|
STUDY
EXAMINES PHYSICIAN OVERCONFIDENCE IN MAKING
DIAGNOSES
Attention
Editors, Reporters: The conference presentation on which
this story is based is available by calling Charlene
Clifford at (740) 593-0946.
Contact:
Hal Arkes, 703-306-1757, ext. 6983; harkes@nsf.org
WASHINGTON,
D.C. -- Physicians hired as expert witnesses in medical
malpractice lawsuits have an advantage over the physician on
trial: When asked how they would have diagnosed a patient's
symptoms, they already know the answer.
Responding
with confidence that they would have come to the correct
medical conclusion in the case before the jury, these
witnesses offer convincing testimony for the prosecution.
But is it accurate testimony?
"The jury
and the expert witnesses are looking back on an event and
they know how it turned out. It biases their perception of
what should have been done," says Hal Arkes, a professor of
psychology at Ohio University who does research on medical
decision making and physician overconfidence.
This
phenomena, called hindsight bias, can predispose a jury in a
medical malpractice suit to a guilty verdict, Arkes says.
But perhaps even more troubling, he adds, is the
consequences it poses for the education of physicians and
medical students.
Teaching
hospitals often require physicians and students to
participate in clinicopathologic conferences (CPCs) --
forums in which one person presents a medical case, offers a
diagnosis and awaits the announcement of the actual cause of
death, presented later by a pathologist.
"Physicians
and students in the audience are supposed to learn things
about treating people with the same symptoms," says Arkes.
But a study done by Arkes and Case Western Reserve
University researcher Neal Dawson and their colleagues
showed that these forums are less effective than they might
otherwise be due to physician overconfidence attributable to
hindsight bias. The study of 160 doctors and medical
students at a Cleveland hospital was presented Feb. 21 at
the American Association for the Advancement of Science
meeting in Washington, D.C.
For the
study, Arkes and Dawson asked half the participants to
complete a questionnaire after the case was presented but
before the cause of death was revealed. Called the foresight
group, these people were given a list of five possible
diagnoses and asked to assign a probability that each was
the correct answer.
The other
half, called the hindsight group, was asked to fill out the
same questionnaire after the pathologist revealed the actual
diagnosis, listing the probability they would have assigned
to each item had they not known the correct answer.
The
results? People in the foresight group were less likely to
select the correct diagnosis, acknowledging that the cases
presented were difficult and wouldn't have been easy to
diagnose. But members of the hindsight group, aided by the
knowledge of the actual diagnosis, were more likely to
choose correctly.
"Rather
than learning as much as they could, the physicians in the
hindsight group felt they knew the right answer all along,"
Arkes says. "They lost a wonderful educational opportunity."
In the
hindsight section, the only group that claimed they wouldn't
have known the right answer were those physicians with more
years experience in the field, Arkes says, which suggests
the younger, more inexperienced physicians -- the ones who
need the educational training the most -- are more likely to
be overconfident.
"In
hindsight, we underestimate how difficult the task was
inforesight," he says. "It's unfair to in hindsight to
evaluate people who only had a foresight perspective."
Teaching
hospitals could eliminate hindsight bias simply by having
physicians and students list probable diagnoses before the
actual diagnosis is offered. In tougher cases, more people
might choose the wrong answer, Arkes says, but those cases
help them make a better diagnosis later for their patients.
Removing
hindsight bias from the courtroom is a more difficult task,
he says, but one the courts should attempt if a fair verdict
is the desired outcome.
"It's
important to try to get people to say what they would have
done without knowing how it turned out."
Neal
Dawson, professor of medicine and associate professor
ofepidemiology and biostatistics at Case Western, was
co-author of the study. Arkes holds an appointment in the
College of Arts and Sciences.
[
30 ]
|
