MEDICAL DECISION MAKING
The patient (or the patients lay representative) must decide what symptoms warrant medical
attention - risk of something being serious - do you need to go to GP?
The medic must decide what diagnosis results from the symptoms
The medic must decide what treatment to recommend
Patient (or lay representative) must decide whether to accept the treatment
Policy makers must decide what treatments to make available
All involve some kind of cost benefit analysis and weighing of probabilities for risk & outcome -
utility
Medical decision making = a form of qualitative inquiry that examines the cognitive thought
processes involved in making medical decisions.
Medical errors:
• In the USA, medical errors are the 8th most common cause of death. Above breast cancer,
AIDS or traffic accidents (Kohn, Corrigan & Donaldson, 1999).
• Zhang et al:
- Created a taxonomy of medical errors based on Normans (1986) theory of action, and
Reasons (1990) theory of errors
- Procedural errors = slip in the script - performing correct procedure incorrectly -
execution errors
- Declarative errors = make a misdiagnosis - evaluation errors
How to measure utility for health:
Policy makers need to know how much quality of life is provided by an intervention so they can
decide whether to fund the intervention or not
We have to trust policy makers to make the best decision for us with our money
Similarly, a physician may want to know how patients regard the potential side effects against
the potential health benefits of a treatment, and the cost
e.g. the utility of a vasectomy :
- costs: £350
- benefits: moral, personal, population management
- risks: failure (1 in 2000), side effects (50% infection, post vasectomy pain syndrome in 5%-35% of
cases)
Three standard measures:
1. Visual Analogue scale
- Patient rates their target health on a 0-100 scale (picture of a thermometer)
- Easy to administer, easy for patients to understand
, - But unrelated to economic notion of utility, is subjective (my 10 could be your 50),
difficult to infer the relative weighing of costs and benefits
2. Standard Gamble
- Patient compares the prospect of experiencing perfect health with p probability against
a 1-p prospect of death
- Questions are iterated until the patient is indifferent between the gamble and perfect
health state
- The resulting p is the measure of utility
- E.g. if a patient is indifferent between living with diabetes for sure and undergoing a
treatment with 85% chance of perfect health but 15% chance of death, the SEU is .85
where death = 0 and perfect health =1
3. Time trade off (TTO) (Torrance 1986)
- Patient compares the prospect of living in target health state for a set number of years
(Y) against living in perfect health for a shorter number of years (X)
- As with the standard gamble, the patient makes a series of pairwise choice until the
indifference point it found
- The utility of the target state is then found by x/y
- So if the prospect of living with diabetes for 10 years followed by death is equivalent to
living in perfect health for 8.5 years followed by death, we would conclude the utility of
diabetes was .85 on a scale of 0 (imminent death) to 1 (perfect health)
- Interesting because based on trade off between Quality and Length of life - e.g. smoker
choose a reduced life for a higher quality of life
- But you might change your mind - changes throughout life - U shaped curve for self
destructive behaviors throughout life
Biases in Medicine
Framing effects (McNeil et al., 1982) - gain/loss frame
Hindsight bias (Arkes et al., 1981) - I knew it all along
Certainty bias (Tversky & Khaneman, 1986) - prefer a certain outcome to a risky one
Preference reversals (Tunney, 2009)
Added alternative (Redelmeier & Schafir, 1995)
Single vs. repeated plays (Redelmeier & Tversky, 1990)
PREFERENCE REVERSALS
Occur when the same risky or uncertain choices are offered in different forms and the preferred
option changes as a consequence
The choices people make should reflect their best interests and so they should be stable,
irrespective of how the preferences are elicited
Violation of Rational Choice Theory
P bet: 80% chance of winning £20.50
$ bet: 20% chance of winning £82.00
EV = £16.40 for both
The patient (or the patients lay representative) must decide what symptoms warrant medical
attention - risk of something being serious - do you need to go to GP?
The medic must decide what diagnosis results from the symptoms
The medic must decide what treatment to recommend
Patient (or lay representative) must decide whether to accept the treatment
Policy makers must decide what treatments to make available
All involve some kind of cost benefit analysis and weighing of probabilities for risk & outcome -
utility
Medical decision making = a form of qualitative inquiry that examines the cognitive thought
processes involved in making medical decisions.
Medical errors:
• In the USA, medical errors are the 8th most common cause of death. Above breast cancer,
AIDS or traffic accidents (Kohn, Corrigan & Donaldson, 1999).
• Zhang et al:
- Created a taxonomy of medical errors based on Normans (1986) theory of action, and
Reasons (1990) theory of errors
- Procedural errors = slip in the script - performing correct procedure incorrectly -
execution errors
- Declarative errors = make a misdiagnosis - evaluation errors
How to measure utility for health:
Policy makers need to know how much quality of life is provided by an intervention so they can
decide whether to fund the intervention or not
We have to trust policy makers to make the best decision for us with our money
Similarly, a physician may want to know how patients regard the potential side effects against
the potential health benefits of a treatment, and the cost
e.g. the utility of a vasectomy :
- costs: £350
- benefits: moral, personal, population management
- risks: failure (1 in 2000), side effects (50% infection, post vasectomy pain syndrome in 5%-35% of
cases)
Three standard measures:
1. Visual Analogue scale
- Patient rates their target health on a 0-100 scale (picture of a thermometer)
- Easy to administer, easy for patients to understand
, - But unrelated to economic notion of utility, is subjective (my 10 could be your 50),
difficult to infer the relative weighing of costs and benefits
2. Standard Gamble
- Patient compares the prospect of experiencing perfect health with p probability against
a 1-p prospect of death
- Questions are iterated until the patient is indifferent between the gamble and perfect
health state
- The resulting p is the measure of utility
- E.g. if a patient is indifferent between living with diabetes for sure and undergoing a
treatment with 85% chance of perfect health but 15% chance of death, the SEU is .85
where death = 0 and perfect health =1
3. Time trade off (TTO) (Torrance 1986)
- Patient compares the prospect of living in target health state for a set number of years
(Y) against living in perfect health for a shorter number of years (X)
- As with the standard gamble, the patient makes a series of pairwise choice until the
indifference point it found
- The utility of the target state is then found by x/y
- So if the prospect of living with diabetes for 10 years followed by death is equivalent to
living in perfect health for 8.5 years followed by death, we would conclude the utility of
diabetes was .85 on a scale of 0 (imminent death) to 1 (perfect health)
- Interesting because based on trade off between Quality and Length of life - e.g. smoker
choose a reduced life for a higher quality of life
- But you might change your mind - changes throughout life - U shaped curve for self
destructive behaviors throughout life
Biases in Medicine
Framing effects (McNeil et al., 1982) - gain/loss frame
Hindsight bias (Arkes et al., 1981) - I knew it all along
Certainty bias (Tversky & Khaneman, 1986) - prefer a certain outcome to a risky one
Preference reversals (Tunney, 2009)
Added alternative (Redelmeier & Schafir, 1995)
Single vs. repeated plays (Redelmeier & Tversky, 1990)
PREFERENCE REVERSALS
Occur when the same risky or uncertain choices are offered in different forms and the preferred
option changes as a consequence
The choices people make should reflect their best interests and so they should be stable,
irrespective of how the preferences are elicited
Violation of Rational Choice Theory
P bet: 80% chance of winning £20.50
$ bet: 20% chance of winning £82.00
EV = £16.40 for both
