chicken pox story

Bayesian decision making in
primary care – or how to stop
people dying of chicken pox
Trisha Greenhalgh
Professor of Primary Health Care
UCL


Before we start:

What is primary health care?


Hospital medicine
“Distinguishing the clear message of the disease
from the interfering noise of the patient as a
person.”

Marshall Marinker. ‘The mythology of Hilda Thompson’
In Greenhalgh T and Hurwitz H (eds)
‘Narrative Based Medicine’. London: BMJ Books, 1998


Primary health care
“In secondary care diseases stay, but patients
come and go, whereas in primary care
patients stay but diseases come and go."

Iona Heath
‘The mystery of general practice’.
London: Nuffield Provincial Hospitals Trust, 1995

Primary health care
“First-contact care, delivered by generalists,
dependent on teamwork, which is
accessible, comprehensive, co-ordinated,
population-based, and activated by
patient choice.”

Pat Gordon and Diane Plamping
‘Extending Primary Care’. Oxford: Radcliffe, 1996


Primary health care
“Doing simple things well, for large numbers
of people, few of whom feel ill.”

Julian Tudor Hart
‘A new kind of doctor’. London: Merlin Press, 1998


Case history

A patient with query chicken pox


A patient with chicken pox
It was Saturday morning. I was on call from
8.30 am. I got a call from one of my
partners, Dr B, at 5.45 am. He was on

holiday 200 miles away but had been called
on his mobile phone by Health Call. One of
his patients had rung Health Call and
demanded a visit by Dr B. No other doctor
would do.


A patient with chicken pox
The family had a child with chicken pox. She
had been seen the day before by another
partner, Dr R, who has 24 years’ experience
in general practice and is also a clinical
assistant in dermatology. She had said it
was “definitely chicken pox” and prescribed
fluids, analgesia and calamine.


A patient with chicken pox
The child had apparently deteriorated and the
parents were worried. They had decided
that only Dr B would know what to do. Dr B
(who was many miles away) asked me to
go round immediately and examine the
child. I was not yet on call and keen to go
for my early morning swim before surgery.
What should my next move be?


Intermission: getting by as a GP
You only need to answer three questions:

1. Are they ill or are they not ill?
2. If ill, can I deal with it or does someone
else need to be involved?
3. If someone else, can it wait 12 weeks or
can’t it?
Cecilia Gould
Crouch End Surgery coffee break, July 1989


Bayesian decision-making
Pre-test odds of
disease X

TEST Y
Post-test odds of
disease X


Bayesian decision-making
Parent phones up to
say “I think my child
has chicken pox”

Pre-testY odds of
disease X

Dr R examines child

TEST Y
Post-testY odds of

disease X

Swab to virology

O.5

O.97

TEST Z
Post-testZ odds of
disease X

O.99


Bayesian decision-making
Assume Disease X = Patient is seriously ill


Bayesian decision-making
Parent phones up
asking for visit to child
with chicken pox

Pre-testP odds of
serious illness

INSERT QUESTION
HERE

TEST P
Post-testP odds of
serious illness

INSERT QUESTION
HERE

O.0005

O.005

TEST Q
Post-testQ odds of
serious illness

O.5


A patient with chicken pox
I asked:
1. “How old is the child?”

[Answer: 15]


Bayesian decision-making
Parent phones up
asking for visit to child
with chicken pox”

Pre-testP odds of
serious illness

How old is the child?
[High risk age group]

TEST P
Post-testP odds of
serious illness

INSERT QUESTION
HERE

O.0005

O.005

TEST Q
Post-testQ odds of
serious illness

O.5


A patient with chicken pox
I asked:
1. “How old is the child?”
2. “Why the $#*! are you so convinced that
these guys are not time wasters?”

A patient with chicken pox
He said:
“For one thing, this family have been on my
list for 17 years and they’ve never asked for
a visit before.
For another thing, they go to the most
orthodox synagogue in Golders Green.”


A patient with chicken pox
“And there’s one more thing I don’t like about
this case. It wasn’t the mother who rang, it
was the father. In that family, the father
never does the kids’ health.”


Probability
•

Of calling the doctor out at night:
=

1 in 17 years (1 in 6205)

x
e
d
in rn.
e

h
t
e
c
e
n
t
o on the Sabbath:
a
• Of using thetimtelephone
c
l
a
t
s
n
E
e
r
= 1 in 10,000?
a
of p

•

Of father rather than mother negotiating:
=

1 in 100?

Bayesian decision-making
Parent phones up
asking for visit to child
with chicken pox”

Pre-testP odds of
serious illness

How old is the child?
[High risk age group]

TEST P
Post-testP odds of
serious illness

How worried are the
parents?

O.0005

O.005

TEST Q
Post-testQ odds of
serious illness

O.5

The illness script theory
•
•
•

•

We start by learning detailed rules about the
cause, course and treatment of diseases
As we gain knowledge we convert these rules
to stereotypical stories (‘scripts’)
We refine our knowledge by accumulating
atypical and alternative stories via experience
and the oral tradition (grand rounds etc)
Knowledge is stored in our memory as stories


Illness scripts: chicken pox visit
•
•
•
•
•

“My febrile child should stay indoors.”
“I think my child has meningitis.”
“This is the first ever illness in my first
baby”
“My husband has got the car and I’m at
home with the 3 kids.”

“My husband and I are both working and
it’s not convenient to take time off.”


Illness scripts: chicken pox visit
•

“My 15 year old daughter definitely has
chicken pox. I’ve seen chicken pox in my
other kids and this is different. I think my
daughter is going to die.”

DOES NOT FIT KNOWN ILLNESS SCRIPT