancester’s theory of value (1996):
‘Subjective evaluations of a good (value/utility) are derived from the good’s characteristics (attributes)’
❖ Theoretical foundation for deconstructing objects in a number of attributes and levels
❖ Strategy: establish weights for each attribute and model the value of each object as the sum of its
parts (or a function of its parts)
McFadden’s Random Utility Theory (1974):
❖ McFadden combined Thurstone’s work with Lancester’s theory of value (1966), which postulates
that the utility of a good is derived from its characteristics
❖ Integrating statistics, he developed the Thurstone model into the tractable econometric Random
Utility Model (RUM), which relates choices to utility of attributes and alternatives available
❖ Practice now known as choice modeling
, MOPP, 2024
McFadden’s Random Utility Theory states that utility is a latent construct that can be partitioned into a
systematic (V) and a random, unexplainable component (ε):
‘Utility’ = V + ε
With people choosing the option that provides the highest utility
The utility function & preferences:
And by making assumptions on the unexplainable component (ε) it becomes possible to statistically
estimate preferences from the observed choices between different alternatives
Step 3: we estimate the model on the data to obtain results
Step 4: we transform the results into interpretable outcomes
Four key components of a choice model
1. Choice context: broader situation in which the choice information is applied and the decision is
made (e.g., choosing a treatment for illness)
2. Decision maker: individual or agent of an organization who is to make a choice (e.g., patient,
physician, policy-maker)
3. Choice set: set of possible alternatives in a choice task from which a decision maker is expected
to choose (e.g., surgery, radiotherapy, active surveillance)
4. Decision rule: behavioral process of choosing an alternative (i.e., mathematical approximation of
real behavior) (e.g., utility maximization, regret minimization)
Example of a choice study in health → flu vaccination (DCE)
❖ Flu vaccination is promoted by many health authorities, as the single option of flu prevention
❖ Recommended: annual flu vaccination to all individuals ≥ 60 yrs
❖ Many countries in Europe do not achieve high coverage in these groups
❖ In several countries, there is even a lowering trend of flu vaccination rate for elderly
❖ To satisfy vaccination coverage requirements in line with the EU and WHO goals, more effective
strategies have to be developed to increase flu vaccination coverage
❖ For developing effective flu vaccination strategies that are highly acceptable, a first important
step is knowing how vaccination and patient characteristics impact on the choice to opt for flu
vaccination, or not
Objective: quantify how vaccination and patient
characteristics impact on flu vaccination uptake
, MOPP, 2024
Methods
(I) define key components
(II) define attributes and levels via:
1. A literature study
2. Expert interviews
3. Focus groups
(III) generate a D-efficient choice design (i.e., how many
choice tasks, what attribute level combinations (week 3)
(IV) generate a survey including a DCE
❖ Brief introduction of the survey
❖ 4 background variable questions (age, education…)
❖ Explanation of attributes and levels of flu vaccination and stated choice task
❖ 1 warm-up question that carefully explained the stated choice task
❖ 8 stated choice tasks
❖ 8 influenza vaccination-related variable questions
❖ Questions related to decision style, health literacy, numeracy
❖ Questions about complexity and length of the survey
(V) data collection (week 2)
(VI) data analysis (week 5-6)
Conclusion
❖ Although vaccination characteristics proved to influence flu vaccination uptake, patient
characteristics had an even higher impact on flu vaccination uptake
❖ Policymakers and GPs can use these insights to improve their communication plans and
information regarding flu vaccination for individuals aged above 60 yrs
❖ Physicians should focus more on patients who experienced side effects due to vaccination in the
past
❖ Policymakers should tailor the standard information folder to patients who had been vaccinated
last year and to patients who had not
