0:00

So we can draw a few conclusions from all of this.

0:03

Most errors that we make

0:05

as radiologists are perceptual errors,

0:07

simply not seeing something.

0:09

Although we might actually see it later, we

0:12

can often see it in retrospect.

0:14

The other errors radiologists make are often related

0:16

to cognitive biases and mental shortcuts or heuristics.

0:20

Very few errors are because of gaps in knowledge.

0:23

And I think that maybe because we're all so well-trained

0:26

and it, it could be that we worry about, um, making sure

0:30

that we have all the knowledge that we

0:31

need before we come to work.

0:32

Um, and it worked.

0:33

You know, maybe if we didn't do CME,

0:35

we'd make even more errors, but it's a very,

0:37

very tiny fraction of the errors we make are due to that.

0:40

Um, sustained vigilance.

0:44

You know, staying in front

0:45

of a screen just looking at this stuff, uh, all day long.

0:48

It may be limited by our biology,

0:51

and maybe this is a feature and not a bug.

0:53

I mean, we didn't evolve to practice radiology.

0:56

We, we evolved really fundamentally

0:59

to appreciate the music of the Beatles.

1:03

One more thing, you know, some error may be due to

1:06

what we call microsleep events.

1:07

These are very short duration, um, less than a second.

1:11

They can last up to 30 seconds. They're detectable by EEG.

1:15

And, uh, it's possible that, um, you know,

1:18

that might be at times when the lights are on

1:20

and, and nobody's home.

1:21

In our own laboratory, we've done, uh, quite a bit

1:24

of research looking at, um, sort of, uh, uh, interplay

1:28

between different networks, uh, uh, a higher brain function,

1:32

you know, the default mode network, the parietal, uh,

1:35

network, uh, salience network,

1:37

and how those can interact to create a situation where, uh,

1:42

you're looking but not seeing.

1:43

And we actually have found, uh,

1:45

that there is an increased risk

1:47

of error under certain mental conditions,

1:50

which are generally not detectable, um, you know,

1:53

without specialized equipment.

1:55

Some suggested reading, uh,

1:57

there's a great book called Being Wrong by Katherine Scholz,

2:00

where she talks about how, um, when you're wrong,

2:04

it feels just exactly the same as when you're right.

2:06

And so, uh, you, you can't necessarily know,

2:09

like I know I'm right 'cause it feels right.

2:11

Um, great book called Why We Make Mistakes, which is about

2:15

how we tend to self justify, uh, another book.

2:18

Along that same line, mistakes were made, but not by me.

2:23

Um, this is a terrific book.

2:24

How Not To Be Wrong by the Mathematic

2:27

mathematician, Jordan Ellison.

2:28

This, actually, this is probably the most fun book you'll

2:30

ever read by a mathematician.

2:32

Um, and, uh, thinking Fast and Slow.

2:35

I mentioned, uh, from Daniel Canman,

2:38

and then there was a terrific book called Snowball

2:40

and a blizzard by, uh, Dr.

2:42

Steven Hatch who talked about, uh,

2:44

how hard might recognize this as a mammogram.

2:47

Uh, how hard does, you know, to find some things

2:49

because they, you know, they're difficult to, we, we deal

2:51

with a high level of uncertainty.

2:53

And again, the whole concept of, of, of uncertainty

2:56

and the high level of uncertainty that we deal with

2:58

Sort of begs the question that how much of the things

3:01

that we call error are just a manifestation

3:04

of the inevitable uncertainty that we find ourselves in.

3:07

Um, terrific book, uh, by Stuart Firestein, the head

3:10

of Biology at Columbia, uh, talking about, uh, how failure

3:14

and repeated failure is what made science successful.

3:18

And, you know, we, we need to do that.

3:21

So, uh, a couple of final thoughts on on closing.

3:24

Uh, when you think about tort law, tort exists, you know,

3:28

basically, uh, for multiple reasons.

3:31

Uh, one is to deter bad behavior.

3:35

Uh, the other is to punish malfeasance.

3:38

And the third is to compensate people that are harmed.

3:41

And when you're dealing with perceptual error,

3:43

the most common errors that radiologists make, you know,

3:46

most of them are outta the radiologist's control

3:49

that we simply don't see something, even though we looked

3:52

and we did all the things that we're supposed to do.

3:54

Um, and you can't get that

3:58

to improve by punishing it.

3:59

You can't, um, change that.

4:02

Uh, you can't just, you know, be more careful

4:05

and it just doesn't work.

4:07

Uh, we're dealing with an underlying biological process,

4:10

so perhaps there are some situations

4:11

that we commonly see radiologists sued for,

4:14

that it's really not appropriate to go through tort law,

4:17

and we ought to find other ways to compensate, uh, people

4:21

who are harmed because people are harmed by our errors.

4:23

And that's, that's a reality,

4:25

and we have to, uh, deal with that

4:27

and, um, make restitution for that.

4:29

Um, but sometimes, you know, blame

4:32

and, you know, concept of negligence

4:35

and harm are not as tightly connected as people may think.

4:38

And, uh, when you can't control it,

4:41

but it's a reality, then you have

4:43

to mitigate it some other way.

4:44

And, you know, that's what the insurance industry,

4:47

you know, is supposed to do.

4:48

Um, and asbestos, we have, you know, a fund that pays people

4:52

who are harmed by asbestos without necessarily pointing the

4:54

finger and blame at people

4:56

who used insulation in the 1940s, you know, for example.

5:00

So, one last thought in closing, uh, about quality.

5:04

The race for quality has no finish line.

5:06

So when you think about it, is kind of like a death march.