Transcript Cognitive Psychology

Heuristics and Medical Decision Making
Clinical Grand Rounds
Aug. 15, 2007
Dr. Shounak Das
Medical Errors
• Classification of Medical Errors:
1. Patient factors
2. Outside systems
3. Access: EMS, transfers
4. Triage
5. Human error
6. Teamwork failure
7. Local environment: the microsystem
8. Hospital environment: the macrosystem
9. Hospital administration and third party factors
10.Community, society, health care policy
Medical Errors
• Human Errors:




Cognitive error
Skill-set error
Task-based error
Personal impairment
Heuristics
• Heuristics (hyu’-ris-tiks)
= an aid to learning or problem solving by
experimental and especially trial-and-error
methods; cognitive “short cuts” or
“rules of thumb”
+ve: increase efficiency
-ve: potential source of diagnostic error
Heuristics
•
Types of heuristics:
1)
2)
3)
4)
5)
6)
7)
8)
9)
Representative heuristic
Sampling heuristic
Saliency heuristic
Simple weighting heuristic
Availability heuristic
Anchoring heuristic
Framing effect
Blind obedience
Premature closure
Representative Heuristic
• Representative heuristic = how well
signs + symptoms fit a “representative”
picture of a particular disease
 i.e.: a patient presents with pleuritic
chest pain, dyspnea, and a low-grade fever
• diagnosis = PE
• ignores pre-test probabilities – i.e. the differential
diagnosis includes pneumonia which is far likelier
Sampling Heuristic
• Sampling heuristic = basing pre-test
probabilities on personal experience
 i.e.: an intern who trains at a tertiary
academic center sees 3 cases of
granulomatous vasculitis during her medicine
rotation
• a patient presents with dyspnea + wheezing
• diagnosis = granulomatous vasculitis!
• overestimates rare disorder (gran. vasc.) and
underestimates common disorder (asthma)
Saliency Heuristic
• Saliency heuristic = focusing on a
“striking” point, feature, or highlight




recency
rarity
novel clinical features
“burned” by missing a case
Simple Weighting Heuristic
• Simple weighting heuristic =
assigning equal value to all factors
 i.e.: a patient presents with chest pain,
a strong +ve family history of CAD, nausea,
and diaphoresis
• nausea + diaphoresis = ?gastroenteritis
• chest pain + strong +ve family history of CAD
= ?acute coronary syndrome
Availability Heuristic
• Availability heuristic = focusing on
diagnoses which are easily available
 common or recently encountered problems
• i.e.: are there more words in the English
language that begin with the letter “r” or have
the letter “r” as their third letter?
• people tend to think that there are more words
that begin with “r” because they’re easier to
“r”ecall even though the true ratio is almost
2:1 the other way
Anchoring Heuristic
• Anchoring heuristic = sticking to first
impression
 i.e. people thinking that their arthritis
symptoms are worse when the weather’s
bad ̶ this may have happened on a
single occasion, but people remember it
and forget the other times they’ve had
symptoms on sunny days
Framing Effect
• Framing effect = coming to different
conclusions depending on how the
information is presented
 i.e. more people chose radiation
treatment over surgery for lung cancer if
it was presented as giving them a 90%
chance of surviving than when it was
presented as giving them a 10% chance
of dying
Blind Obedience
• Blind obedience = obeying another
authority
 attending physician or consultant
 prior diagnoses
 lab or x-ray finding
Premature Closure
• Premature closure = stop thinking of
alternative diagnoses or explanations
(this is a type of anchoring bias)
 premature closure is paradoxically more
compelling when there are several
choices vs. 1 choice: in one study –
clinicians chose surgery over medications
to treat hip pain (72% vs. 53%) when 2
alternative medications were offered as
opposed to just 1 medication
*reminder to self – make a joke about orthopedic surgeons
Case Study
• Mr. Davis is a 65 year-old African
American man who presented to the
ER of an academic medical center
with back pain, general body aches,
and a sore throat. He was given a
diagnosis of a “viral syndrome” and
sent home on ibuprofen.
Case Study
• Availability heuristic
 “viral syndromes” are common, so
alternative diagnoses are not considered
• Anchoring heuristic
 once a diagnosis is made, other data are
ignored (?back pain)
Case Study
• Mr. Davis ends up having a positive
blood culture for Staph. aureus
Case Study
• Here, if the data were framed as a case of:
“pharyngitis, myalgias, and a blood culture
positive for Staphylococcus,” one might
stick with the diagnosis of a viral syndrome
and explain away the positive blood culture
as a skin contaminant. This would be an
example of both the framing effect, and the
anchoring heuristic (consider how the
differential diagnosis changes if the case is
presented as “fever, back pain, and
hematuria”).
Case Study
• However, Mr. Davis is called and told
to come back to the ER
Case Study
• Mr. Davis is admitted to hospital and started
on vancomycin. He has a normal
transthoracic echocardiogram. Plain films of
his cervical and lumbar spines just show
degenerative changes. He has a long history
of moderately severe lichen planus, and it is
assumed that this is the source of his
infection. The Staph. comes back methicillinsensitive, so he is switched to nafcillin, and
discharged home in 4 days. He is instructed
to complete a 2-week course of dicloxacillin,
and to follow-up with his PCP in 2-3 weeks.
Case Study
• In this case, blind obedience to the findings of
a normal echocardiogram and negative plain
films have resulted in the premature closure
of both endocarditis and osteomyelitis as
possible diagnoses.
• Neither test has sufficient sensitivity to
completely rule out these diagnoses.
• Consider that Staph. bacteremia is unusual in
a non-diabetic patient. Also, lichen planus
leading to bacteremia is unusual. One also
wonders why this situation has never
developed previously when Mr. Davis has such
a long history of lichen planus.
Case Study
• Mr. Davis sees his PCP as instructed.
Since being discharged, symptoms of
generalized fatigue, neck and back
pain have recurred. He also reports
tingling sensations in his fingers and
difficulty urinating. Blood cultures
are drawn, and he is sent home.
Case Study
• One of 2 surveillance blood cultures is
positive for S. aureus, so Mr. Davis is
readmitted to hospital.
• This time he has an MRI of the spine,
and is diagnosed with osteomyelitis at
C6-7 with an epidural abscess and
impingement of the spinal cord. He
declines surgery and is instead
treated with 6 weeks of IV antibiotics.
Case Study
• The PCP may have fallen trap to the following
heuristics:
 Blind obedience

accepting the diagnoses given to him by Mr. Davis’
physicians in the hospital
 Anchoring heuristic

sticking with the initial diagnosis of Staph. bacteremia
secondary to lichen planus
 Sampling heuristic

he may never have seen a case of spinal osteomyelitis with
an epidural abscess, so he does not consider this diagnosis
 Simple weighting heuristic

giving fatigue equal weight with difficulty urinating in Mr.
Davis’ symptom complex
Case Study
• A traditional critique of this case might be
to say: “never forget osteomyelitis”
• Looking at the case from the perspective of
cognitive psychology allows for analysis of
decision making, and where shortcuts can
lead one down the wrong path
• A seasoned clinician can still rely on her
heuristics, but awareness of them can add
safeguards to the diagnostic process
Using Follow-up To Overcome
Cognitive Fallibilities
• Follow-up is a feasible strategy to prevent
cognitive shortcuts from causing harm
• Follow-up would give time to read up on a
subject to counteract the availability
heuristic
• Follow-up would also give time and distance
from a case to counteract the anchoring
heuristic and premature closure
• One caveat to follow-up is that delay in
diagnosis of certain conditions can cause
irreparable damage
Some Corrective Strategies to
Counteract Heuristics
References
1.
2.
3.
4.
5.
Redelmeier DA. The cognitive psychology of missed
diagnoses. Ann Int Med 2005; 142: 115-120.
Redelmeier DA. Problems for clinical judgement:
introducing cognitive psychology as one more basic
science. CMAJ 2001; 164: 358-360.
Elstein AS. Heuristics and biases: selected errors in
clinical reasoning. Acad Med 1999; 74: 791-794.
Kohn LT et al. eds. “To err is human: building a safer
health system.” Washington, DC: National Academy
Press; 1999.
Redelmeier DA, Shafir E. Medical decision making in
situations that offer multiple alternatives. JAMA. 1995;
273: 302-5.