0:01

So, uh, in terms of diagnostic errors

0:04

that we have in radiology,

0:05

there's really two big categories.

0:08

Visual perception is the big one,

0:09

and then cognitive interpretation.

0:11

So just seeing things and then making sense of them.

0:15

The radiologist error rate of interpretation,

0:17

which includes both of those is been estimated around three

0:21

and 4%, and it's been measured, you know, at that level

0:25

for a really long time.

0:26

Uh, of course the error rate is higher.

0:29

It goes up to 30% if you only sample exams

0:32

with abnormal findings.

0:33

So the three to 4% error rate

0:35

that we're talking about is based on a usual case mix in,

0:39

in a typical daily, uh, you know, workflow of a radiologist,

0:43

uh, if you, if you artificially change, you know

0:46

what you're looking at,

0:47

and then measure the error rate,

0:48

you'll get higher rates of error.

0:50

Uh, you know, if you look at specific modalities, you know,

0:52

x-rays, the, the error rate's been described at 20% CT 7%.

0:58

Interestingly, the error rate increases

0:59

with low target prevalence.

1:01

Uh, so if you have, um, you know, fewer, uh,

1:06

abnormals in your pile, you're, you're more likely

1:08

to miss the few that you have.

1:10

Uh, and the first person to study errors was Dr.

1:13

Leo Henry Garland, who is a, uh,

1:16

radiologist in San Francisco.

1:17

And he had been the president of the RSNA,

1:19

and he lived, uh, from 1903 to 1966.

1:23

Uh, so, uh, I overlapped him a little

1:25

bit, but not professionally.

1:27

And, uh, uh, he didn't make himself, uh, exactly, um,

1:31

popular by, uh, turning the lens

1:34

of science on radiologist errors.

1:36

I think people wanted to believe, you know,

1:38

like my colleague before that, uh, you know,

1:40

they never made errors, but he, he showed that we do,

1:43

and he measured it for the first time.

1:45

Uh, so there are part mistakes of particular interest

1:47

and concern for us in radiology practice.

1:49

And, uh, you know, this is the list of them.

1:52

Uh, I would say that, uh, of all

1:55

of these simply missed findings, the perceptual error is

1:58

by far the most important.

2:01

Uh, but, uh, you know, obviously we have the same sorts

2:03

of things, uh, that other specialties do, like in terms

2:06

of general safety, like patient falls, medication errors,

2:09

uh, you know, failure to manage

2:11

adverse events appropriately.

2:13

But, uh, uniquely we have, uh,

2:15

just simply not seeing something.

2:17

There is a certain baseline variability

2:20

in radiology interpretation.

2:22

And, you know, this is just the fact of, of our life.

2:25

We don't all agree. Uh, Hani Ajuah and,

2:28

and, uh, friends found that there was about a 25% chance

2:32

of a radiologist interpreting the same exam

2:35

differently on a second reading.

2:37

And, uh, he did a study where he, uh,

2:39

had volunteers from his, his colleagues at Mass General,

2:42

and, uh, they, uh, they reread a set, a, a, a set of, uh,

2:46

body cts, which included ones that they themselves had read

2:50

and, um, and that their colleagues had read.

2:52

And they disagree with themselves about 25%

2:54

of, uh, of the time.

2:56

And they disagree with each other about 30% of the time.

2:59

So here's the reference on that,

3:01

if you wanna look up the study.

3:02

Um, not surprisingly,

3:04

he had a hard time getting this published,

3:05

even though it was absolutely solid work,

3:07

but it, it ultimately found a home.

3:10

It was published in a European journal.

3:11

He couldn't get it published in the United States.

3:13

So, uh, here's the great American philosopher, Homer Simpson

3:16

again, uh, reminding us how we feel when we find out, uh,

3:20

that we've made an error.